Merge branch 'stripped-models' into git-lfs

.github/workflows/frontend-checks.yml

@@ -44,7 +44,12 @@ jobs:
       - name: check for changed frontend files
         if: ${{ inputs.always_run != true }}
         id: changed-files
-        uses: tj-actions/changed-files@v42
+        # Pinned to the _hash_ for v45.0.9 to prevent supply-chain attacks.
+        # See:
+        # - CVE-2025-30066
+        # - https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised
+        # - https://github.com/tj-actions/changed-files/issues/2463
+        uses: tj-actions/changed-files@a284dc1814e3fd07f2e34267fc8f81227ed29fb8
         with:
           files_yaml: |
             frontend:

.github/workflows/frontend-tests.yml

@@ -44,7 +44,12 @@ jobs:
       - name: check for changed frontend files
         if: ${{ inputs.always_run != true }}
         id: changed-files
-        uses: tj-actions/changed-files@v42
+        # Pinned to the _hash_ for v45.0.9 to prevent supply-chain attacks.
+        # See:
+        # - CVE-2025-30066
+        # - https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised
+        # - https://github.com/tj-actions/changed-files/issues/2463
+        uses: tj-actions/changed-files@a284dc1814e3fd07f2e34267fc8f81227ed29fb8
         with:
           files_yaml: |
             frontend:

.github/workflows/python-checks.yml

@@ -43,7 +43,12 @@ jobs:
       - name: check for changed python files
         if: ${{ inputs.always_run != true }}
         id: changed-files
-        uses: tj-actions/changed-files@v42
+        # Pinned to the _hash_ for v45.0.9 to prevent supply-chain attacks.
+        # See:
+        # - CVE-2025-30066
+        # - https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised
+        # - https://github.com/tj-actions/changed-files/issues/2463
+        uses: tj-actions/changed-files@a284dc1814e3fd07f2e34267fc8f81227ed29fb8
         with:
           files_yaml: |
             python:

.github/workflows/python-tests.yml

@@ -77,7 +77,12 @@ jobs:
       - name: check for changed python files
         if: ${{ inputs.always_run != true }}
         id: changed-files
-        uses: tj-actions/changed-files@v42
+        # Pinned to the _hash_ for v45.0.9 to prevent supply-chain attacks.
+        # See:
+        # - CVE-2025-30066
+        # - https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised
+        # - https://github.com/tj-actions/changed-files/issues/2463
+        uses: tj-actions/changed-files@a284dc1814e3fd07f2e34267fc8f81227ed29fb8
         with:
           files_yaml: |
             python:

.github/workflows/typegen-checks.yml

@@ -42,7 +42,12 @@ jobs:
       - name: check for changed files
         if: ${{ inputs.always_run != true }}
         id: changed-files
-        uses: tj-actions/changed-files@v42
+        # Pinned to the _hash_ for v45.0.9 to prevent supply-chain attacks.
+        # See:
+        # - CVE-2025-30066
+        # - https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised
+        # - https://github.com/tj-actions/changed-files/issues/2463
+        uses: tj-actions/changed-files@a284dc1814e3fd07f2e34267fc8f81227ed29fb8
         with:
           files_yaml: |
             src:

configs/controlnet/cldm_v15.yaml

@@ -1,79 +0,0 @@
-model:
-  target: cldm.cldm.ControlLDM
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    control_key: "hint"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-    only_mid_control: False
-
-    control_stage_config:
-      target: cldm.cldm.ControlNet
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        hint_channels: 3
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    unet_config:
-      target: cldm.cldm.ControlledUnetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: ldm.modules.encoders.modules.FrozenCLIPEmbedder

configs/controlnet/cldm_v21.yaml

@@ -1,85 +0,0 @@
-model:
-  target: cldm.cldm.ControlLDM
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    control_key: "hint"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-    only_mid_control: False
-
-    control_stage_config:
-      target: cldm.cldm.ControlNet
-      params:
-        use_checkpoint: True
-        image_size: 32 # unused
-        in_channels: 4
-        hint_channels: 3
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    unet_config:
-      target: cldm.cldm.ControlledUnetModel
-      params:
-        use_checkpoint: True
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"

configs/stable-diffusion/sd_xl_base.yaml

@@ -1,98 +0,0 @@
-model:
-  target: sgm.models.diffusion.DiffusionEngine
-  params:
-    scale_factor: 0.13025
-    disable_first_stage_autocast: True
-
-    denoiser_config:
-      target: sgm.modules.diffusionmodules.denoiser.DiscreteDenoiser
-      params:
-        num_idx: 1000
-
-        weighting_config:
-          target: sgm.modules.diffusionmodules.denoiser_weighting.EpsWeighting
-        scaling_config:
-          target: sgm.modules.diffusionmodules.denoiser_scaling.EpsScaling
-        discretization_config:
-          target: sgm.modules.diffusionmodules.discretizer.LegacyDDPMDiscretization
-
-    network_config:
-      target: sgm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        adm_in_channels: 2816
-        num_classes: sequential
-        use_checkpoint: True
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [4, 2]
-        num_res_blocks: 2
-        channel_mult: [1, 2, 4]
-        num_head_channels: 64
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: [1, 2, 10]  # note: the first is unused (due to attn_res starting at 2) 32, 16, 8 --> 64, 32, 16
-        context_dim: 2048
-        spatial_transformer_attn_type: softmax-xformers
-        legacy: False
-
-    conditioner_config:
-      target: sgm.modules.GeneralConditioner
-      params:
-        emb_models:
-          # crossattn cond
-          - is_trainable: False
-            input_key: txt
-            target: sgm.modules.encoders.modules.FrozenCLIPEmbedder
-            params:
-              layer: hidden
-              layer_idx: 11
-          # crossattn and vector cond
-          - is_trainable: False
-            input_key: txt
-            target: sgm.modules.encoders.modules.FrozenOpenCLIPEmbedder2
-            params:
-              arch: ViT-bigG-14
-              version: laion2b_s39b_b160k
-              freeze: True
-              layer: penultimate
-              always_return_pooled: True
-              legacy: False
-          # vector cond
-          - is_trainable: False
-            input_key: original_size_as_tuple
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: crop_coords_top_left
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: target_size_as_tuple
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-
-    first_stage_config:
-      target: sgm.models.autoencoder.AutoencoderKLInferenceWrapper
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          attn_type: vanilla-xformers
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult: [1, 2, 4, 4]
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity

configs/stable-diffusion/sd_xl_inpaint.yaml

@@ -1,98 +0,0 @@
-model:
-  target: sgm.models.diffusion.DiffusionEngine
-  params:
-    scale_factor: 0.13025
-    disable_first_stage_autocast: True
-
-    denoiser_config:
-      target: sgm.modules.diffusionmodules.denoiser.DiscreteDenoiser
-      params:
-        num_idx: 1000
-
-        weighting_config:
-          target: sgm.modules.diffusionmodules.denoiser_weighting.EpsWeighting
-        scaling_config:
-          target: sgm.modules.diffusionmodules.denoiser_scaling.EpsScaling
-        discretization_config:
-          target: sgm.modules.diffusionmodules.discretizer.LegacyDDPMDiscretization
-
-    network_config:
-      target: sgm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        adm_in_channels: 2816
-        num_classes: sequential
-        use_checkpoint: True
-        in_channels: 9
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [4, 2]
-        num_res_blocks: 2
-        channel_mult: [1, 2, 4]
-        num_head_channels: 64
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: [1, 2, 10]  # note: the first is unused (due to attn_res starting at 2) 32, 16, 8 --> 64, 32, 16
-        context_dim: 2048
-        spatial_transformer_attn_type: softmax-xformers
-        legacy: False
-
-    conditioner_config:
-      target: sgm.modules.GeneralConditioner
-      params:
-        emb_models:
-          # crossattn cond
-          - is_trainable: False
-            input_key: txt
-            target: sgm.modules.encoders.modules.FrozenCLIPEmbedder
-            params:
-              layer: hidden
-              layer_idx: 11
-          # crossattn and vector cond
-          - is_trainable: False
-            input_key: txt
-            target: sgm.modules.encoders.modules.FrozenOpenCLIPEmbedder2
-            params:
-              arch: ViT-bigG-14
-              version: laion2b_s39b_b160k
-              freeze: True
-              layer: penultimate
-              always_return_pooled: True
-              legacy: False
-          # vector cond
-          - is_trainable: False
-            input_key: original_size_as_tuple
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: crop_coords_top_left
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: target_size_as_tuple
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-
-    first_stage_config:
-      target: sgm.models.autoencoder.AutoencoderKLInferenceWrapper
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          attn_type: vanilla-xformers
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult: [1, 2, 4, 4]
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity

configs/stable-diffusion/sd_xl_refiner.yaml

@@ -1,91 +0,0 @@
-model:
-  target: sgm.models.diffusion.DiffusionEngine
-  params:
-    scale_factor: 0.13025
-    disable_first_stage_autocast: True
-
-    denoiser_config:
-      target: sgm.modules.diffusionmodules.denoiser.DiscreteDenoiser
-      params:
-        num_idx: 1000
-
-        weighting_config:
-          target: sgm.modules.diffusionmodules.denoiser_weighting.EpsWeighting
-        scaling_config:
-          target: sgm.modules.diffusionmodules.denoiser_scaling.EpsScaling
-        discretization_config:
-          target: sgm.modules.diffusionmodules.discretizer.LegacyDDPMDiscretization
-
-    network_config:
-      target: sgm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        adm_in_channels: 2560
-        num_classes: sequential
-        use_checkpoint: True
-        in_channels: 4
-        out_channels: 4
-        model_channels: 384
-        attention_resolutions: [4, 2]
-        num_res_blocks: 2
-        channel_mult: [1, 2, 4, 4]
-        num_head_channels: 64
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 4
-        context_dim: [1280, 1280, 1280, 1280]  # 1280
-        spatial_transformer_attn_type: softmax-xformers
-        legacy: False
-
-    conditioner_config:
-      target: sgm.modules.GeneralConditioner
-      params:
-        emb_models:
-          # crossattn and vector cond
-          - is_trainable: False
-            input_key: txt
-            target: sgm.modules.encoders.modules.FrozenOpenCLIPEmbedder2
-            params:
-              arch: ViT-bigG-14
-              version: laion2b_s39b_b160k
-              legacy: False
-              freeze: True
-              layer: penultimate
-              always_return_pooled: True
-          # vector cond
-          - is_trainable: False
-            input_key: original_size_as_tuple
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: crop_coords_top_left
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by two
-          # vector cond
-          - is_trainable: False
-            input_key: aesthetic_score
-            target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND
-            params:
-              outdim: 256  # multiplied by one
-
-    first_stage_config:
-      target: sgm.models.autoencoder.AutoencoderKLInferenceWrapper
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          attn_type: vanilla-xformers
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult: [1, 2, 4, 4]
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity

configs/stable-diffusion/v1-finetune.yaml

@@ -1,110 +0,0 @@
-model:
-  base_learning_rate: 5.0e-03
-  target: invokeai.backend.stable_diffusion.diffusion.ddpm.LatentDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: image
-    cond_stage_key: caption
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: true   # Note: different from the one we trained before
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-    embedding_reg_weight: 0.0
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ["sculpture"]
-        per_image_tokens: false
-        num_vectors_per_token: 1
-        progressive_words: False
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.FrozenCLIPEmbedder
-
-data:
-  target: main.DataModuleFromConfig
-  params:
-    batch_size: 1
-    num_workers: 2
-    wrap: false
-    train:
-      target: invokeai.backend.stable_diffusion.data.personalized.PersonalizedBase
-      params:
-        size: 512
-        set: train
-        per_image_tokens: false
-        repeats: 100
-    validation:
-      target: invokeai.backend.stable_diffusion.data.personalized.PersonalizedBase
-      params:
-        size: 512
-        set: val
-        per_image_tokens: false
-        repeats: 10
-
-lightning:
-  modelcheckpoint:
-    params:
-      every_n_train_steps: 500
-  callbacks:
-    image_logger:
-      target: main.ImageLogger
-      params:
-        batch_frequency: 500
-        max_images: 8
-        increase_log_steps: False
-
-  trainer:
-    benchmark: True
-    max_steps: 4000000
-#    max_steps: 4000
-

configs/stable-diffusion/v1-finetune_style.yaml

@@ -1,103 +0,0 @@
-model:
-  base_learning_rate: 5.0e-03
-  target: invokeai.backend.models.diffusion.ddpm.LatentDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: image
-    cond_stage_key: caption
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: true   # Note: different from the one we trained before
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-    embedding_reg_weight: 0.0
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ["painting"]
-        per_image_tokens: false
-        num_vectors_per_token: 1
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.FrozenCLIPEmbedder
-
-data:
-  target: main.DataModuleFromConfig
-  params:
-    batch_size: 2
-    num_workers: 16
-    wrap: false
-    train:
-      target: invokeai.backend.stable_diffusion.data.personalized_style.PersonalizedBase
-      params:
-        size: 512
-        set: train
-        per_image_tokens: false
-        repeats: 100
-    validation:
-      target: invokeai.backend.stable_diffusion.data.personalized_style.PersonalizedBase
-      params:
-        size: 512
-        set: val
-        per_image_tokens: false
-        repeats: 10
-
-lightning:
-  callbacks:
-    image_logger:
-      target: main.ImageLogger
-      params:
-        batch_frequency: 500
-        max_images: 8
-        increase_log_steps: False
-
-  trainer:
-    benchmark: True

configs/stable-diffusion/v1-inference-v.yaml

@@ -1,80 +0,0 @@
-model:
-  base_learning_rate: 1.0e-04
-  target: invokeai.backend.models.diffusion.ddpm.LatentDiffusion
-  params:
-    parameterization: "v"
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false   # Note: different from the one we trained before
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-
-    scheduler_config: # 10000 warmup steps
-      target: invokeai.backend.stable_diffusion.lr_scheduler.LambdaLinearScheduler
-      params:
-        warm_up_steps: [ 10000 ]
-        cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases
-        f_start: [ 1.e-6 ]
-        f_max: [ 1. ]
-        f_min: [ 1. ]
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ['sculpture']
-        per_image_tokens: false
-        num_vectors_per_token: 1
-        progressive_words: False
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.WeightedFrozenCLIPEmbedder

configs/stable-diffusion/v1-inference.yaml

@@ -1,79 +0,0 @@
-model:
-  base_learning_rate: 1.0e-04
-  target: invokeai.backend.models.diffusion.ddpm.LatentDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false   # Note: different from the one we trained before
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-
-    scheduler_config: # 10000 warmup steps
-      target: invokeai.backend.stable_diffusion.lr_scheduler.LambdaLinearScheduler
-      params:
-        warm_up_steps: [ 10000 ]
-        cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases
-        f_start: [ 1.e-6 ]
-        f_max: [ 1. ]
-        f_min: [ 1. ]
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ['sculpture']
-        per_image_tokens: false
-        num_vectors_per_token: 1
-        progressive_words: False
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.WeightedFrozenCLIPEmbedder

configs/stable-diffusion/v1-inpainting-inference.yaml

@@ -1,79 +0,0 @@
-model:
-  base_learning_rate: 7.5e-05
-  target: invokeai.backend.models.diffusion.ddpm.LatentInpaintDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false   # Note: different from the one we trained before
-    conditioning_key: hybrid   # important
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    finetune_keys: null
-
-    scheduler_config: # 10000 warmup steps
-      target: invokeai.backend.stable_diffusion.lr_scheduler.LambdaLinearScheduler
-      params:
-        warm_up_steps: [ 2500 ] # NOTE for resuming. use 10000 if starting from scratch
-        cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases
-        f_start: [ 1.e-6 ]
-        f_max: [ 1. ]
-        f_min: [ 1. ]
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ['sculpture']
-        per_image_tokens: false
-        num_vectors_per_token: 8
-        progressive_words: False
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 9  # 4 data + 4 downscaled image + 1 mask
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.WeightedFrozenCLIPEmbedder

configs/stable-diffusion/v1-m1-finetune.yaml

@@ -1,110 +0,0 @@
-model:
-  base_learning_rate: 5.0e-03
-  target: invokeai.backend.models.diffusion.ddpm.LatentDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: image
-    cond_stage_key: caption
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: true   # Note: different from the one we trained before
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False
-    embedding_reg_weight: 0.0
-
-    personalization_config:
-      target: invokeai.backend.stable_diffusion.embedding_manager.EmbeddingManager
-      params:
-        placeholder_strings: ["*"]
-        initializer_words: ['sculpture']
-        per_image_tokens: false
-        num_vectors_per_token: 6
-        progressive_words: False
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_heads: 8
-        use_spatial_transformer: True
-        transformer_depth: 1
-        context_dim: 768
-        use_checkpoint: True
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.FrozenCLIPEmbedder
-
-data:
-  target: main.DataModuleFromConfig
-  params:
-    batch_size: 1
-    num_workers: 2
-    wrap: false
-    train:
-      target: invokeai.backend.stable_diffusion.data.personalized.PersonalizedBase
-      params:
-        size: 512
-        set: train
-        per_image_tokens: false
-        repeats: 100
-    validation:
-      target: invokeai.backend.stable_diffusion.data.personalized.PersonalizedBase
-      params:
-        size: 512
-        set: val
-        per_image_tokens: false
-        repeats: 10
-
-lightning:
-  modelcheckpoint:
-    params:
-      every_n_train_steps: 500
-  callbacks:
-    image_logger:
-      target: main.ImageLogger
-      params:
-        batch_frequency: 500
-        max_images: 5
-        increase_log_steps: False
-
-  trainer:
-    benchmark: False
-    max_steps: 6200
-#    max_steps: 4000
-

configs/stable-diffusion/v2-inference-v.yaml

@@ -1,68 +0,0 @@
-model:
-  base_learning_rate: 1.0e-4
-  target: invokeai.backend.stable_diffusion.diffusion.ddpm.LatentDiffusion
-  params:
-    parameterization: "v"
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False # we set this to false because this is an inference only config
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        use_checkpoint: True
-        use_fp16: True
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"

configs/stable-diffusion/v2-inference.yaml

@@ -1,67 +0,0 @@
-model:
-  base_learning_rate: 1.0e-4
-  target: invokeai.backend.stable_diffusion.diffusion.ddpm.LatentDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: crossattn
-    monitor: val/loss_simple_ema
-    scale_factor: 0.18215
-    use_ema: False # we set this to false because this is an inference only config
-
-    unet_config:
-      target: invokeai.backend.stable_diffusion.diffusionmodules.openaimodel.UNetModel
-      params:
-        use_checkpoint: True
-        use_fp16: True
-        image_size: 32 # unused
-        in_channels: 4
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: invokeai.backend.stable_diffusion.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-          - 1
-          - 2
-          - 4
-          - 4
-          num_res_blocks: 2
-          attn_resolutions: []
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: invokeai.backend.stable_diffusion.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"

configs/stable-diffusion/v2-inpainting-inference-v.yaml

@@ -1,159 +0,0 @@
-model:
-  base_learning_rate: 5.0e-05
-  target: ldm.models.diffusion.ddpm.LatentInpaintDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    parameterization: "v"
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: hybrid
-    scale_factor: 0.18215
-    monitor: val/loss_simple_ema
-    finetune_keys: null
-    use_ema: False
-
-    unet_config:
-      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        use_checkpoint: True
-        image_size: 32 # unused
-        in_channels: 9
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-            - 1
-            - 2
-            - 4
-            - 4
-          num_res_blocks: 2
-          attn_resolutions: [ ]
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"
-
-
-data:
-  target: ldm.data.laion.WebDataModuleFromConfig
-  params:
-    tar_base: null  # for concat as in LAION-A
-    p_unsafe_threshold: 0.1
-    filter_word_list: "data/filters.yaml"
-    max_pwatermark: 0.45
-    batch_size: 8
-    num_workers: 6
-    multinode: True
-    min_size: 512
-    train:
-      shards:
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-0/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-1/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-2/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-3/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-4/{00000..18699}.tar -"  #{00000-94333}.tar"
-      shuffle: 10000
-      image_key: jpg
-      image_transforms:
-      - target: torchvision.transforms.Resize
-        params:
-          size: 512
-          interpolation: 3
-      - target: torchvision.transforms.RandomCrop
-        params:
-          size: 512
-      postprocess:
-        target: ldm.data.laion.AddMask
-        params:
-          mode: "512train-large"
-          p_drop: 0.25
-    # NOTE use enough shards to avoid empty validation loops in workers
-    validation:
-      shards:
-        - "pipe:aws s3 cp s3://deep-floyd-s3/datasets/laion_cleaned-part5/{93001..94333}.tar - "
-      shuffle: 0
-      image_key: jpg
-      image_transforms:
-      - target: torchvision.transforms.Resize
-        params:
-          size: 512
-          interpolation: 3
-      - target: torchvision.transforms.CenterCrop
-        params:
-          size: 512
-      postprocess:
-        target: ldm.data.laion.AddMask
-        params:
-          mode: "512train-large"
-          p_drop: 0.25
-
-lightning:
-  find_unused_parameters: True
-  modelcheckpoint:
-    params:
-      every_n_train_steps: 5000
-
-  callbacks:
-    metrics_over_trainsteps_checkpoint:
-      params:
-        every_n_train_steps: 10000
-
-    image_logger:
-      target: main.ImageLogger
-      params:
-        enable_autocast: False
-        disabled: False
-        batch_frequency: 1000
-        max_images: 4
-        increase_log_steps: False
-        log_first_step: False
-        log_images_kwargs:
-          use_ema_scope: False
-          inpaint: False
-          plot_progressive_rows: False
-          plot_diffusion_rows: False
-          N: 4
-          unconditional_guidance_scale: 5.0
-          unconditional_guidance_label: [""]
-          ddim_steps: 50  # todo check these out for depth2img,
-          ddim_eta: 0.0   # todo check these out for depth2img,
-
-  trainer:
-    benchmark: True
-    val_check_interval: 5000000
-    num_sanity_val_steps: 0
-    accumulate_grad_batches: 1

configs/stable-diffusion/v2-inpainting-inference.yaml

@@ -1,158 +0,0 @@
-model:
-  base_learning_rate: 5.0e-05
-  target: ldm.models.diffusion.ddpm.LatentInpaintDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: hybrid
-    scale_factor: 0.18215
-    monitor: val/loss_simple_ema
-    finetune_keys: null
-    use_ema: False
-
-    unet_config:
-      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        use_checkpoint: True
-        image_size: 32 # unused
-        in_channels: 9
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-            - 1
-            - 2
-            - 4
-            - 4
-          num_res_blocks: 2
-          attn_resolutions: [ ]
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"
-
-
-data:
-  target: ldm.data.laion.WebDataModuleFromConfig
-  params:
-    tar_base: null  # for concat as in LAION-A
-    p_unsafe_threshold: 0.1
-    filter_word_list: "data/filters.yaml"
-    max_pwatermark: 0.45
-    batch_size: 8
-    num_workers: 6
-    multinode: True
-    min_size: 512
-    train:
-      shards:
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-0/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-1/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-2/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-3/{00000..18699}.tar -"
-        - "pipe:aws s3 cp s3://stability-aws/laion-a-native/part-4/{00000..18699}.tar -"  #{00000-94333}.tar"
-      shuffle: 10000
-      image_key: jpg
-      image_transforms:
-      - target: torchvision.transforms.Resize
-        params:
-          size: 512
-          interpolation: 3
-      - target: torchvision.transforms.RandomCrop
-        params:
-          size: 512
-      postprocess:
-        target: ldm.data.laion.AddMask
-        params:
-          mode: "512train-large"
-          p_drop: 0.25
-    # NOTE use enough shards to avoid empty validation loops in workers
-    validation:
-      shards:
-        - "pipe:aws s3 cp s3://deep-floyd-s3/datasets/laion_cleaned-part5/{93001..94333}.tar - "
-      shuffle: 0
-      image_key: jpg
-      image_transforms:
-      - target: torchvision.transforms.Resize
-        params:
-          size: 512
-          interpolation: 3
-      - target: torchvision.transforms.CenterCrop
-        params:
-          size: 512
-      postprocess:
-        target: ldm.data.laion.AddMask
-        params:
-          mode: "512train-large"
-          p_drop: 0.25
-
-lightning:
-  find_unused_parameters: True
-  modelcheckpoint:
-    params:
-      every_n_train_steps: 5000
-
-  callbacks:
-    metrics_over_trainsteps_checkpoint:
-      params:
-        every_n_train_steps: 10000
-
-    image_logger:
-      target: main.ImageLogger
-      params:
-        enable_autocast: False
-        disabled: False
-        batch_frequency: 1000
-        max_images: 4
-        increase_log_steps: False
-        log_first_step: False
-        log_images_kwargs:
-          use_ema_scope: False
-          inpaint: False
-          plot_progressive_rows: False
-          plot_diffusion_rows: False
-          N: 4
-          unconditional_guidance_scale: 5.0
-          unconditional_guidance_label: [""]
-          ddim_steps: 50  # todo check these out for depth2img,
-          ddim_eta: 0.0   # todo check these out for depth2img,
-
-  trainer:
-    benchmark: True
-    val_check_interval: 5000000
-    num_sanity_val_steps: 0
-    accumulate_grad_batches: 1

configs/stable-diffusion/v2-midas-inference.yaml

@@ -1,74 +0,0 @@
-model:
-  base_learning_rate: 5.0e-07
-  target: ldm.models.diffusion.ddpm.LatentDepth2ImageDiffusion
-  params:
-    linear_start: 0.00085
-    linear_end: 0.0120
-    num_timesteps_cond: 1
-    log_every_t: 200
-    timesteps: 1000
-    first_stage_key: "jpg"
-    cond_stage_key: "txt"
-    image_size: 64
-    channels: 4
-    cond_stage_trainable: false
-    conditioning_key: hybrid
-    scale_factor: 0.18215
-    monitor: val/loss_simple_ema
-    finetune_keys: null
-    use_ema: False
-
-    depth_stage_config:
-      target: ldm.modules.midas.api.MiDaSInference
-      params:
-        model_type: "dpt_hybrid"
-
-    unet_config:
-      target: ldm.modules.diffusionmodules.openaimodel.UNetModel
-      params:
-        use_checkpoint: True
-        image_size: 32 # unused
-        in_channels: 5
-        out_channels: 4
-        model_channels: 320
-        attention_resolutions: [ 4, 2, 1 ]
-        num_res_blocks: 2
-        channel_mult: [ 1, 2, 4, 4 ]
-        num_head_channels: 64 # need to fix for flash-attn
-        use_spatial_transformer: True
-        use_linear_in_transformer: True
-        transformer_depth: 1
-        context_dim: 1024
-        legacy: False
-
-    first_stage_config:
-      target: ldm.models.autoencoder.AutoencoderKL
-      params:
-        embed_dim: 4
-        monitor: val/rec_loss
-        ddconfig:
-          #attn_type: "vanilla-xformers"
-          double_z: true
-          z_channels: 4
-          resolution: 256
-          in_channels: 3
-          out_ch: 3
-          ch: 128
-          ch_mult:
-            - 1
-            - 2
-            - 4
-            - 4
-          num_res_blocks: 2
-          attn_resolutions: [ ]
-          dropout: 0.0
-        lossconfig:
-          target: torch.nn.Identity
-
-    cond_stage_config:
-      target: ldm.modules.encoders.modules.FrozenOpenCLIPEmbedder
-      params:
-        freeze: True
-        layer: "penultimate"
-
-

invokeai.example.yaml

@@ -1,59 +0,0 @@
-# This is an example file with default and example settings.
-# You should not copy this whole file into your config.
-# Only add the settings you need to change to your config file.
-
-# Internal metadata - do not edit:
-schema_version: 4.0.2
-
-# Put user settings here - see https://invoke-ai.github.io/InvokeAI/configuration/:
-host: 127.0.0.1
-port: 9090
-allow_origins: []
-allow_credentials: true
-allow_methods:
-- '*'
-allow_headers:
-- '*'
-log_tokenization: false
-patchmatch: true
-models_dir: models
-convert_cache_dir: models/.convert_cache
-download_cache_dir: models/.download_cache
-legacy_conf_dir: configs
-db_dir: databases
-outputs_dir: outputs
-custom_nodes_dir: nodes
-style_presets_dir: style_presets
-workflow_thumbnails_dir: workflow_thumbnails
-log_handlers:
-- console
-log_format: color
-log_level: info
-log_sql: false
-log_level_network: warning
-use_memory_db: false
-dev_reload: false
-profile_graphs: false
-profiles_dir: profiles
-log_memory_usage: false
-device_working_mem_gb: 3.0
-enable_partial_loading: false
-keep_ram_copy_of_weights: true
-lazy_offload: true
-device: auto
-precision: auto
-sequential_guidance: false
-attention_type: auto
-attention_slice_size: auto
-force_tiled_decode: false
-pil_compress_level: 1
-max_queue_size: 10000
-clear_queue_on_startup: false
-node_cache_size: 512
-hashing_algorithm: blake3_single
-remote_api_tokens:
-- url_regex: cool-models.com
-  token: my_secret_token
-- url_regex: nifty-models.com
-  token: some_other_token
-scan_models_on_startup: false

invokeai.yaml

@@ -1,4 +0,0 @@
-# Internal metadata - do not edit:
-schema_version: 4.0.2
-
-# Put user settings here - see https://invoke-ai.github.io/InvokeAI/configuration/:

invokeai/app/invocations/compel.py

@@ -40,10 +40,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "compel",
-    title="Prompt",
+    title="Prompt - SD1.5",
     tags=["prompt", "compel"],
     category="conditioning",
-    version="1.2.0",
+    version="1.2.1",
 )
 class CompelInvocation(BaseInvocation):
     """Parse prompt using compel package to conditioning."""
@@ -233,10 +233,10 @@ class SDXLPromptInvocationBase:
 
 @invocation(
     "sdxl_compel_prompt",
-    title="SDXL Prompt",
+    title="Prompt - SDXL",
     tags=["sdxl", "compel", "prompt"],
     category="conditioning",
-    version="1.2.0",
+    version="1.2.1",
 )
 class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
     """Parse prompt using compel package to conditioning."""
@@ -327,10 +327,10 @@ class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
 
 @invocation(
     "sdxl_refiner_compel_prompt",
-    title="SDXL Refiner Prompt",
+    title="Prompt - SDXL Refiner",
     tags=["sdxl", "compel", "prompt"],
     category="conditioning",
-    version="1.1.1",
+    version="1.1.2",
 )
 class SDXLRefinerCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
     """Parse prompt using compel package to conditioning."""
@@ -376,10 +376,10 @@ class CLIPSkipInvocationOutput(BaseInvocationOutput):
 
 @invocation(
     "clip_skip",
-    title="CLIP Skip",
+    title="Apply CLIP Skip - SD1.5, SDXL",
     tags=["clipskip", "clip", "skip"],
     category="conditioning",
-    version="1.1.0",
+    version="1.1.1",
 )
 class CLIPSkipInvocation(BaseInvocation):
     """Skip layers in clip text_encoder model."""

invokeai/app/invocations/controlnet_image_processors.py

@@ -87,7 +87,7 @@ class ControlOutput(BaseInvocationOutput):
     control: ControlField = OutputField(description=FieldDescriptions.control)
 
 
-@invocation("controlnet", title="ControlNet", tags=["controlnet"], category="controlnet", version="1.1.2")
+@invocation("controlnet", title="ControlNet - SD1.5, SDXL", tags=["controlnet"], category="controlnet", version="1.1.3")
 class ControlNetInvocation(BaseInvocation):
     """Collects ControlNet info to pass to other nodes"""
 

invokeai/app/invocations/denoise_latents.py

@@ -127,10 +127,10 @@ def get_scheduler(
 
 @invocation(
     "denoise_latents",
-    title="Denoise Latents",
+    title="Denoise - SD1.5, SDXL",
     tags=["latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l"],
     category="latents",
-    version="1.5.3",
+    version="1.5.4",
 )
 class DenoiseLatentsInvocation(BaseInvocation):
     """Denoises noisy latents to decodable images"""

invokeai/app/invocations/fields.py

@@ -59,6 +59,7 @@ class UIType(str, Enum, metaclass=MetaEnum):
     ControlLoRAModel = "ControlLoRAModelField"
     SigLipModel = "SigLipModelField"
     FluxReduxModel = "FluxReduxModelField"
+    LlavaOnevisionModel = "LLaVAModelField"
     # endregion
 
     # region Misc Field Types
@@ -205,6 +206,7 @@ class FieldDescriptions:
     freeu_b2 = "Scaling factor for stage 2 to amplify the contributions of backbone features."
     instantx_control_mode = "The control mode for InstantX ControlNet union models. Ignored for other ControlNet models. The standard mapping is: canny (0), tile (1), depth (2), blur (3), pose (4), gray (5), low quality (6). Negative values will be treated as 'None'."
     flux_redux_conditioning = "FLUX Redux conditioning tensor"
+    vllm_model = "The VLLM model to use"
 
 
 class ImageField(BaseModel):

invokeai/app/invocations/flux_control_lora_loader.py

@@ -21,10 +21,10 @@ class FluxControlLoRALoaderOutput(BaseInvocationOutput):
 
 @invocation(
     "flux_control_lora_loader",
-    title="Flux Control LoRA",
+    title="Control LoRA - FLUX",
     tags=["lora", "model", "flux"],
     category="model",
-    version="1.1.0",
+    version="1.1.1",
     classification=Classification.Prototype,
 )
 class FluxControlLoRALoaderInvocation(BaseInvocation):

invokeai/app/invocations/flux_model_loader.py

@@ -37,10 +37,10 @@ class FluxModelLoaderOutput(BaseInvocationOutput):
 
 @invocation(
     "flux_model_loader",
-    title="Flux Main Model",
+    title="Main Model - FLUX",
     tags=["model", "flux"],
     category="model",
-    version="1.0.5",
+    version="1.0.6",
     classification=Classification.Prototype,
 )
 class FluxModelLoaderInvocation(BaseInvocation):

invokeai/app/invocations/flux_text_encoder.py

@@ -26,10 +26,10 @@ from invokeai.backend.stable_diffusion.diffusion.conditioning_data import Condit
 
 @invocation(
     "flux_text_encoder",
-    title="FLUX Text Encoding",
+    title="Prompt - FLUX",
     tags=["prompt", "conditioning", "flux"],
     category="conditioning",
-    version="1.1.1",
+    version="1.1.2",
     classification=Classification.Prototype,
 )
 class FluxTextEncoderInvocation(BaseInvocation):

invokeai/app/invocations/flux_vae_decode.py

@@ -22,10 +22,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "flux_vae_decode",
-    title="FLUX Latents to Image",
+    title="Latents to Image - FLUX",
     tags=["latents", "image", "vae", "l2i", "flux"],
     category="latents",
-    version="1.0.1",
+    version="1.0.2",
 )
 class FluxVaeDecodeInvocation(BaseInvocation, WithMetadata, WithBoard):
     """Generates an image from latents."""

invokeai/app/invocations/flux_vae_encode.py

@@ -19,10 +19,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "flux_vae_encode",
-    title="FLUX Image to Latents",
+    title="Image to Latents - FLUX",
     tags=["latents", "image", "vae", "i2l", "flux"],
     category="latents",
-    version="1.0.0",
+    version="1.0.1",
 )
 class FluxVaeEncodeInvocation(BaseInvocation):
     """Encodes an image into latents."""

invokeai/app/invocations/ideal_size.py

@@ -19,9 +19,9 @@ class IdealSizeOutput(BaseInvocationOutput):
 
 @invocation(
     "ideal_size",
-    title="Ideal Size",
+    title="Ideal Size - SD1.5, SDXL",
     tags=["latents", "math", "ideal_size"],
-    version="1.0.4",
+    version="1.0.5",
 )
 class IdealSizeInvocation(BaseInvocation):
     """Calculates the ideal size for generation to avoid duplication"""

invokeai/app/invocations/image_to_latents.py

@@ -31,10 +31,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "i2l",
-    title="Image to Latents",
+    title="Image to Latents - SD1.5, SDXL",
     tags=["latents", "image", "vae", "i2l"],
     category="latents",
-    version="1.1.0",
+    version="1.1.1",
 )
 class ImageToLatentsInvocation(BaseInvocation):
     """Encodes an image into latents."""

invokeai/app/invocations/ip_adapter.py

@@ -69,7 +69,13 @@ CLIP_VISION_MODEL_MAP: dict[Literal["ViT-L", "ViT-H", "ViT-G"], StarterModel] =
 }
 
 
-@invocation("ip_adapter", title="IP-Adapter", tags=["ip_adapter", "control"], category="ip_adapter", version="1.5.0")
+@invocation(
+    "ip_adapter",
+    title="IP-Adapter - SD1.5, SDXL",
+    tags=["ip_adapter", "control"],
+    category="ip_adapter",
+    version="1.5.1",
+)
 class IPAdapterInvocation(BaseInvocation):
     """Collects IP-Adapter info to pass to other nodes."""
 

invokeai/app/invocations/latents_to_image.py

@@ -31,10 +31,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "l2i",
-    title="Latents to Image",
+    title="Latents to Image - SD1.5, SDXL",
     tags=["latents", "image", "vae", "l2i"],
     category="latents",
-    version="1.3.1",
+    version="1.3.2",
 )
 class LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
     """Generates an image from latents."""

invokeai/app/invocations/llava_onevision_vllm.py

@@ -0,0 +1,60 @@
+from typing import Any
+
+import torch
+from PIL.Image import Image
+from pydantic import field_validator
+
+from invokeai.app.invocations.baseinvocation import BaseInvocation, invocation
+from invokeai.app.invocations.fields import FieldDescriptions, ImageField, InputField, UIComponent, UIType
+from invokeai.app.invocations.model import ModelIdentifierField
+from invokeai.app.invocations.primitives import StringOutput
+from invokeai.app.services.shared.invocation_context import InvocationContext
+from invokeai.backend.llava_onevision_model import LlavaOnevisionModel
+from invokeai.backend.util.devices import TorchDevice
+
+
+@invocation("llava_onevision_vllm", title="LLaVA OneVision VLLM", tags=["vllm"], category="vllm", version="1.0.0")
+class LlavaOnevisionVllmInvocation(BaseInvocation):
+    """Run a LLaVA OneVision VLLM model."""
+
+    images: list[ImageField] | ImageField | None = InputField(default=None, max_length=3, description="Input image.")
+    prompt: str = InputField(
+        default="",
+        description="Input text prompt.",
+        ui_component=UIComponent.Textarea,
+    )
+    vllm_model: ModelIdentifierField = InputField(
+        title="LLaVA Model Type",
+        description=FieldDescriptions.vllm_model,
+        ui_type=UIType.LlavaOnevisionModel,
+    )
+
+    @field_validator("images", mode="before")
+    def listify_images(cls, v: Any) -> list:
+        if v is None:
+            return v
+        if not isinstance(v, list):
+            return [v]
+        return v
+
+    def _get_images(self, context: InvocationContext) -> list[Image]:
+        if self.images is None:
+            return []
+
+        image_fields = self.images if isinstance(self.images, list) else [self.images]
+        return [context.images.get_pil(image_field.image_name, "RGB") for image_field in image_fields]
+
+    @torch.no_grad()
+    def invoke(self, context: InvocationContext) -> StringOutput:
+        images = self._get_images(context)
+
+        with context.models.load(self.vllm_model) as vllm_model:
+            assert isinstance(vllm_model, LlavaOnevisionModel)
+            output = vllm_model.run(
+                prompt=self.prompt,
+                images=images,
+                device=TorchDevice.choose_torch_device(),
+                dtype=TorchDevice.choose_torch_dtype(),
+            )
+
+        return StringOutput(value=output)

invokeai/app/invocations/metadata_linked.py

@@ -610,10 +610,10 @@ class LatentsMetaOutput(LatentsOutput, MetadataOutput):
 
 @invocation(
     "denoise_latents_meta",
-    title="Denoise Latents + metadata",
+    title=f"{DenoiseLatentsInvocation.UIConfig.title} + Metadata",
     tags=["latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l"],
     category="latents",
-    version="1.1.0",
+    version="1.1.1",
 )
 class DenoiseLatentsMetaInvocation(DenoiseLatentsInvocation, WithMetadata):
     def invoke(self, context: InvocationContext) -> LatentsMetaOutput:
@@ -675,10 +675,10 @@ class DenoiseLatentsMetaInvocation(DenoiseLatentsInvocation, WithMetadata):
 
 @invocation(
     "flux_denoise_meta",
-    title="Flux Denoise + metadata",
+    title=f"{FluxDenoiseInvocation.UIConfig.title} + Metadata",
     tags=["flux", "latents", "denoise", "txt2img", "t2i", "t2l", "img2img", "i2i", "l2l"],
     category="latents",
-    version="1.0.0",
+    version="1.0.1",
 )
 class FluxDenoiseLatentsMetaInvocation(FluxDenoiseInvocation, WithMetadata):
     """Run denoising process with a FLUX transformer model + metadata."""

invokeai/app/invocations/model.py

@@ -122,10 +122,10 @@ class ModelIdentifierOutput(BaseInvocationOutput):
 
 @invocation(
     "model_identifier",
-    title="Model identifier",
+    title="Any Model",
     tags=["model"],
     category="model",
-    version="1.0.0",
+    version="1.0.1",
     classification=Classification.Prototype,
 )
 class ModelIdentifierInvocation(BaseInvocation):
@@ -144,10 +144,10 @@ class ModelIdentifierInvocation(BaseInvocation):
 
 @invocation(
     "main_model_loader",
-    title="Main Model",
+    title="Main Model - SD1.5",
     tags=["model"],
     category="model",
-    version="1.0.3",
+    version="1.0.4",
 )
 class MainModelLoaderInvocation(BaseInvocation):
     """Loads a main model, outputting its submodels."""
@@ -244,7 +244,7 @@ class LoRASelectorOutput(BaseInvocationOutput):
     lora: LoRAField = OutputField(description="LoRA model and weight", title="LoRA")
 
 
-@invocation("lora_selector", title="LoRA Selector", tags=["model"], category="model", version="1.0.1")
+@invocation("lora_selector", title="LoRA Model - SD1.5", tags=["model"], category="model", version="1.0.2")
 class LoRASelectorInvocation(BaseInvocation):
     """Selects a LoRA model and weight."""
 
@@ -257,7 +257,9 @@ class LoRASelectorInvocation(BaseInvocation):
         return LoRASelectorOutput(lora=LoRAField(lora=self.lora, weight=self.weight))
 
 
-@invocation("lora_collection_loader", title="LoRA Collection Loader", tags=["model"], category="model", version="1.1.0")
+@invocation(
+    "lora_collection_loader", title="LoRA Collection - SD1.5", tags=["model"], category="model", version="1.1.1"
+)
 class LoRACollectionLoader(BaseInvocation):
     """Applies a collection of LoRAs to the provided UNet and CLIP models."""
 
@@ -320,10 +322,10 @@ class SDXLLoRALoaderOutput(BaseInvocationOutput):
 
 @invocation(
     "sdxl_lora_loader",
-    title="SDXL LoRA",
+    title="LoRA Model - SDXL",
     tags=["lora", "model"],
     category="model",
-    version="1.0.3",
+    version="1.0.4",
 )
 class SDXLLoRALoaderInvocation(BaseInvocation):
     """Apply selected lora to unet and text_encoder."""
@@ -400,10 +402,10 @@ class SDXLLoRALoaderInvocation(BaseInvocation):
 
 @invocation(
     "sdxl_lora_collection_loader",
-    title="SDXL LoRA Collection Loader",
+    title="LoRA Collection - SDXL",
     tags=["model"],
     category="model",
-    version="1.1.0",
+    version="1.1.1",
 )
 class SDXLLoRACollectionLoader(BaseInvocation):
     """Applies a collection of SDXL LoRAs to the provided UNet and CLIP models."""
@@ -469,7 +471,9 @@ class SDXLLoRACollectionLoader(BaseInvocation):
         return output
 
 
-@invocation("vae_loader", title="VAE", tags=["vae", "model"], category="model", version="1.0.3")
+@invocation(
+    "vae_loader", title="VAE Model - SD1.5, SDXL, SD3, FLUX", tags=["vae", "model"], category="model", version="1.0.4"
+)
 class VAELoaderInvocation(BaseInvocation):
     """Loads a VAE model, outputting a VaeLoaderOutput"""
 
@@ -496,10 +500,10 @@ class SeamlessModeOutput(BaseInvocationOutput):
 
 @invocation(
     "seamless",
-    title="Seamless",
+    title="Apply Seamless - SD1.5, SDXL",
     tags=["seamless", "model"],
     category="model",
-    version="1.0.1",
+    version="1.0.2",
 )
 class SeamlessModeInvocation(BaseInvocation):
     """Applies the seamless transformation to the Model UNet and VAE."""
@@ -539,7 +543,7 @@ class SeamlessModeInvocation(BaseInvocation):
         return SeamlessModeOutput(unet=unet, vae=vae)
 
 
-@invocation("freeu", title="FreeU", tags=["freeu"], category="unet", version="1.0.1")
+@invocation("freeu", title="Apply FreeU - SD1.5, SDXL", tags=["freeu"], category="unet", version="1.0.2")
 class FreeUInvocation(BaseInvocation):
     """
     Applies FreeU to the UNet. Suggested values (b1/b2/s1/s2):

invokeai/app/invocations/noise.py

@@ -72,10 +72,10 @@ class NoiseOutput(BaseInvocationOutput):
 
 @invocation(
     "noise",
-    title="Noise",
+    title="Create Latent Noise",
     tags=["latents", "noise"],
     category="latents",
-    version="1.0.2",
+    version="1.0.3",
 )
 class NoiseInvocation(BaseInvocation):
     """Generates latent noise."""

invokeai/app/invocations/sd3_denoise.py

@@ -32,10 +32,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "sd3_denoise",
-    title="SD3 Denoise",
+    title="Denoise - SD3",
     tags=["image", "sd3"],
     category="image",
-    version="1.1.0",
+    version="1.1.1",
     classification=Classification.Prototype,
 )
 class SD3DenoiseInvocation(BaseInvocation, WithMetadata, WithBoard):

invokeai/app/invocations/sd3_image_to_latents.py

@@ -21,10 +21,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "sd3_i2l",
-    title="SD3 Image to Latents",
+    title="Image to Latents - SD3",
     tags=["image", "latents", "vae", "i2l", "sd3"],
     category="image",
-    version="1.0.0",
+    version="1.0.1",
     classification=Classification.Prototype,
 )
 class SD3ImageToLatentsInvocation(BaseInvocation, WithMetadata, WithBoard):

invokeai/app/invocations/sd3_latents_to_image.py

@@ -24,10 +24,10 @@ from invokeai.backend.util.devices import TorchDevice
 
 @invocation(
     "sd3_l2i",
-    title="SD3 Latents to Image",
+    title="Latents to Image - SD3",
     tags=["latents", "image", "vae", "l2i", "sd3"],
     category="latents",
-    version="1.3.1",
+    version="1.3.2",
 )
 class SD3LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
     """Generates an image from latents."""

invokeai/app/invocations/sd3_model_loader.py

@@ -30,10 +30,10 @@ class Sd3ModelLoaderOutput(BaseInvocationOutput):
 
 @invocation(
     "sd3_model_loader",
-    title="SD3 Main Model",
+    title="Main Model - SD3",
     tags=["model", "sd3"],
     category="model",
-    version="1.0.0",
+    version="1.0.1",
     classification=Classification.Prototype,
 )
 class Sd3ModelLoaderInvocation(BaseInvocation):

invokeai/app/invocations/sd3_text_encoder.py

@@ -29,10 +29,10 @@ SD3_T5_MAX_SEQ_LEN = 256
 
 @invocation(
     "sd3_text_encoder",
-    title="SD3 Text Encoding",
+    title="Prompt - SD3",
     tags=["prompt", "conditioning", "sd3"],
     category="conditioning",
-    version="1.0.0",
+    version="1.0.1",
     classification=Classification.Prototype,
 )
 class Sd3TextEncoderInvocation(BaseInvocation):

invokeai/app/invocations/sdxl.py

@@ -24,7 +24,7 @@ class SDXLRefinerModelLoaderOutput(BaseInvocationOutput):
     vae: VAEField = OutputField(description=FieldDescriptions.vae, title="VAE")
 
 
-@invocation("sdxl_model_loader", title="SDXL Main Model", tags=["model", "sdxl"], category="model", version="1.0.3")
+@invocation("sdxl_model_loader", title="Main Model - SDXL", tags=["model", "sdxl"], category="model", version="1.0.4")
 class SDXLModelLoaderInvocation(BaseInvocation):
     """Loads an sdxl base model, outputting its submodels."""
 
@@ -58,10 +58,10 @@ class SDXLModelLoaderInvocation(BaseInvocation):
 
 @invocation(
     "sdxl_refiner_model_loader",
-    title="SDXL Refiner Model",
+    title="Refiner Model - SDXL",
     tags=["model", "sdxl", "refiner"],
     category="model",
-    version="1.0.3",
+    version="1.0.4",
 )
 class SDXLRefinerModelLoaderInvocation(BaseInvocation):
     """Loads an sdxl refiner model, outputting its submodels."""

invokeai/app/invocations/t2i_adapter.py

@@ -45,7 +45,11 @@ class T2IAdapterOutput(BaseInvocationOutput):
 
 
 @invocation(
-    "t2i_adapter", title="T2I-Adapter", tags=["t2i_adapter", "control"], category="t2i_adapter", version="1.0.3"
+    "t2i_adapter",
+    title="T2I-Adapter - SD1.5, SDXL",
+    tags=["t2i_adapter", "control"],
+    category="t2i_adapter",
+    version="1.0.4",
 )
 class T2IAdapterInvocation(BaseInvocation):
     """Collects T2I-Adapter info to pass to other nodes."""

invokeai/app/invocations/tiled_multi_diffusion_denoise_latents.py

@@ -53,11 +53,11 @@ def crop_controlnet_data(control_data: ControlNetData, latent_region: TBLR) -> C
 
 @invocation(
     "tiled_multi_diffusion_denoise_latents",
-    title="Tiled Multi-Diffusion Denoise Latents",
+    title="Tiled Multi-Diffusion Denoise - SD1.5, SDXL",
     tags=["upscale", "denoise"],
     category="latents",
     classification=Classification.Beta,
-    version="1.0.0",
+    version="1.0.1",
 )
 class TiledMultiDiffusionDenoiseLatents(BaseInvocation):
     """Tiled Multi-Diffusion denoising.

invokeai/app/services/session_queue/session_queue_common.py

@@ -570,7 +570,10 @@ ValueToInsertTuple: TypeAlias = tuple[
     str | None,  # destination (optional)
     int | None,  # retried_from_item_id (optional, this is always None for new items)
 ]
-"""A type alias for the tuple of values to insert into the session queue table."""
+"""A type alias for the tuple of values to insert into the session queue table.
+
+**If you change this, be sure to update the `enqueue_batch` and `retry_items_by_id` methods in the session queue service!**
+"""
 
 
 def prepare_values_to_insert(

invokeai/app/services/session_queue/session_queue_sqlite.py

@@ -27,6 +27,7 @@ from invokeai.app.services.session_queue.session_queue_common import (
     SessionQueueItemDTO,
     SessionQueueItemNotFoundError,
     SessionQueueStatus,
+    ValueToInsertTuple,
     calc_session_count,
     prepare_values_to_insert,
 )
@@ -689,7 +690,7 @@ class SqliteSessionQueue(SessionQueueBase):
         """Retries the given queue items"""
         try:
             cursor = self._conn.cursor()
-            values_to_insert: list[tuple] = []
+            values_to_insert: list[ValueToInsertTuple] = []
             retried_item_ids: list[int] = []
 
             for item_id in item_ids:
@@ -715,16 +716,16 @@ class SqliteSessionQueue(SessionQueueBase):
                     else queue_item.item_id
                 )
 
-                value_to_insert = (
+                value_to_insert: ValueToInsertTuple = (
                     queue_item.queue_id,
-                    queue_item.batch_id,
-                    queue_item.destination,
-                    field_values_json,
-                    queue_item.origin,
-                    queue_item.priority,
-                    workflow_json,
                     cloned_session_json,
                     cloned_session.id,
+                    queue_item.batch_id,
+                    field_values_json,
+                    queue_item.priority,
+                    workflow_json,
+                    queue_item.origin,
+                    queue_item.destination,
                     retried_from_item_id,
                 )
                 values_to_insert.append(value_to_insert)

invokeai/app/services/workflow_records/default_workflows/ESRGAN Upscaling with Canny ControlNet.json

@@ -1,11 +1,11 @@
 {
   "id": "default_686bb1d0-d086-4c70-9fa3-2f600b922023",
-  "name": "ESRGAN Upscaling with Canny ControlNet",
+  "name": "Upscaler - SD1.5, ESRGAN",
   "author": "InvokeAI",
-  "description": "Sample workflow for using Upscaling with ControlNet with SD1.5",
+  "description": "Sample workflow for using ESRGAN to upscale with ControlNet with SD1.5",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "upscaling, controlnet, default",
+  "tags": "sd1.5, upscaling, control",
   "notes": "",
   "exposedFields": [
     {
@@ -185,14 +185,7 @@
           },
           "control_model": {
             "name": "control_model",
-            "label": "Control Model (select Canny)",
-            "value": {
-              "key": "a7b9c76f-4bc5-42aa-b918-c1c458a5bb24",
-              "hash": "blake3:260c7f8e10aefea9868cfc68d89970e91033bd37132b14b903e70ee05ebf530e",
-              "name": "sd-controlnet-canny",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "Control Model (select Canny)"
           },
           "control_weight": {
             "name": "control_weight",
@@ -295,14 +288,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "",
-            "value": {
-              "key": "5cd43ca0-dd0a-418d-9f7e-35b2b9d5e106",
-              "hash": "blake3:6987f323017f597213cc3264250edf57056d21a40a0a85d83a1a33a7d44dc41a",
-              "name": "Deliberate_v5",
-              "base": "sd-1",
-              "type": "main"
-            }
+            "label": ""
           }
         },
         "isOpen": true,
@@ -849,4 +835,4 @@
       "targetHandle": "image_resolution"
     }
   ]
-}
+}

invokeai/app/services/workflow_records/default_workflows/FLUX Image to Image.json

@@ -1,11 +1,11 @@
 {
   "id": "default_cbf0e034-7b54-4b2c-b670-3b1e2e4b4a88",
-  "name": "FLUX Image to Image",
+  "name": "Image to Image - FLUX",
   "author": "InvokeAI",
   "description": "A simple image-to-image workflow using a FLUX dev model. ",
   "version": "1.1.0",
   "contact": "",
-  "tags": "image2image, flux, image-to-image, image to image",
+  "tags": "flux, image to image",
   "notes": "Prerequisite model downloads: T5 Encoder, CLIP-L Encoder, and FLUX VAE. Quantized and un-quantized versions can be found in the starter models tab within your Model Manager. We recommend using FLUX dev models for image-to-image workflows. The image-to-image performance with FLUX schnell models is poor.",
   "exposedFields": [
     {
@@ -201,36 +201,15 @@
           },
           "t5_encoder_model": {
             "name": "t5_encoder_model",
-            "label": "",
-            "value": {
-              "key": "d18d5575-96b6-4da3-b3d8-eb58308d6705",
-              "hash": "random:f2f9ed74acdfb4bf6fec200e780f6c25f8dd8764a35e65d425d606912fdf573a",
-              "name": "t5_bnb_int8_quantized_encoder",
-              "base": "any",
-              "type": "t5_encoder"
-            }
+            "label": ""
           },
           "clip_embed_model": {
             "name": "clip_embed_model",
-            "label": "",
-            "value": {
-              "key": "5a19d7e5-8d98-43cd-8a81-87515e4b3b4e",
-              "hash": "random:4bd08514c08fb6ff04088db9aeb45def3c488e8b5fd09a35f2cc4f2dc346f99f",
-              "name": "clip-vit-large-patch14",
-              "base": "any",
-              "type": "clip_embed"
-            }
+            "label": ""
           },
           "vae_model": {
             "name": "vae_model",
-            "label": "",
-            "value": {
-              "key": "9172beab-5c1d-43f0-b2f0-6e0b956710d9",
-              "hash": "random:c54dde288e5fa2e6137f1c92e9d611f598049e6f16e360207b6d96c9f5a67ba0",
-              "name": "FLUX.1-schnell_ae",
-              "base": "flux",
-              "type": "vae"
-            }
+            "label": ""
           }
         }
       },

invokeai/app/services/workflow_records/default_workflows/Face Detailer with IP-Adapter & Canny (See Note in Details).json

@@ -1,11 +1,11 @@
 {
   "id": "default_dec5a2e9-f59c-40d9-8869-a056751d79b8",
-  "name": "Face Detailer with IP-Adapter & Canny (See Note in Details)",
+  "name": "Face Detailer - SD1.5",
   "author": "kosmoskatten",
   "description": "A workflow to add detail to and improve faces. This workflow is most effective when used with a model that creates realistic outputs. ",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "face detailer, IP-Adapter, Canny",
+  "tags": "sd1.5, reference image, control",
   "notes": "Set this image as the blur mask: https://i.imgur.com/Gxi61zP.png",
   "exposedFields": [
     {
@@ -136,14 +136,7 @@
           },
           "control_model": {
             "name": "control_model",
-            "label": "Control Model (select canny)",
-            "value": {
-              "key": "5bdaacf7-a7a3-4fb8-b394-cc0ffbb8941d",
-              "hash": "blake3:260c7f8e10aefea9868cfc68d89970e91033bd37132b14b903e70ee05ebf530e",
-              "name": "sd-controlnet-canny",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "Control Model (select canny)"
           },
           "control_weight": {
             "name": "control_weight",
@@ -197,14 +190,7 @@
           },
           "ip_adapter_model": {
             "name": "ip_adapter_model",
-            "label": "IP-Adapter Model (select IP Adapter Face)",
-            "value": {
-              "key": "1cc210bb-4d0a-4312-b36c-b5d46c43768e",
-              "hash": "blake3:3d669dffa7471b357b4df088b99ffb6bf4d4383d5e0ef1de5ec1c89728a3d5a5",
-              "name": "ip_adapter_sd15",
-              "base": "sd-1",
-              "type": "ip_adapter"
-            }
+            "label": "IP-Adapter Model (select IP Adapter Face)"
           },
           "clip_vision_model": {
             "name": "clip_vision_model",

invokeai/app/services/workflow_records/default_workflows/Flux Text to Image.json

@@ -1,11 +1,11 @@
 {
   "id": "default_444fe292-896b-44fd-bfc6-c0b5d220fffc",
-  "name": "FLUX Text to Image",
+  "name": "Text to Image - FLUX",
   "author": "InvokeAI",
   "description": "A simple text-to-image workflow using FLUX dev or schnell models.",
   "version": "1.1.0",
   "contact": "",
-  "tags": "text2image, flux, text to image",
+  "tags": "flux, text to image",
   "notes": "Prerequisite model downloads: T5 Encoder, CLIP-L Encoder, and FLUX VAE. Quantized and un-quantized versions can be found in the starter models tab within your Model Manager. We recommend 4 steps for FLUX schnell models and 30 steps for FLUX dev models.",
   "exposedFields": [
     {
@@ -169,36 +169,15 @@
           },
           "t5_encoder_model": {
             "name": "t5_encoder_model",
-            "label": "",
-            "value": {
-              "key": "d18d5575-96b6-4da3-b3d8-eb58308d6705",
-              "hash": "random:f2f9ed74acdfb4bf6fec200e780f6c25f8dd8764a35e65d425d606912fdf573a",
-              "name": "t5_bnb_int8_quantized_encoder",
-              "base": "any",
-              "type": "t5_encoder"
-            }
+            "label": ""
           },
           "clip_embed_model": {
             "name": "clip_embed_model",
-            "label": "",
-            "value": {
-              "key": "5a19d7e5-8d98-43cd-8a81-87515e4b3b4e",
-              "hash": "random:4bd08514c08fb6ff04088db9aeb45def3c488e8b5fd09a35f2cc4f2dc346f99f",
-              "name": "clip-vit-large-patch14",
-              "base": "any",
-              "type": "clip_embed"
-            }
+            "label": ""
           },
           "vae_model": {
             "name": "vae_model",
-            "label": "",
-            "value": {
-              "key": "9172beab-5c1d-43f0-b2f0-6e0b956710d9",
-              "hash": "random:c54dde288e5fa2e6137f1c92e9d611f598049e6f16e360207b6d96c9f5a67ba0",
-              "name": "FLUX.1-schnell_ae",
-              "base": "flux",
-              "type": "vae"
-            }
+            "label": ""
           }
         }
       },

invokeai/app/services/workflow_records/default_workflows/Multi ControlNet (Canny & Depth).json

@@ -1,11 +1,11 @@
 {
   "id": "default_2d05e719-a6b9-4e64-9310-b875d3b2f9d2",
-  "name": "Multi ControlNet (Canny & Depth)",
+  "name": "Text to Image -  SD1.5, Control",
   "author": "InvokeAI",
   "description": "A sample workflow using canny & depth ControlNets to guide the generation process. ",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "ControlNet, canny, depth",
+  "tags": "sd1.5, control, text to image",
   "notes": "",
   "exposedFields": [
     {
@@ -217,14 +217,7 @@
           },
           "control_model": {
             "name": "control_model",
-            "label": "Control Model (select canny)",
-            "value": {
-              "key": "5bdaacf7-a7a3-4fb8-b394-cc0ffbb8941d",
-              "hash": "blake3:260c7f8e10aefea9868cfc68d89970e91033bd37132b14b903e70ee05ebf530e",
-              "name": "sd-controlnet-canny",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "Control Model (select canny)"
           },
           "control_weight": {
             "name": "control_weight",
@@ -371,14 +364,7 @@
           },
           "control_model": {
             "name": "control_model",
-            "label": "Control Model (select depth)",
-            "value": {
-              "key": "87e8855c-671f-4c9e-bbbb-8ed47ccb4aac",
-              "hash": "blake3:2550bf22a53942dfa28ab2fed9d10d80851112531f44d977168992edf9d0534c",
-              "name": "control_v11f1p_sd15_depth",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "Control Model (select depth)"
           },
           "control_weight": {
             "name": "control_weight",

invokeai/app/services/workflow_records/default_workflows/MultiDiffusion SD1.5.json

@@ -1,11 +1,11 @@
 {
   "id": "default_f96e794f-eb3e-4d01-a960-9b4e43402bcf",
-  "name": "MultiDiffusion SD1.5",
+  "name": "Upscaler - SD1.5, MultiDiffusion",
   "author": "Invoke",
   "description": "A workflow to upscale an input image with tiled upscaling, using SD1.5 based models.",
   "version": "1.0.0",
   "contact": "invoke@invoke.ai",
-  "tags": "tiled, upscaling, sdxl",
+  "tags": "sd1.5, upscaling",
   "notes": "",
   "exposedFields": [
     {
@@ -135,14 +135,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "",
-            "value": {
-              "key": "e7b402e5-62e5-4acb-8c39-bee6bdb758ab",
-              "hash": "c8659e796168d076368256b57edbc1b48d6dafc1712f1bb37cc57c7c06889a6b",
-              "name": "526mix",
-              "base": "sd-1",
-              "type": "main"
-            }
+            "label": ""
           }
         }
       },
@@ -384,21 +377,11 @@
           },
           "image": {
             "name": "image",
-            "label": "Image to Upscale",
-            "value": {
-              "image_name": "ee7009f7-a35d-488b-a2a6-21237ef5ae05.png"
-            }
+            "label": "Image to Upscale"
           },
           "image_to_image_model": {
             "name": "image_to_image_model",
-            "label": "",
-            "value": {
-              "key": "38bb1a29-8ede-42ba-b77f-64b3478896eb",
-              "hash": "blake3:e52fdbee46a484ebe9b3b20ea0aac0a35a453ab6d0d353da00acfd35ce7a91ed",
-              "name": "4xNomosWebPhoto_esrgan",
-              "base": "sdxl",
-              "type": "spandrel_image_to_image"
-            }
+            "label": ""
           },
           "tile_size": {
             "name": "tile_size",
@@ -437,14 +420,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "ControlNet Model - Choose a Tile ControlNet",
-            "value": {
-              "key": "20645e4d-ef97-4c5a-9243-b834a3483925",
-              "hash": "f0812e13758f91baf4e54b7dbb707b70642937d3b2098cd2b94cc36d3eba308e",
-              "name": "tile",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "ControlNet Model - Choose a Tile ControlNet"
           }
         }
       },

invokeai/app/services/workflow_records/default_workflows/MultiDiffusion SDXL.json

@@ -1,11 +1,11 @@
 {
   "id": "default_35658541-6d41-4a20-8ec5-4bf2561faed0",
-  "name": "MultiDiffusion SDXL",
+  "name": "Upscaler - SDXL, MultiDiffusion",
   "author": "Invoke",
   "description": "A workflow to upscale an input image with tiled upscaling, using SDXL based models.",
   "version": "1.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "tiled, upscaling, sdxl",
+  "tags": "sdxl, upscaling",
   "notes": "",
   "exposedFields": [
     {
@@ -341,14 +341,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "ControlNet Model - Choose a Tile ControlNet",
-            "value": {
-              "key": "74f4651f-0ace-4b7b-b616-e98360257797",
-              "hash": "blake3:167a5b84583aaed3e5c8d660b45830e82e1c602743c689d3c27773c6c8b85b4a",
-              "name": "controlnet-tile-sdxl-1.0",
-              "base": "sdxl",
-              "type": "controlnet"
-            }
+            "label": "ControlNet Model - Choose a Tile ControlNet"
           }
         }
       },
@@ -801,14 +794,7 @@
         "inputs": {
           "vae_model": {
             "name": "vae_model",
-            "label": "",
-            "value": {
-              "key": "ff926845-090e-4d46-b81e-30289ee47474",
-              "hash": "9705ab1c31fa96b308734214fb7571a958621c7a9247eed82b7d277145f8d9fa",
-              "name": "VAEFix",
-              "base": "sdxl",
-              "type": "vae"
-            }
+            "label": ""
           }
         }
       },
@@ -832,14 +818,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "SDXL Model",
-            "value": {
-              "key": "ab191f73-68d2-492c-8aec-b438a8cf0f45",
-              "hash": "blake3:2d50e940627e3bf555f015280ec0976d5c1fa100f7bc94e95ffbfc770e98b6fe",
-              "name": "CustomXLv7",
-              "base": "sdxl",
-              "type": "main"
-            }
+            "label": "SDXL Model"
           }
         }
       },

invokeai/app/services/workflow_records/default_workflows/Prompt from File.json

@@ -1,11 +1,11 @@
 {
   "id": "default_d7a1c60f-ca2f-4f90-9e33-75a826ca6d8f",
-  "name": "Prompt from File",
+  "name": "Text to Image - SD1.5, Prompt from File",
   "author": "InvokeAI",
   "description": "Sample workflow using Prompt from File node",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "text2image, prompt from file, default, text to image",
+  "tags": "sd1.5, text to image",
   "notes": "",
   "exposedFields": [
     {
@@ -513,4 +513,4 @@
       "targetHandle": "vae"
     }
   ]
-}
+}

invokeai/app/services/workflow_records/default_workflows/README.md

@@ -10,6 +10,7 @@ _default workflows_ on app startup.
   An exception will be raised during sync if this is not set correctly.
 - Default workflows appear on the "Default Workflows" tab of the Workflow
   Library.
+- Default workflows should not reference any resources that are user-created or installed. That includes images and models. For example, if a default workflow references Juggernaut as an SDXL model, when a user loads the workflow, even if they have a version of Juggernaut installed, it will have a different UUID. They may see a warning. So, it's best to ship default workflows without any references to these types of resources.
 
 After adding or updating default workflows, you **must** start the app up and
 load them to ensure:

invokeai/app/services/workflow_records/default_workflows/SD3.5 Text to Image.json

@@ -1,11 +1,11 @@
 {
   "id": "default_dbe46d95-22aa-43fb-9c16-94400d0ce2fd",
-  "name": "SD3.5 Text to Image",
+  "name": "Text to Image - SD3.5",
   "author": "InvokeAI",
   "description": "Sample text to image workflow for Stable Diffusion 3.5",
   "version": "1.0.0",
   "contact": "invoke@invoke.ai",
-  "tags": "text2image, SD3.5, text to image",
+  "tags": "SD3.5, text to image",
   "notes": "",
   "exposedFields": [
     {
@@ -38,14 +38,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "",
-            "value": {
-              "key": "f7b20be9-92a8-4cfb-bca4-6c3b5535c10b",
-              "hash": "placeholder",
-              "name": "stable-diffusion-3.5-medium",
-              "base": "sd-3",
-              "type": "main"
-            }
+            "label": ""
           },
           "t5_encoder_model": {
             "name": "t5_encoder_model",

invokeai/app/services/workflow_records/default_workflows/Text to Image - SD1.5.json

@@ -5,7 +5,7 @@
   "description": "Sample text to image workflow for Stable Diffusion 1.5/2",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "text2image, SD1.5, SD2, text to image",
+  "tags": "SD1.5, text to image",
   "notes": "",
   "exposedFields": [
     {
@@ -417,4 +417,4 @@
       "targetHandle": "vae"
     }
   ]
-}
+}

invokeai/app/services/workflow_records/default_workflows/Text to Image - SDXL.json

@@ -5,7 +5,7 @@
   "description": "Sample text to image workflow for SDXL",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "text2image, SDXL, text to image",
+  "tags": "SDXL, text to image",
   "notes": "",
   "exposedFields": [
     {
@@ -46,14 +46,7 @@
         "inputs": {
           "vae_model": {
             "name": "vae_model",
-            "label": "VAE (use the FP16 model)",
-            "value": {
-              "key": "f20f9e5c-1bce-4c46-a84d-34ebfa7df069",
-              "hash": "blake3:9705ab1c31fa96b308734214fb7571a958621c7a9247eed82b7d277145f8d9fa",
-              "name": "sdxl-vae-fp16-fix",
-              "base": "sdxl",
-              "type": "vae"
-            }
+            "label": "VAE (use the FP16 model)"
           }
         },
         "isOpen": true,
@@ -203,14 +196,7 @@
         "inputs": {
           "model": {
             "name": "model",
-            "label": "",
-            "value": {
-              "key": "4a63b226-e8ff-4da4-854e-0b9f04b562ba",
-              "hash": "blake3:d279309ea6e5ee6e8fd52504275865cc280dac71cbf528c5b07c98b888bddaba",
-              "name": "dreamshaper-xl-v2-turbo",
-              "base": "sdxl",
-              "type": "main"
-            }
+            "label": ""
           }
         },
         "isOpen": true,
@@ -715,4 +701,4 @@
       "targetHandle": "style"
     }
   ]
-}
+}

invokeai/app/services/workflow_records/default_workflows/Text to Image with LoRA.json

@@ -1,11 +1,11 @@
 {
   "id": "default_e71d153c-2089-43c7-bd2c-f61f37d4c1c1",
-  "name": "Text to Image with LoRA",
+  "name": "Text to Image - SD1.5, LoRA",
   "author": "InvokeAI",
   "description": "Simple text to image workflow with a LoRA",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "text to image, lora, text to image",
+  "tags": "sd1.5, text to image, lora",
   "notes": "",
   "exposedFields": [
     {

invokeai/app/services/workflow_records/default_workflows/Tiled Upscaling (Beta).json

@@ -1,11 +1,11 @@
 {
   "id": "default_43b0d7f7-6a12-4dcf-a5a4-50c940cbee29",
-  "name": "Tiled Upscaling (Beta)",
+  "name": "Upscaler - SD1.5, Tiled",
   "author": "Invoke",
   "description": "A workflow to upscale an input image with tiled upscaling. ",
   "version": "2.1.0",
   "contact": "invoke@invoke.ai",
-  "tags": "tiled, upscaling, sd1.5",
+  "tags": "sd1.5, upscaling",
   "notes": "",
   "exposedFields": [
     {
@@ -86,14 +86,7 @@
           },
           "ip_adapter_model": {
             "name": "ip_adapter_model",
-            "label": "IP-Adapter Model (select ip_adapter_sd15)",
-            "value": {
-              "key": "1cc210bb-4d0a-4312-b36c-b5d46c43768e",
-              "hash": "blake3:3d669dffa7471b357b4df088b99ffb6bf4d4383d5e0ef1de5ec1c89728a3d5a5",
-              "name": "ip_adapter_sd15",
-              "base": "sd-1",
-              "type": "ip_adapter"
-            }
+            "label": "IP-Adapter Model (select ip_adapter_sd15)"
           },
           "clip_vision_model": {
             "name": "clip_vision_model",
@@ -201,14 +194,7 @@
           },
           "control_model": {
             "name": "control_model",
-            "label": "Control Model (select contro_v11f1e_sd15_tile)",
-            "value": {
-              "key": "773843c8-db1f-4502-8f65-59782efa7960",
-              "hash": "blake3:f0812e13758f91baf4e54b7dbb707b70642937d3b2098cd2b94cc36d3eba308e",
-              "name": "control_v11f1e_sd15_tile",
-              "base": "sd-1",
-              "type": "controlnet"
-            }
+            "label": "Control Model (select control_v11f1e_sd15_tile)"
           },
           "control_weight": {
             "name": "control_weight",
@@ -1816,4 +1802,4 @@
       "targetHandle": "unet"
     }
   ]
-}
+}

invokeai/app/services/workflow_thumbnails/workflow_thumbnails_base.py

@@ -8,12 +8,12 @@ class WorkflowThumbnailServiceBase(ABC):
     """Base class for workflow thumbnail services"""
 
     @abstractmethod
-    def get_path(self, workflow_id: str) -> Path:
+    def get_path(self, workflow_id: str, with_hash: bool = True) -> Path:
         """Gets the path to a workflow thumbnail"""
         pass
 
     @abstractmethod
-    def get_url(self, workflow_id: str) -> str | None:
+    def get_url(self, workflow_id: str, with_hash: bool = True) -> str | None:
         """Gets the URL of a workflow thumbnail"""
         pass
 

invokeai/app/services/workflow_thumbnails/workflow_thumbnails_disk.py

@@ -41,7 +41,7 @@ class WorkflowThumbnailFileStorageDisk(WorkflowThumbnailServiceBase):
         except Exception as e:
             raise WorkflowThumbnailFileSaveException from e
 
-    def get_path(self, workflow_id: str) -> Path:
+    def get_path(self, workflow_id: str, with_hash: bool = True) -> Path:
         workflow = self._invoker.services.workflow_records.get(workflow_id).workflow
         if workflow.meta.category is WorkflowCategory.Default:
             default_thumbnails_dir = Path(__file__).parent / Path("default_workflow_thumbnails")
@@ -51,7 +51,7 @@ class WorkflowThumbnailFileStorageDisk(WorkflowThumbnailServiceBase):
 
         return path
 
-    def get_url(self, workflow_id: str) -> str | None:
+    def get_url(self, workflow_id: str, with_hash: bool = True) -> str | None:
         path = self.get_path(workflow_id)
         if not self._validate_path(path):
             return
@@ -59,7 +59,8 @@ class WorkflowThumbnailFileStorageDisk(WorkflowThumbnailServiceBase):
         url = self._invoker.services.urls.get_workflow_thumbnail_url(workflow_id)
 
         # The image URL never changes, so we must add random query string to it to prevent caching
-        url += f"?{uuid_string()}"
+        if with_hash:
+            url += f"?{uuid_string()}"
 
         return url
 

invokeai/backend/llava_onevision_model.py

@@ -0,0 +1,49 @@
+from pathlib import Path
+from typing import Optional
+
+import torch
+from PIL.Image import Image
+from transformers import AutoProcessor, LlavaOnevisionForConditionalGeneration, LlavaOnevisionProcessor
+
+from invokeai.backend.raw_model import RawModel
+
+
+class LlavaOnevisionModel(RawModel):
+    def __init__(self, vllm_model: LlavaOnevisionForConditionalGeneration, processor: LlavaOnevisionProcessor):
+        self._vllm_model = vllm_model
+        self._processor = processor
+
+    @classmethod
+    def load_from_path(cls, path: str | Path):
+        vllm_model = LlavaOnevisionForConditionalGeneration.from_pretrained(path, local_files_only=True)
+        assert isinstance(vllm_model, LlavaOnevisionForConditionalGeneration)
+        processor = AutoProcessor.from_pretrained(path, local_files_only=True)
+        assert isinstance(processor, LlavaOnevisionProcessor)
+        return cls(vllm_model, processor)
+
+    def run(self, prompt: str, images: list[Image], device: torch.device, dtype: torch.dtype) -> str:
+        # TODO(ryand): Tune the max number of images that are useful for the model.
+        if len(images) > 3:
+            raise ValueError(
+                f"{len(images)} images were provided as input to the LLaVA OneVision model. "
+                "Pass <=3 images for good performance."
+            )
+
+        # Define a chat history and use `apply_chat_template` to get correctly formatted prompt.
+        # "content" is a list of dicts with types "text" or "image".
+        content = [{"type": "text", "text": prompt}]
+        # Add the correct number of images.
+        for _ in images:
+            content.append({"type": "image"})
+
+        conversation = [{"role": "user", "content": content}]
+        prompt = self._processor.apply_chat_template(conversation, add_generation_prompt=True)
+        inputs = self._processor(images=images or None, text=prompt, return_tensors="pt").to(device=device, dtype=dtype)
+        output = self._vllm_model.generate(**inputs, max_new_tokens=400, do_sample=False)
+        output_str: str = self._processor.decode(output[0][2:], skip_special_tokens=True)
+        # The output_str will include the prompt, so we extract the response.
+        response = output_str.split("assistant\n", 1)[1].strip()
+        return response
+
+    def to(self, device: Optional[torch.device] = None, dtype: Optional[torch.dtype] = None) -> None:
+        self._vllm_model.to(device=device, dtype=dtype)

invokeai/backend/model_manager/config.py

@@ -90,6 +90,7 @@ class ModelType(str, Enum):
     SpandrelImageToImage = "spandrel_image_to_image"
     SigLIP = "siglip"
     FluxRedux = "flux_redux"
+    LlavaOnevision = "llava_onevision"
 
 
 class SubModelType(str, Enum):
@@ -624,6 +625,13 @@ class FluxReduxConfig(LegacyProbeMixin, ModelConfigBase):
     format: Literal[ModelFormat.Checkpoint] = ModelFormat.Checkpoint
 
 
+class LlavaOnevisionConfig(DiffusersConfigBase, LegacyProbeMixin, ModelConfigBase):
+    """Model config for Llava Onevision models."""
+
+    type: Literal[ModelType.LlavaOnevision] = ModelType.LlavaOnevision
+    format: Literal[ModelFormat.Diffusers] = ModelFormat.Diffusers
+
+
 def get_model_discriminator_value(v: Any) -> str:
     """
     Computes the discriminator value for a model config.
@@ -690,6 +698,7 @@ AnyModelConfig = Annotated[
         Annotated[CLIPGEmbedDiffusersConfig, CLIPGEmbedDiffusersConfig.get_tag()],
         Annotated[SigLIPConfig, SigLIPConfig.get_tag()],
         Annotated[FluxReduxConfig, FluxReduxConfig.get_tag()],
+        Annotated[LlavaOnevisionConfig, LlavaOnevisionConfig.get_tag()],
     ],
     Discriminator(get_model_discriminator_value),
 ]

invokeai/backend/model_manager/legacy_probe.py

@@ -141,6 +141,7 @@ class ModelProbe(object):
         "SD3Transformer2DModel": ModelType.Main,
         "CLIPTextModelWithProjection": ModelType.CLIPEmbed,
         "SiglipModel": ModelType.SigLIP,
+        "LlavaOnevisionForConditionalGeneration": ModelType.LlavaOnevision,
     }
 
     TYPE2VARIANT: Dict[ModelType, Callable[[str], Optional[AnyVariant]]] = {ModelType.CLIPEmbed: get_clip_variant_type}
@@ -767,6 +768,11 @@ class FluxReduxCheckpointProbe(CheckpointProbeBase):
         return BaseModelType.Flux
 
 
+class LlavaOnevisionCheckpointProbe(CheckpointProbeBase):
+    def get_base_type(self) -> BaseModelType:
+        raise NotImplementedError()
+
+
 ########################################################
 # classes for probing folders
 #######################################################
@@ -1047,6 +1053,11 @@ class FluxReduxFolderProbe(FolderProbeBase):
         raise NotImplementedError()
 
 
+class LlaveOnevisionFolderProbe(FolderProbeBase):
+    def get_base_type(self) -> BaseModelType:
+        return BaseModelType.Any
+
+
 class T2IAdapterFolderProbe(FolderProbeBase):
     def get_base_type(self) -> BaseModelType:
         config_file = self.model_path / "config.json"
@@ -1082,6 +1093,7 @@ ModelProbe.register_probe("diffusers", ModelType.T2IAdapter, T2IAdapterFolderPro
 ModelProbe.register_probe("diffusers", ModelType.SpandrelImageToImage, SpandrelImageToImageFolderProbe)
 ModelProbe.register_probe("diffusers", ModelType.SigLIP, SigLIPFolderProbe)
 ModelProbe.register_probe("diffusers", ModelType.FluxRedux, FluxReduxFolderProbe)
+ModelProbe.register_probe("diffusers", ModelType.LlavaOnevision, LlaveOnevisionFolderProbe)
 
 ModelProbe.register_probe("checkpoint", ModelType.Main, PipelineCheckpointProbe)
 ModelProbe.register_probe("checkpoint", ModelType.VAE, VaeCheckpointProbe)
@@ -1095,5 +1107,6 @@ ModelProbe.register_probe("checkpoint", ModelType.T2IAdapter, T2IAdapterCheckpoi
 ModelProbe.register_probe("checkpoint", ModelType.SpandrelImageToImage, SpandrelImageToImageCheckpointProbe)
 ModelProbe.register_probe("checkpoint", ModelType.SigLIP, SigLIPCheckpointProbe)
 ModelProbe.register_probe("checkpoint", ModelType.FluxRedux, FluxReduxCheckpointProbe)
+ModelProbe.register_probe("checkpoint", ModelType.LlavaOnevision, LlavaOnevisionCheckpointProbe)
 
 ModelProbe.register_probe("onnx", ModelType.ONNX, ONNXFolderProbe)

invokeai/backend/model_manager/load/model_loaders/llava_onevision.py

@@ -0,0 +1,32 @@
+from pathlib import Path
+from typing import Optional
+
+from invokeai.backend.llava_onevision_model import LlavaOnevisionModel
+from invokeai.backend.model_manager.config import (
+    AnyModel,
+    AnyModelConfig,
+    BaseModelType,
+    ModelFormat,
+    ModelType,
+    SubModelType,
+)
+from invokeai.backend.model_manager.load.load_default import ModelLoader
+from invokeai.backend.model_manager.load.model_loader_registry import ModelLoaderRegistry
+
+
+@ModelLoaderRegistry.register(base=BaseModelType.Any, type=ModelType.LlavaOnevision, format=ModelFormat.Diffusers)
+class LlavaOnevisionModelLoader(ModelLoader):
+    """Class for loading LLaVA Onevision VLLM models."""
+
+    def _load_model(
+        self,
+        config: AnyModelConfig,
+        submodel_type: Optional[SubModelType] = None,
+    ) -> AnyModel:
+        if submodel_type is not None:
+            raise ValueError("Unexpected submodel requested for LLaVA OneVision model.")
+
+        model_path = Path(config.path)
+        model = LlavaOnevisionModel.load_from_path(model_path)
+        model.to(dtype=self._torch_dtype)
+        return model

invokeai/backend/model_manager/starter_models.py

@@ -614,6 +614,16 @@ flux_redux = StarterModel(
 )
 # endregion
 
+# region LlavaOnevisionModel
+llava_onevision = StarterModel(
+    name="LLaVA Onevision Qwen2 0.5B",
+    base=BaseModelType.Any,
+    source="llava-hf/llava-onevision-qwen2-0.5b-ov-hf",
+    description="LLaVA Onevision VLLM model",
+    type=ModelType.LlavaOnevision,
+)
+# endregion
+
 # List of starter models, displayed on the frontend.
 # The order/sort of this list is not changed by the frontend - set it how you want it here.
 STARTER_MODELS: list[StarterModel] = [
@@ -683,6 +693,7 @@ STARTER_MODELS: list[StarterModel] = [
     clip_l_encoder,
     siglip,
     flux_redux,
+    llava_onevision,
 ]
 
 sd1_bundle: list[StarterModel] = [

invokeai/frontend/web/public/locales/en.json

@@ -116,6 +116,8 @@
         "dontShowMeThese": "Don't show me these",
         "editor": "Editor",
         "error": "Error",
+        "error_withCount_one": "{{count}} error",
+        "error_withCount_other": "{{count}} errors",
         "file": "File",
         "folder": "Folder",
         "format": "format",
@@ -844,6 +846,7 @@
         "starterModels": "Starter Models",
         "starterModelsInModelManager": "Starter Models can be found in Model Manager",
         "controlLora": "Control LoRA",
+        "llavaOnevision": "LLaVA OneVision",
         "syncModels": "Sync Models",
         "textualInversions": "Textual Inversions",
         "triggerPhrases": "Trigger Phrases",
@@ -1205,6 +1208,7 @@
         "informationalPopoversDisabled": "Informational Popovers Disabled",
         "informationalPopoversDisabledDesc": "Informational popovers have been disabled. Enable them in Settings.",
         "enableModelDescriptions": "Enable Model Descriptions in Dropdowns",
+        "enableHighlightFocusedRegions": "Highlight Focused Regions",
         "modelDescriptionsDisabled": "Model Descriptions in Dropdowns Disabled",
         "modelDescriptionsDisabledDesc": "Model descriptions in dropdowns have been disabled. Enable them in Settings.",
         "enableInvisibleWatermark": "Enable Invisible Watermark",
@@ -1765,7 +1769,9 @@
             "containerPlaceholder": "Empty Container",
             "headingPlaceholder": "Empty Heading",
             "textPlaceholder": "Empty Text",
-            "workflowBuilderAlphaWarning": "The workflow builder is currently in alpha. There may be breaking changes before the stable release."
+            "workflowBuilderAlphaWarning": "The workflow builder is currently in alpha. There may be breaking changes before the stable release.",
+            "minimum": "Minimum",
+            "maximum": "Maximum"
         }
     },
     "controlLayers": {
@@ -2317,6 +2323,7 @@
     "newUserExperience": {
         "toGetStartedLocal": "To get started, make sure to download or import models needed to run Invoke. Then, enter a prompt in the box and click <StrongComponent>Invoke</StrongComponent> to generate your first image. Select a prompt template to improve results. You can choose to save your images directly to the <StrongComponent>Gallery</StrongComponent> or edit them to the <StrongComponent>Canvas</StrongComponent>.",
         "toGetStarted": "To get started, enter a prompt in the box and click <StrongComponent>Invoke</StrongComponent> to generate your first image. Select a prompt template to improve results. You can choose to save your images directly to the <StrongComponent>Gallery</StrongComponent> or edit them to the <StrongComponent>Canvas</StrongComponent>.",
+        "toGetStartedWorkflow": "To get started, fill in the fields on the left and press <StrongComponent>Invoke</StrongComponent> to generate your image. Want to explore more workflows? Click the <StrongComponent>folder icon</StrongComponent> next to the workflow title to see a list of other templates you can try.",
         "gettingStartedSeries": "Want more guidance? Check out our <LinkComponent>Getting Started Series</LinkComponent> for tips on unlocking the full potential of the Invoke Studio.",
         "lowVRAMMode": "For best performance, follow our <LinkComponent>Low VRAM guide</LinkComponent>.",
         "noModelsInstalled": "It looks like you don't have any models installed! You can <DownloadStarterModelsButton>download a starter model bundle</DownloadStarterModelsButton> or <ImportModelsButton>import models</ImportModelsButton>."
@@ -2324,8 +2331,8 @@
     "whatsNew": {
         "whatsNewInInvoke": "What's New in Invoke",
         "items": [
-            "Memory Management: New setting for users with Nvidia GPUs to reduce VRAM usage.",
-            "Performance: Continued improvements to overall application performance and responsiveness."
+            "Workflows: New and improved Workflow Library.",
+            "FLUX: Support for FLUX Redux in Workflows and Canvas."
         ],
         "readReleaseNotes": "Read Release Notes",
         "watchRecentReleaseVideos": "Watch Recent Release Videos",

invokeai/frontend/web/public/locales/it.json

@@ -110,7 +110,10 @@
         "layout": "Schema",
         "row": "Riga",
         "column": "Colonna",
-        "saveChanges": "Salva modifiche"
+        "saveChanges": "Salva modifiche",
+        "error_withCount_one": "{{count}} errore",
+        "error_withCount_many": "{{count}} errori",
+        "error_withCount_other": "{{count}} errori"
     },
     "gallery": {
         "galleryImageSize": "Dimensione dell'immagine",
@@ -1771,20 +1774,24 @@
             "divider": "Divisore",
             "container": "Contenitore",
             "text": "Testo",
-            "numberInput": "Ingresso numerico"
+            "numberInput": "Ingresso numerico",
+            "containerRowLayout": "Contenitore (disposizione riga)",
+            "containerColumnLayout": "Contenitore (disposizione colonna)"
         },
         "loadMore": "Carica altro",
         "searchPlaceholder": "Cerca per nome, descrizione o etichetta",
         "filterByTags": "Filtra per etichetta",
         "shared": "Condiviso",
         "browseWorkflows": "Sfoglia i flussi di lavoro",
-        "resetTags": "Reimposta le etichette",
         "allLoaded": "Tutti i flussi di lavoro caricati",
         "saveChanges": "Salva modifiche",
         "yourWorkflows": "I tuoi flussi di lavoro",
         "recentlyOpened": "Aperto di recente",
         "workflowThumbnail": "Miniatura del flusso di lavoro",
-        "private": "Privato"
+        "private": "Privato",
+        "deselectAll": "Deseleziona tutto",
+        "noRecentWorkflows": "Nessun flusso di lavoro recente",
+        "view": "Visualizza"
     },
     "accordions": {
         "compositing": {
@@ -2334,7 +2341,8 @@
         "toGetStarted": "Per iniziare, inserisci un prompt nella casella e fai clic su <StrongComponent>Invoke</StrongComponent> per generare la tua prima immagine. Seleziona un modello di prompt per migliorare i risultati. Puoi scegliere di salvare le tue immagini direttamente nella <StrongComponent>Galleria</StrongComponent> o modificarle nella <StrongComponent>Tela</StrongComponent>.",
         "noModelsInstalled": "Sembra che non hai installato alcun modello! Puoi <DownloadStarterModelsButton>scaricare un pacchetto di modelli di avvio</DownloadStarterModelsButton> o <ImportModelsButton>importare modelli</ImportModelsButton>.",
         "toGetStartedLocal": "Per iniziare, assicurati di scaricare o importare i modelli necessari per eseguire Invoke. Quindi, inserisci un prompt nella casella e fai clic su <StrongComponent>Invoke</StrongComponent> per generare la tua prima immagine. Seleziona un modello di prompt per migliorare i risultati. Puoi scegliere di salvare le tue immagini direttamente nella <StrongComponent>Galleria</StrongComponent> o modificarle nella <StrongComponent>Tela</StrongComponent>.",
-        "lowVRAMMode": "Per prestazioni ottimali, segui la nostra <LinkComponent>guida per bassa VRAM</LinkComponent>."
+        "lowVRAMMode": "Per prestazioni ottimali, segui la nostra <LinkComponent>guida per bassa VRAM</LinkComponent>.",
+        "toGetStartedWorkflow": "Per iniziare, compila i campi a sinistra e premi <StrongComponent>Invoke</StrongComponent> per generare la tua immagine. Vuoi esplorare altri flussi di lavoro? Fai clic sull'<StrongComponent>icona della cartella</StrongComponent> accanto al titolo del flusso di lavoro per visualizzare un elenco di altri modelli che puoi provare."
     },
     "whatsNew": {
         "whatsNewInInvoke": "Novità in Invoke",

invokeai/frontend/web/public/locales/vi.json

@@ -2300,7 +2300,6 @@
         "filterByTags": "Lọc Theo Nhãn",
         "recentlyOpened": "Mở Gần Đây",
         "private": "Cá Nhân",
-        "resetTags": "Khởi Động Lại Nhãn",
         "loadMore": "Tải Thêm"
     },
     "upscaling": {

invokeai/frontend/web/src/app/components/InvokeAIUI.tsx

@@ -19,11 +19,13 @@ import { $store } from 'app/store/nanostores/store';
 import { createStore } from 'app/store/store';
 import type { PartialAppConfig } from 'app/types/invokeai';
 import Loading from 'common/components/Loading/Loading';
-import type { WorkflowTagCategory } from 'features/nodes/store/workflowLibrarySlice';
+import type { WorkflowSortOption, WorkflowTagCategory } from 'features/nodes/store/workflowLibrarySlice';
 import {
   $workflowLibraryCategoriesOptions,
+  $workflowLibrarySortOptions,
   $workflowLibraryTagCategoriesOptions,
   DEFAULT_WORKFLOW_LIBRARY_CATEGORIES,
+  DEFAULT_WORKFLOW_LIBRARY_SORT_OPTIONS,
   DEFAULT_WORKFLOW_LIBRARY_TAG_CATEGORIES,
 } from 'features/nodes/store/workflowLibrarySlice';
 import type { WorkflowCategory } from 'features/nodes/types/workflow';
@@ -55,6 +57,7 @@ interface Props extends PropsWithChildren {
   logo?: ReactNode;
   workflowCategories?: WorkflowCategory[];
   workflowTagCategories?: WorkflowTagCategory[];
+  workflowSortOptions?: WorkflowSortOption[];
   loggingOverrides?: LoggingOverrides;
 }
 
@@ -76,6 +79,7 @@ const InvokeAIUI = ({
   logo,
   workflowCategories,
   workflowTagCategories,
+  workflowSortOptions,
   loggingOverrides,
 }: Props) => {
   useLayoutEffect(() => {
@@ -221,6 +225,16 @@ const InvokeAIUI = ({
     };
   }, [workflowTagCategories]);
 
+  useEffect(() => {
+    if (workflowSortOptions) {
+      $workflowLibrarySortOptions.set(workflowSortOptions);
+    }
+
+    return () => {
+      $workflowLibrarySortOptions.set(DEFAULT_WORKFLOW_LIBRARY_SORT_OPTIONS);
+    };
+  }, [workflowSortOptions]);
+
   useEffect(() => {
     if (socketOptions) {
       $socketOptions.set(socketOptions);

invokeai/frontend/web/src/app/store/use-debounced-app-selector.ts

@@ -0,0 +1,43 @@
+import type { Selector } from '@reduxjs/toolkit';
+import { useAppStore } from 'app/store/nanostores/store';
+import type { RootState } from 'app/store/store';
+import { useEffect, useState } from 'react';
+
+/**
+ * A hook that returns a debounced value from the app state.
+ *
+ * @param selector The redux selector
+ * @param debounceMs The debounce time in milliseconds
+ * @returns The debounced value
+ */
+export const useDebouncedAppSelector = <T>(selector: Selector<RootState, T>, debounceMs: number = 300) => {
+  const store = useAppStore();
+  const [value, setValue] = useState<T>(() => selector(store.getState()));
+
+  useEffect(() => {
+    let prevValue = selector(store.getState());
+    let timeout: number | null = null;
+
+    const unsubscribe = store.subscribe(() => {
+      const value = selector(store.getState());
+      if (value !== prevValue) {
+        if (timeout !== null) {
+          window.clearTimeout(timeout);
+        }
+        timeout = window.setTimeout(() => {
+          setValue(value);
+          prevValue = value;
+        }, debounceMs);
+      }
+    });
+
+    return () => {
+      unsubscribe();
+      if (timeout !== null) {
+        window.clearTimeout(timeout);
+      }
+    };
+  }, [debounceMs, selector, store]);
+
+  return value;
+};

invokeai/frontend/web/src/app/types/invokeai.ts

@@ -27,7 +27,8 @@ export type AppFeature =
   | 'bulkDownload'
   | 'starterModels'
   | 'hfToken'
-  | 'retryQueueItem';
+  | 'retryQueueItem'
+  | 'cancelAndClearAll';
 /**
  * A disable-able Stable Diffusion feature
  */

invokeai/frontend/web/src/common/components/FocusRegionWrapper.tsx

@@ -0,0 +1,56 @@
+import { Box, type BoxProps, type SystemStyleObject } from '@invoke-ai/ui-library';
+import { useAppSelector } from 'app/store/storeHooks';
+import { type FocusRegionName, useFocusRegion, useIsRegionFocused } from 'common/hooks/focus';
+import { selectSystemShouldEnableHighlightFocusedRegions } from 'features/system/store/systemSlice';
+import { memo, useMemo, useRef } from 'react';
+
+interface FocusRegionWrapperProps extends BoxProps {
+  region: FocusRegionName;
+  focusOnMount?: boolean;
+}
+
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  position: 'relative',
+  '&[data-highlighted="true"]::after': {
+    borderColor: 'blue.700',
+  },
+  '&::after': {
+    content: '""',
+    position: 'absolute',
+    inset: 0,
+    zIndex: 1,
+    borderRadius: 'base',
+    border: '2px solid',
+    borderColor: 'transparent',
+    pointerEvents: 'none',
+    transition: 'border-color 0.1s ease-in-out',
+  },
+};
+
+export const FocusRegionWrapper = memo(
+  ({ region, focusOnMount = false, sx, children, ...boxProps }: FocusRegionWrapperProps) => {
+    const shouldHighlightFocusedRegions = useAppSelector(selectSystemShouldEnableHighlightFocusedRegions);
+
+    const ref = useRef<HTMLDivElement>(null);
+
+    const options = useMemo(() => ({ focusOnMount }), [focusOnMount]);
+
+    useFocusRegion(region, ref, options);
+    const isFocused = useIsRegionFocused(region);
+    const isHighlighted = isFocused && shouldHighlightFocusedRegions;
+
+    return (
+      <Box
+        ref={ref}
+        tabIndex={-1}
+        sx={useMemo(() => ({ ...FOCUS_REGION_STYLES, ...sx }), [sx])}
+        data-highlighted={isHighlighted}
+        {...boxProps}
+      >
+        {children}
+      </Box>
+    );
+  }
+);
+
+FocusRegionWrapper.displayName = 'FocusRegionWrapper';

invokeai/frontend/web/src/common/hooks/focus.ts

@@ -30,7 +30,7 @@ const log = logger('system');
 /**
  * The names of the focus regions.
  */
-type FocusRegionName = 'gallery' | 'layers' | 'canvas' | 'workflows' | 'viewer';
+export type FocusRegionName = 'gallery' | 'layers' | 'canvas' | 'workflows' | 'viewer';
 
 /**
  * A map of focus regions to the elements that are part of that region.

invokeai/frontend/web/src/features/controlLayers/components/CanvasLayersPanelContent.tsx

@@ -1,28 +1,33 @@
-import { Divider, Flex } from '@invoke-ai/ui-library';
+import { Divider, Flex, type SystemStyleObject } from '@invoke-ai/ui-library';
 import { useAppSelector } from 'app/store/storeHooks';
-import { useFocusRegion } from 'common/hooks/focus';
+import { FocusRegionWrapper } from 'common/components/FocusRegionWrapper';
 import { CanvasAddEntityButtons } from 'features/controlLayers/components/CanvasAddEntityButtons';
 import { CanvasEntityList } from 'features/controlLayers/components/CanvasEntityList/CanvasEntityList';
 import { EntityListSelectedEntityActionBar } from 'features/controlLayers/components/CanvasEntityList/EntityListSelectedEntityActionBar';
 import { selectHasEntities } from 'features/controlLayers/store/selectors';
-import { memo, useRef } from 'react';
+import { memo } from 'react';
 
 import { ParamDenoisingStrength } from './ParamDenoisingStrength';
 
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  width: 'full',
+  height: 'full',
+};
+
 export const CanvasLayersPanelContent = memo(() => {
   const hasEntities = useAppSelector(selectHasEntities);
-  const layersPanelFocusRef = useRef<HTMLDivElement>(null);
-  useFocusRegion('layers', layersPanelFocusRef);
 
   return (
-    <Flex ref={layersPanelFocusRef} flexDir="column" gap={2} w="full" h="full">
-      <EntityListSelectedEntityActionBar />
-      <Divider py={0} />
-      <ParamDenoisingStrength />
-      <Divider py={0} />
-      {!hasEntities && <CanvasAddEntityButtons />}
-      {hasEntities && <CanvasEntityList />}
-    </Flex>
+    <FocusRegionWrapper region="layers" sx={FOCUS_REGION_STYLES}>
+      <Flex flexDir="column" gap={2} w="full" h="full">
+        <EntityListSelectedEntityActionBar />
+        <Divider py={0} />
+        <ParamDenoisingStrength />
+        <Divider py={0} />
+        {!hasEntities && <CanvasAddEntityButtons />}
+        {hasEntities && <CanvasEntityList />}
+      </Flex>
+    </FocusRegionWrapper>
   );
 });
 

invokeai/frontend/web/src/features/controlLayers/components/CanvasMainPanelContent.tsx

@@ -1,6 +1,14 @@
-import { ContextMenu, Flex, IconButton, Menu, MenuButton, MenuList } from '@invoke-ai/ui-library';
+import {
+  ContextMenu,
+  Flex,
+  IconButton,
+  Menu,
+  MenuButton,
+  MenuList,
+  type SystemStyleObject,
+} from '@invoke-ai/ui-library';
 import { useAppSelector } from 'app/store/storeHooks';
-import { useFocusRegion } from 'common/hooks/focus';
+import { FocusRegionWrapper } from 'common/components/FocusRegionWrapper';
 import { CanvasAlertsPreserveMask } from 'features/controlLayers/components/CanvasAlerts/CanvasAlertsPreserveMask';
 import { CanvasAlertsSelectedEntityStatus } from 'features/controlLayers/components/CanvasAlerts/CanvasAlertsSelectedEntityStatus';
 import { CanvasAlertsSendingToGallery } from 'features/controlLayers/components/CanvasAlerts/CanvasAlertsSendingTo';
@@ -18,11 +26,16 @@ import { Transform } from 'features/controlLayers/components/Transform/Transform
 import { CanvasManagerProviderGate } from 'features/controlLayers/contexts/CanvasManagerProviderGate';
 import { selectDynamicGrid, selectShowHUD } from 'features/controlLayers/store/canvasSettingsSlice';
 import { GatedImageViewer } from 'features/gallery/components/ImageViewer/ImageViewer';
-import { memo, useCallback, useRef } from 'react';
+import { memo, useCallback } from 'react';
 import { PiDotsThreeOutlineVerticalFill } from 'react-icons/pi';
 
 import { CanvasAlertsInvocationProgress } from './CanvasAlerts/CanvasAlertsInvocationProgress';
 
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  width: 'full',
+  height: 'full',
+};
+
 const MenuContent = () => {
   return (
     <CanvasManagerProviderGate>
@@ -35,7 +48,6 @@ const MenuContent = () => {
 };
 
 export const CanvasMainPanelContent = memo(() => {
-  const ref = useRef<HTMLDivElement>(null);
   const dynamicGrid = useAppSelector(selectDynamicGrid);
   const showHUD = useAppSelector(selectShowHUD);
 
@@ -43,82 +55,81 @@ export const CanvasMainPanelContent = memo(() => {
     return <MenuContent />;
   }, []);
 
-  useFocusRegion('canvas', ref);
-
   return (
-    <Flex
-      tabIndex={-1}
-      ref={ref}
-      borderRadius="base"
-      position="relative"
-      flexDirection="column"
-      height="full"
-      width="full"
-      gap={2}
-      alignItems="center"
-      justifyContent="center"
-      overflow="hidden"
-    >
-      <CanvasManagerProviderGate>
-        <CanvasToolbar />
-      </CanvasManagerProviderGate>
-      <ContextMenu<HTMLDivElement> renderMenu={renderMenu} withLongPress={false}>
-        {(ref) => (
-          <Flex
-            ref={ref}
-            position="relative"
-            w="full"
-            h="full"
-            bg={dynamicGrid ? 'base.850' : 'base.900'}
-            borderRadius="base"
-            overflow="hidden"
-          >
-            <InvokeCanvasComponent />
-            <CanvasManagerProviderGate>
-              <Flex
-                position="absolute"
-                flexDir="column"
-                top={1}
-                insetInlineStart={1}
-                pointerEvents="none"
-                gap={2}
-                alignItems="flex-start"
-              >
-                {showHUD && <CanvasHUD />}
-                <CanvasAlertsSelectedEntityStatus />
-                <CanvasAlertsPreserveMask />
-                <CanvasAlertsSendingToGallery />
-                <CanvasAlertsInvocationProgress />
-              </Flex>
-              <Flex position="absolute" top={1} insetInlineEnd={1}>
-                <Menu>
-                  <MenuButton as={IconButton} icon={<PiDotsThreeOutlineVerticalFill />} colorScheme="base" />
-                  <MenuContent />
-                </Menu>
-              </Flex>
-            </CanvasManagerProviderGate>
-          </Flex>
-        )}
-      </ContextMenu>
-      <Flex position="absolute" bottom={4} gap={2} align="center" justify="center">
+    <FocusRegionWrapper region="canvas" sx={FOCUS_REGION_STYLES}>
+      <Flex
+        tabIndex={-1}
+        borderRadius="base"
+        position="relative"
+        flexDirection="column"
+        height="full"
+        width="full"
+        gap={2}
+        alignItems="center"
+        justifyContent="center"
+        overflow="hidden"
+      >
         <CanvasManagerProviderGate>
-          <StagingAreaIsStagingGate>
-            <StagingAreaToolbar />
-          </StagingAreaIsStagingGate>
+          <CanvasToolbar />
         </CanvasManagerProviderGate>
-      </Flex>
-      <Flex position="absolute" bottom={4}>
+        <ContextMenu<HTMLDivElement> renderMenu={renderMenu} withLongPress={false}>
+          {(ref) => (
+            <Flex
+              ref={ref}
+              position="relative"
+              w="full"
+              h="full"
+              bg={dynamicGrid ? 'base.850' : 'base.900'}
+              borderRadius="base"
+              overflow="hidden"
+            >
+              <InvokeCanvasComponent />
+              <CanvasManagerProviderGate>
+                <Flex
+                  position="absolute"
+                  flexDir="column"
+                  top={1}
+                  insetInlineStart={1}
+                  pointerEvents="none"
+                  gap={2}
+                  alignItems="flex-start"
+                >
+                  {showHUD && <CanvasHUD />}
+                  <CanvasAlertsSelectedEntityStatus />
+                  <CanvasAlertsPreserveMask />
+                  <CanvasAlertsSendingToGallery />
+                  <CanvasAlertsInvocationProgress />
+                </Flex>
+                <Flex position="absolute" top={1} insetInlineEnd={1}>
+                  <Menu>
+                    <MenuButton as={IconButton} icon={<PiDotsThreeOutlineVerticalFill />} colorScheme="base" />
+                    <MenuContent />
+                  </Menu>
+                </Flex>
+              </CanvasManagerProviderGate>
+            </Flex>
+          )}
+        </ContextMenu>
+        <Flex position="absolute" bottom={4} gap={2} align="center" justify="center">
+          <CanvasManagerProviderGate>
+            <StagingAreaIsStagingGate>
+              <StagingAreaToolbar />
+            </StagingAreaIsStagingGate>
+          </CanvasManagerProviderGate>
+        </Flex>
+        <Flex position="absolute" bottom={4}>
+          <CanvasManagerProviderGate>
+            <Filter />
+            <Transform />
+            <SelectObject />
+          </CanvasManagerProviderGate>
+        </Flex>
         <CanvasManagerProviderGate>
-          <Filter />
-          <Transform />
-          <SelectObject />
+          <CanvasDropArea />
         </CanvasManagerProviderGate>
+        <GatedImageViewer />
       </Flex>
-      <CanvasManagerProviderGate>
-        <CanvasDropArea />
-      </CanvasManagerProviderGate>
-      <GatedImageViewer />
-    </Flex>
+    </FocusRegionWrapper>
   );
 });
 

invokeai/frontend/web/src/features/dnd/FullscreenDropzone.tsx

@@ -7,6 +7,7 @@ import { Box, Flex, Heading } from '@invoke-ai/ui-library';
 import { useStore } from '@nanostores/react';
 import { getStore } from 'app/store/nanostores/store';
 import { useAppSelector } from 'app/store/storeHooks';
+import { $focusedRegion } from 'common/hooks/focus';
 import { setFileToPaste } from 'features/controlLayers/components/CanvasPasteModal';
 import { DndDropOverlay } from 'features/dnd/DndDropOverlay';
 import type { DndTargetState } from 'features/dnd/types';
@@ -99,10 +100,18 @@ export const FullscreenDropzone = memo(() => {
         return;
       }
 
+      const focusedRegion = $focusedRegion.get();
+
       // While on the canvas tab and when pasting a single image, canvas may want to create a new layer. Let it handle
       // the paste event.
       const [firstImageFile] = files;
-      if (!isImageViewerOpen && activeTab === 'canvas' && files.length === 1 && firstImageFile) {
+      if (
+        focusedRegion === 'canvas' &&
+        !isImageViewerOpen &&
+        activeTab === 'canvas' &&
+        files.length === 1 &&
+        firstImageFile
+      ) {
         setFileToPaste(firstImageFile);
         return;
       }

invokeai/frontend/web/src/features/gallery/components/GalleryPanelContent.tsx

@@ -1,6 +1,15 @@
-import { Box, Button, Collapse, Divider, Flex, IconButton, useDisclosure } from '@invoke-ai/ui-library';
+import {
+  Box,
+  Button,
+  Collapse,
+  Divider,
+  Flex,
+  IconButton,
+  type SystemStyleObject,
+  useDisclosure,
+} from '@invoke-ai/ui-library';
 import { useAppDispatch, useAppSelector } from 'app/store/storeHooks';
-import { useFocusRegion } from 'common/hooks/focus';
+import { FocusRegionWrapper } from 'common/components/FocusRegionWrapper';
 import { GalleryHeader } from 'features/gallery/components/GalleryHeader';
 import { selectBoardSearchText } from 'features/gallery/store/gallerySelectors';
 import { boardSearchTextChanged } from 'features/gallery/store/gallerySlice';
@@ -20,14 +29,20 @@ import { Gallery } from './Gallery';
 
 const COLLAPSE_STYLES: CSSProperties = { flexShrink: 0, minHeight: 0 };
 
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  width: 'full',
+  height: 'full',
+  position: 'relative',
+  flexDirection: 'column',
+  display: 'flex',
+};
+
 const GalleryPanelContent = () => {
   const { t } = useTranslation();
   const boardSearchText = useAppSelector(selectBoardSearchText);
   const dispatch = useAppDispatch();
   const boardSearchDisclosure = useDisclosure({ defaultIsOpen: !!boardSearchText.length });
   const imperativePanelGroupRef = useRef<ImperativePanelGroupHandle>(null);
-  const galleryPanelFocusRef = useRef<HTMLDivElement>(null);
-  useFocusRegion('gallery', galleryPanelFocusRef);
 
   const boardsListPanelOptions = useMemo<UsePanelOptions>(
     () => ({
@@ -50,7 +65,7 @@ const GalleryPanelContent = () => {
   }, [boardSearchText.length, boardSearchDisclosure, boardsListPanel, dispatch]);
 
   return (
-    <Flex ref={galleryPanelFocusRef} position="relative" flexDirection="column" h="full" w="full" tabIndex={-1}>
+    <FocusRegionWrapper region="gallery" sx={FOCUS_REGION_STYLES}>
       <Flex alignItems="center" justifyContent="space-between" w="full">
         <Flex flexGrow={1} flexBasis={0}>
           <Button
@@ -99,7 +114,7 @@ const GalleryPanelContent = () => {
           <Gallery />
         </Panel>
       </PanelGroup>
-    </Flex>
+    </FocusRegionWrapper>
   );
 };
 

invokeai/frontend/web/src/features/gallery/components/ImageViewer/ImageViewer.tsx

@@ -1,6 +1,6 @@
-import { Box, Flex, IconButton } from '@invoke-ai/ui-library';
+import { Box, IconButton, type SystemStyleObject } from '@invoke-ai/ui-library';
 import { useAppSelector } from 'app/store/storeHooks';
-import { useFocusRegion } from 'common/hooks/focus';
+import { FocusRegionWrapper } from 'common/components/FocusRegionWrapper';
 import { useAssertSingleton } from 'common/hooks/useAssertSingleton';
 import { CompareToolbar } from 'features/gallery/components/ImageViewer/CompareToolbar';
 import CurrentImagePreview from 'features/gallery/components/ImageViewer/CurrentImagePreview';
@@ -9,7 +9,7 @@ import { ImageComparisonDroppable } from 'features/gallery/components/ImageViewe
 import { ViewerToolbar } from 'features/gallery/components/ImageViewer/ViewerToolbar';
 import { selectHasImageToCompare } from 'features/gallery/store/gallerySelectors';
 import type { ReactNode } from 'react';
-import { memo, useRef } from 'react';
+import { memo } from 'react';
 import { useHotkeys } from 'react-hotkeys-hook';
 import { useTranslation } from 'react-i18next';
 import { PiXBold } from 'react-icons/pi';
@@ -25,29 +25,24 @@ const useFocusRegionOptions = {
   focusOnMount: true,
 };
 
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  width: 'full',
+  height: 'full',
+  position: 'absolute',
+  flexDirection: 'column',
+  inset: 0,
+  alignItems: 'center',
+  justifyContent: 'center',
+  overflow: 'hidden',
+};
+
 export const ImageViewer = memo(({ closeButton }: Props) => {
   useAssertSingleton('ImageViewer');
   const hasImageToCompare = useAppSelector(selectHasImageToCompare);
   const [containerRef, containerDims] = useMeasure<HTMLDivElement>();
-  const ref = useRef<HTMLDivElement>(null);
-  useFocusRegion('viewer', ref, useFocusRegionOptions);
 
   return (
-    <Flex
-      ref={ref}
-      tabIndex={-1}
-      layerStyle="first"
-      borderRadius="base"
-      position="absolute"
-      flexDirection="column"
-      top={0}
-      right={0}
-      bottom={0}
-      left={0}
-      alignItems="center"
-      justifyContent="center"
-      overflow="hidden"
-    >
+    <FocusRegionWrapper region="viewer" sx={FOCUS_REGION_STYLES} layerStyle="first" {...useFocusRegionOptions}>
       {hasImageToCompare && <CompareToolbar />}
       {!hasImageToCompare && <ViewerToolbar closeButton={closeButton} />}
       <Box ref={containerRef} w="full" h="full" p={2}>
@@ -55,7 +50,7 @@ export const ImageViewer = memo(({ closeButton }: Props) => {
         {hasImageToCompare && <ImageComparison containerDims={containerDims} />}
       </Box>
       <ImageComparisonDroppable />
-    </Flex>
+    </FocusRegionWrapper>
   );
 });
 

invokeai/frontend/web/src/features/gallery/components/ImageViewer/NoContentForViewer.tsx

@@ -7,6 +7,7 @@ import { LOADING_SYMBOL, useHasImages } from 'features/gallery/hooks/useHasImage
 import { $installModelsTab } from 'features/modelManagerV2/subpanels/InstallModels';
 import { useFeatureStatus } from 'features/system/hooks/useFeatureStatus';
 import { selectIsLocal } from 'features/system/store/configSlice';
+import { selectActiveTab } from 'features/ui/store/uiSelectors';
 import { setActiveTab } from 'features/ui/store/uiSlice';
 import type { PropsWithChildren } from 'react';
 import { memo, useCallback, useMemo } from 'react';
@@ -19,6 +20,7 @@ export const NoContentForViewer = memo(() => {
   const [mainModels, { data }] = useMainModels();
   const isLocal = useAppSelector(selectIsLocal);
   const isEnabled = useFeatureStatus('starterModels');
+  const activeTab = useAppSelector(selectActiveTab);
   const { t } = useTranslation();
 
   const showStarterBundles = useMemo(() => {
@@ -38,10 +40,10 @@ export const NoContentForViewer = memo(() => {
   }
 
   return (
-    <Flex flexDir="column" gap={8} alignItems="center" textAlign="center" maxW="600px">
+    <Flex flexDir="column" gap={8} alignItems="center" textAlign="center" maxW="400px">
       <InvokeLogoIcon w={32} h={32} />
       <Flex flexDir="column" gap={4} alignItems="center" textAlign="center">
-        {isLocal ? <GetStartedLocal /> : <GetStartedCommercial />}
+        {isLocal ? <GetStartedLocal /> : activeTab === 'workflows' ? <GetStartedWorkflows /> : <GetStartedCommercial />}
         {showStarterBundles && <StarterBundlesCallout />}
         <Divider />
         <GettingStartedVideosCallout />
@@ -103,6 +105,14 @@ const GetStartedCommercial = () => {
   );
 };
 
+const GetStartedWorkflows = () => {
+  return (
+    <Text fontSize="md" color="base.200">
+      <Trans i18nKey="newUserExperience.toGetStartedWorkflow" components={{ StrongComponent }} />
+    </Text>
+  );
+};
+
 const GettingStartedVideosCallout = () => {
   return (
     <Text fontSize="md" color="base.200">

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelManagerPanel/ModelList.tsx

@@ -16,6 +16,7 @@ import {
   useEmbeddingModels,
   useFluxReduxModels,
   useIPAdapterModels,
+  useLLaVAModels,
   useLoRAModels,
   useMainModels,
   useRefinerModels,
@@ -126,6 +127,12 @@ const ModelList = () => {
     [fluxReduxModels, searchTerm, filteredModelType]
   );
 
+  const [llavaOneVisionModels, { isLoading: isLoadingLlavaOneVisionModels }] = useLLaVAModels();
+  const filteredLlavaOneVisionModels = useMemo(
+    () => modelsFilter(llavaOneVisionModels, searchTerm, filteredModelType),
+    [llavaOneVisionModels, searchTerm, filteredModelType]
+  );
+
   const totalFilteredModels = useMemo(() => {
     return (
       filteredMainModels.length +
@@ -236,6 +243,17 @@ const ModelList = () => {
         {!isLoadingClipEmbedModels && filteredClipEmbedModels.length > 0 && (
           <ModelListWrapper title={t('modelManager.clipEmbed')} modelList={filteredClipEmbedModels} key="clip-embed" />
         )}
+
+        {/* LLaVA OneVision List */}
+        {isLoadingLlavaOneVisionModels && <FetchingModelsLoader loadingMessage="Loading LLaVA OneVision Models..." />}
+        {!isLoadingLlavaOneVisionModels && filteredLlavaOneVisionModels.length > 0 && (
+          <ModelListWrapper
+            title={t('modelManager.llavaOnevision')}
+            modelList={filteredLlavaOneVisionModels}
+            key="llava-onevision"
+          />
+        )}
+
         {/* Spandrel Image to Image List */}
         {isLoadingSpandrelImageToImageModels && (
           <FetchingModelsLoader loadingMessage="Loading Image-to-Image Models..." />

invokeai/frontend/web/src/features/modelManagerV2/subpanels/ModelManagerPanel/ModelTypeFilter.tsx

@@ -27,6 +27,7 @@ export const ModelTypeFilter = memo(() => {
       control_lora: t('modelManager.controlLora'),
       siglip: t('modelManager.siglip'),
       flux_redux: t('modelManager.fluxRedux'),
+      llava_onevision: t('modelManager.llavaOnevision'),
     }),
     [t]
   );

invokeai/frontend/web/src/features/nodes/components/NodeEditor.tsx

@@ -1,10 +1,10 @@
-import { Flex } from '@invoke-ai/ui-library';
+import type { SystemStyleObject } from '@invoke-ai/ui-library';
+import { FocusRegionWrapper } from 'common/components/FocusRegionWrapper';
 import { IAINoContentFallback } from 'common/components/IAIImageFallback';
-import { useFocusRegion } from 'common/hooks/focus';
 import { AddNodeCmdk } from 'features/nodes/components/flow/AddNodeCmdk/AddNodeCmdk';
 import TopPanel from 'features/nodes/components/flow/panels/TopPanel/TopPanel';
 import WorkflowEditorSettings from 'features/nodes/components/flow/panels/TopRightPanel/WorkflowEditorSettings';
-import { memo, useRef } from 'react';
+import { memo } from 'react';
 import { useTranslation } from 'react-i18next';
 import { PiFlowArrowBold } from 'react-icons/pi';
 import { useGetOpenAPISchemaQuery } from 'services/api/endpoints/appInfo';
@@ -13,24 +13,20 @@ import { Flow } from './flow/Flow';
 import BottomLeftPanel from './flow/panels/BottomLeftPanel/BottomLeftPanel';
 import MinimapPanel from './flow/panels/MinimapPanel/MinimapPanel';
 
+const FOCUS_REGION_STYLES: SystemStyleObject = {
+  position: 'relative',
+  width: 'full',
+  height: 'full',
+  alignItems: 'center',
+  justifyContent: 'center',
+};
+
 const NodeEditor = () => {
   const { data, isLoading } = useGetOpenAPISchemaQuery();
   const { t } = useTranslation();
-  const ref = useRef<HTMLDivElement>(null);
-  useFocusRegion('workflows', ref);
 
   return (
-    <Flex
-      tabIndex={-1}
-      ref={ref}
-      layerStyle="first"
-      position="relative"
-      width="full"
-      height="full"
-      borderRadius="base"
-      alignItems="center"
-      justifyContent="center"
-    >
+    <FocusRegionWrapper region="workflows" layerStyle="first" sx={FOCUS_REGION_STYLES}>
       {data && (
         <>
           <Flow />
@@ -42,7 +38,7 @@ const NodeEditor = () => {
       )}
       <WorkflowEditorSettings />
       {isLoading && <IAINoContentFallback label={t('nodes.loadingNodes')} icon={PiFlowArrowBold} />}
-    </Flex>
+    </FocusRegionWrapper>
   );
 };
 

invokeai/frontend/web/src/features/nodes/components/flow/Flow.tsx

@@ -252,7 +252,7 @@ export const Flow = memo(() => {
     id: 'pasteSelection',
     category: 'workflows',
     callback: pasteSelection,
-    options: { preventDefault: true },
+    options: { enabled: isWorkflowsFocused, preventDefault: true },
     dependencies: [pasteSelection],
   });
 
@@ -260,7 +260,7 @@ export const Flow = memo(() => {
     id: 'pasteSelectionWithEdges',
     category: 'workflows',
     callback: pasteSelectionWithEdges,
-    options: { preventDefault: true },
+    options: { enabled: isWorkflowsFocused, preventDefault: true },
     dependencies: [pasteSelectionWithEdges],
   });
 
@@ -270,7 +270,7 @@ export const Flow = memo(() => {
     callback: () => {
       dispatch(undo());
     },
-    options: { enabled: mayUndo, preventDefault: true },
+    options: { enabled: isWorkflowsFocused && mayUndo, preventDefault: true },
     dependencies: [mayUndo],
   });
 
@@ -280,7 +280,7 @@ export const Flow = memo(() => {
     callback: () => {
       dispatch(redo());
     },
-    options: { enabled: mayRedo, preventDefault: true },
+    options: { enabled: isWorkflowsFocused && mayRedo, preventDefault: true },
     dependencies: [mayRedo],
   });
 

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/FloatField/FloatFieldInput.tsx

@@ -3,24 +3,35 @@ import { useFloatField } from 'features/nodes/components/flow/nodes/Invocation/f
 import type { FieldComponentProps } from 'features/nodes/components/flow/nodes/Invocation/fields/inputs/types';
 import { NO_DRAG_CLASS } from 'features/nodes/types/constants';
 import type { FloatFieldInputInstance, FloatFieldInputTemplate } from 'features/nodes/types/field';
+import type { NodeFieldFloatSettings } from 'features/nodes/types/workflow';
 import { memo } from 'react';
 
-export const FloatFieldInput = memo((props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate>) => {
-  const { defaultValue, onChange, value, min, max, step, fineStep } = useFloatField(props);
+export const FloatFieldInput = memo(
+  (
+    props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate, { settings?: NodeFieldFloatSettings }>
+  ) => {
+    const { nodeId, field, fieldTemplate, settings } = props;
+    const { defaultValue, onChange, min, max, step, fineStep } = useFloatField(
+      nodeId,
+      field.name,
+      fieldTemplate,
+      settings
+    );
 
-  return (
-    <CompositeNumberInput
-      defaultValue={defaultValue}
-      onChange={onChange}
-      value={value}
-      min={min}
-      max={max}
-      step={step}
-      fineStep={fineStep}
-      className={NO_DRAG_CLASS}
-      flex="1 1 0"
-    />
-  );
-});
+    return (
+      <CompositeNumberInput
+        defaultValue={defaultValue}
+        onChange={onChange}
+        value={field.value}
+        min={min}
+        max={max}
+        step={step}
+        fineStep={fineStep}
+        className={NO_DRAG_CLASS}
+        flex="1 1 0"
+      />
+    );
+  }
+);
 
 FloatFieldInput.displayName = 'FloatFieldInput ';

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/FloatField/FloatFieldInputAndSlider.tsx

@@ -3,18 +3,27 @@ import { useFloatField } from 'features/nodes/components/flow/nodes/Invocation/f
 import type { FieldComponentProps } from 'features/nodes/components/flow/nodes/Invocation/fields/inputs/types';
 import { NO_DRAG_CLASS } from 'features/nodes/types/constants';
 import type { FloatFieldInputInstance, FloatFieldInputTemplate } from 'features/nodes/types/field';
+import type { NodeFieldFloatSettings } from 'features/nodes/types/workflow';
 import { memo } from 'react';
 
 export const FloatFieldInputAndSlider = memo(
-  (props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate>) => {
-    const { defaultValue, onChange, value, min, max, step, fineStep } = useFloatField(props);
+  (
+    props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate, { settings?: NodeFieldFloatSettings }>
+  ) => {
+    const { nodeId, field, fieldTemplate, settings } = props;
+    const { defaultValue, onChange, min, max, step, fineStep } = useFloatField(
+      nodeId,
+      field.name,
+      fieldTemplate,
+      settings
+    );
 
     return (
       <>
         <CompositeSlider
           defaultValue={defaultValue}
           onChange={onChange}
-          value={value}
+          value={field.value}
           min={min}
           max={max}
           step={step}
@@ -27,7 +36,7 @@ export const FloatFieldInputAndSlider = memo(
         <CompositeNumberInput
           defaultValue={defaultValue}
           onChange={onChange}
-          value={value}
+          value={field.value}
           min={min}
           max={max}
           step={step}

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/FloatField/FloatFieldSlider.tsx

@@ -3,26 +3,37 @@ import { useFloatField } from 'features/nodes/components/flow/nodes/Invocation/f
 import type { FieldComponentProps } from 'features/nodes/components/flow/nodes/Invocation/fields/inputs/types';
 import { NO_DRAG_CLASS } from 'features/nodes/types/constants';
 import type { FloatFieldInputInstance, FloatFieldInputTemplate } from 'features/nodes/types/field';
+import type { NodeFieldFloatSettings } from 'features/nodes/types/workflow';
 import { memo } from 'react';
 
-export const FloatFieldSlider = memo((props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate>) => {
-  const { defaultValue, onChange, value, min, max, step, fineStep } = useFloatField(props);
+export const FloatFieldSlider = memo(
+  (
+    props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate, { settings?: NodeFieldFloatSettings }>
+  ) => {
+    const { nodeId, field, fieldTemplate, settings } = props;
+    const { defaultValue, onChange, min, max, step, fineStep } = useFloatField(
+      nodeId,
+      field.name,
+      fieldTemplate,
+      settings
+    );
 
-  return (
-    <CompositeSlider
-      defaultValue={defaultValue}
-      onChange={onChange}
-      value={value}
-      min={min}
-      max={max}
-      step={step}
-      fineStep={fineStep}
-      className={NO_DRAG_CLASS}
-      marks
-      withThumbTooltip
-      flex="1 1 0"
-    />
-  );
-});
+    return (
+      <CompositeSlider
+        defaultValue={defaultValue}
+        onChange={onChange}
+        value={field.value}
+        min={min}
+        max={max}
+        step={step}
+        fineStep={fineStep}
+        className={NO_DRAG_CLASS}
+        marks
+        withThumbTooltip
+        flex="1 1 0"
+      />
+    );
+  }
+);
 
 FloatFieldSlider.displayName = 'FloatFieldSlider ';

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/FloatField/useFloatField.ts

@@ -1,65 +1,89 @@
 import { NUMPY_RAND_MAX } from 'app/constants';
 import { useAppDispatch } from 'app/store/storeHooks';
-import type { FieldComponentProps } from 'features/nodes/components/flow/nodes/Invocation/fields/inputs/types';
+import { roundDownToMultiple, roundUpToMultiple } from 'common/util/roundDownToMultiple';
 import { fieldFloatValueChanged } from 'features/nodes/store/nodesSlice';
-import type { FloatFieldInputInstance, FloatFieldInputTemplate } from 'features/nodes/types/field';
+import type { FloatFieldInputTemplate } from 'features/nodes/types/field';
+import { constrainNumber } from 'features/nodes/util/constrainNumber';
 import { isNil } from 'lodash-es';
 import { useCallback, useMemo } from 'react';
 
-export const useFloatField = (props: FieldComponentProps<FloatFieldInputInstance, FloatFieldInputTemplate>) => {
-  const { nodeId, field, fieldTemplate } = props;
+export const useFloatField = (
+  nodeId: string,
+  fieldName: string,
+  fieldTemplate: FloatFieldInputTemplate,
+  overrides: { min?: number; max?: number; step?: number } = {}
+) => {
+  const { min: overrideMin, max: overrideMax, step: overrideStep } = overrides;
   const dispatch = useAppDispatch();
 
-  const onChange = useCallback(
-    (value: number) => {
-      dispatch(fieldFloatValueChanged({ nodeId, fieldName: field.name, value }));
-    },
-    [dispatch, field.name, nodeId]
-  );
-
-  const min = useMemo(() => {
-    let min = -NUMPY_RAND_MAX;
-    if (!isNil(fieldTemplate.minimum)) {
-      min = fieldTemplate.minimum;
-    }
-    if (!isNil(fieldTemplate.exclusiveMinimum)) {
-      min = fieldTemplate.exclusiveMinimum + 0.01;
-    }
-    return min;
-  }, [fieldTemplate.exclusiveMinimum, fieldTemplate.minimum]);
-
-  const max = useMemo(() => {
-    let max = NUMPY_RAND_MAX;
-    if (!isNil(fieldTemplate.maximum)) {
-      max = fieldTemplate.maximum;
-    }
-    if (!isNil(fieldTemplate.exclusiveMaximum)) {
-      max = fieldTemplate.exclusiveMaximum - 0.01;
-    }
-    return max;
-  }, [fieldTemplate.exclusiveMaximum, fieldTemplate.maximum]);
-
   const step = useMemo(() => {
+    if (overrideStep !== undefined) {
+      return overrideStep;
+    }
     if (isNil(fieldTemplate.multipleOf)) {
       return 0.1;
     }
     return fieldTemplate.multipleOf;
-  }, [fieldTemplate.multipleOf]);
+  }, [fieldTemplate.multipleOf, overrideStep]);
 
   const fineStep = useMemo(() => {
+    if (overrideStep !== undefined) {
+      return overrideStep;
+    }
     if (isNil(fieldTemplate.multipleOf)) {
       return 0.01;
     }
     return fieldTemplate.multipleOf;
-  }, [fieldTemplate.multipleOf]);
+  }, [fieldTemplate.multipleOf, overrideStep]);
+
+  const min = useMemo(() => {
+    let min = -NUMPY_RAND_MAX;
+
+    if (overrideMin !== undefined) {
+      min = overrideMin;
+    } else if (!isNil(fieldTemplate.minimum)) {
+      min = fieldTemplate.minimum;
+    } else if (!isNil(fieldTemplate.exclusiveMinimum)) {
+      min = fieldTemplate.exclusiveMinimum + 0.01;
+    }
+
+    return roundUpToMultiple(min, step);
+  }, [fieldTemplate.exclusiveMinimum, fieldTemplate.minimum, overrideMin, step]);
+
+  const max = useMemo(() => {
+    let max = NUMPY_RAND_MAX;
+
+    if (overrideMax !== undefined) {
+      max = overrideMax;
+    } else if (!isNil(fieldTemplate.maximum)) {
+      max = fieldTemplate.maximum;
+    } else if (!isNil(fieldTemplate.exclusiveMaximum)) {
+      max = fieldTemplate.exclusiveMaximum - 0.01;
+    }
+
+    return roundDownToMultiple(max, step);
+  }, [fieldTemplate.exclusiveMaximum, fieldTemplate.maximum, overrideMax, step]);
+
+  const constrainValue = useCallback(
+    (v: number) =>
+      constrainNumber(v, { min, max, step: step }, { min: overrideMin, max: overrideMax, step: overrideStep }),
+    [max, min, overrideMax, overrideMin, overrideStep, step]
+  );
+
+  const onChange = useCallback(
+    (value: number) => {
+      dispatch(fieldFloatValueChanged({ nodeId, fieldName, value }));
+    },
+    [dispatch, fieldName, nodeId]
+  );
 
   return {
     defaultValue: fieldTemplate.default,
     onChange,
-    value: field.value,
     min,
     max,
     step,
     fineStep,
+    constrainValue,
   };
 };

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/InputFieldRenderer.tsx

@@ -62,6 +62,8 @@ import {
   isIntegerGeneratorFieldInputTemplate,
   isIPAdapterModelFieldInputInstance,
   isIPAdapterModelFieldInputTemplate,
+  isLLaVAModelFieldInputInstance,
+  isLLaVAModelFieldInputTemplate,
   isLoRAModelFieldInputInstance,
   isLoRAModelFieldInputTemplate,
   isMainModelFieldInputInstance,
@@ -95,6 +97,8 @@ import {
 } from 'features/nodes/types/field';
 import type { NodeFieldElement } from 'features/nodes/types/workflow';
 import { memo } from 'react';
+import type { Equals } from 'tsafe';
+import { assert } from 'tsafe';
 
 import BoardFieldInputComponent from './inputs/BoardFieldInputComponent';
 import BooleanFieldInputComponent from './inputs/BooleanFieldInputComponent';
@@ -110,6 +114,7 @@ import FluxReduxModelFieldInputComponent from './inputs/FluxReduxModelFieldInput
 import FluxVAEModelFieldInputComponent from './inputs/FluxVAEModelFieldInputComponent';
 import ImageFieldInputComponent from './inputs/ImageFieldInputComponent';
 import IPAdapterModelFieldInputComponent from './inputs/IPAdapterModelFieldInputComponent';
+import LLaVAModelFieldInputComponent from './inputs/LLaVAModelFieldInputComponent';
 import LoRAModelFieldInputComponent from './inputs/LoRAModelFieldInputComponent';
 import MainModelFieldInputComponent from './inputs/MainModelFieldInputComponent';
 import RefinerModelFieldInputComponent from './inputs/RefinerModelFieldInputComponent';
@@ -157,6 +162,8 @@ export const InputFieldRenderer = memo(({ nodeId, fieldName, settings }: Props)
       return <StringFieldInput nodeId={nodeId} field={field} fieldTemplate={template} />;
     } else if (settings.component === 'textarea') {
       return <StringFieldTextarea nodeId={nodeId} field={field} fieldTemplate={template} />;
+    } else {
+      assert<Equals<never, typeof settings.component>>(false, 'Unexpected settings.component');
     }
   }
 
@@ -171,32 +178,47 @@ export const InputFieldRenderer = memo(({ nodeId, fieldName, settings }: Props)
     if (!isIntegerFieldInputInstance(field)) {
       return null;
     }
-    if (settings?.type !== 'integer-field-config') {
+    if (!settings || settings.type !== 'integer-field-config') {
       return <IntegerFieldInput nodeId={nodeId} field={field} fieldTemplate={template} />;
     }
+
     if (settings.component === 'number-input') {
-      return <IntegerFieldInput nodeId={nodeId} field={field} fieldTemplate={template} />;
-    } else if (settings.component === 'slider') {
-      return <IntegerFieldSlider nodeId={nodeId} field={field} fieldTemplate={template} />;
-    } else if (settings.component === 'number-input-and-slider') {
-      return <IntegerFieldInputAndSlider nodeId={nodeId} field={field} fieldTemplate={template} />;
+      return <IntegerFieldInput nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
     }
+
+    if (settings.component === 'slider') {
+      return <IntegerFieldSlider nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
+    }
+
+    if (settings.component === 'number-input-and-slider') {
+      return <IntegerFieldInputAndSlider nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
+    }
+
+    assert<Equals<never, typeof settings.component>>(false, 'Unexpected settings.component');
   }
 
   if (isFloatFieldInputTemplate(template)) {
     if (!isFloatFieldInputInstance(field)) {
       return null;
     }
-    if (settings?.type !== 'float-field-config') {
+
+    if (!settings || settings.type !== 'float-field-config') {
       return <FloatFieldInput nodeId={nodeId} field={field} fieldTemplate={template} />;
     }
+
     if (settings.component === 'number-input') {
-      return <FloatFieldInput nodeId={nodeId} field={field} fieldTemplate={template} />;
-    } else if (settings.component === 'slider') {
-      return <FloatFieldSlider nodeId={nodeId} field={field} fieldTemplate={template} />;
-    } else if (settings.component === 'number-input-and-slider') {
-      return <FloatFieldInputAndSlider nodeId={nodeId} field={field} fieldTemplate={template} />;
+      return <FloatFieldInput nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
     }
+
+    if (settings.component === 'slider') {
+      return <FloatFieldSlider nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
+    }
+
+    if (settings.component === 'number-input-and-slider') {
+      return <FloatFieldInputAndSlider nodeId={nodeId} field={field} fieldTemplate={template} settings={settings} />;
+    }
+
+    assert<Equals<never, typeof settings.component>>(false, 'Unexpected settings.component');
   }
 
   if (isIntegerFieldCollectionInputTemplate(template)) {
@@ -303,6 +325,13 @@ export const InputFieldRenderer = memo(({ nodeId, fieldName, settings }: Props)
     return <ControlLoRAModelFieldInputComponent nodeId={nodeId} field={field} fieldTemplate={template} />;
   }
 
+  if (isLLaVAModelFieldInputTemplate(template)) {
+    if (!isLLaVAModelFieldInputInstance(field)) {
+      return null;
+    }
+    return <LLaVAModelFieldInputComponent nodeId={nodeId} field={field} fieldTemplate={template} />;
+  }
+
   if (isFluxVAEModelFieldInputTemplate(template)) {
     if (!isFluxVAEModelFieldInputInstance(field)) {
       return null;

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/InputFieldTooltipContent.tsx

@@ -1,4 +1,5 @@
-import { Flex, Text } from '@invoke-ai/ui-library';
+import { Flex, ListItem, Text, UnorderedList } from '@invoke-ai/ui-library';
+import { useInputFieldErrors } from 'features/nodes/hooks/useInputFieldErrors';
 import { useInputFieldInstance } from 'features/nodes/hooks/useInputFieldInstance';
 import { useInputFieldTemplate } from 'features/nodes/hooks/useInputFieldTemplate';
 import { useFieldTypeName } from 'features/nodes/hooks/usePrettyFieldType';
@@ -17,6 +18,7 @@ export const InputFieldTooltipContent = memo(({ nodeId, fieldName }: Props) => {
   const fieldInstance = useInputFieldInstance(nodeId, fieldName);
   const fieldTemplate = useInputFieldTemplate(nodeId, fieldName);
   const fieldTypeName = useFieldTypeName(fieldTemplate.type);
+  const fieldErrors = useInputFieldErrors(nodeId, fieldName);
 
   const fieldTitle = useMemo(() => {
     if (fieldInstance.label && fieldTemplate.title) {
@@ -42,6 +44,16 @@ export const InputFieldTooltipContent = memo(({ nodeId, fieldName }: Props) => {
       <Text>
         {t('common.input')}: {startCase(fieldTemplate.input)}
       </Text>
+      {fieldErrors.length > 0 && (
+        <>
+          <Text color="error.500">{t('common.error_withCount', { count: fieldErrors.length })}:</Text>
+          <UnorderedList>
+            {fieldErrors.map(({ issue }) => (
+              <ListItem key={issue}>{issue}</ListItem>
+            ))}
+          </UnorderedList>
+        </>
+      )}
     </Flex>
   );
 });

invokeai/frontend/web/src/features/nodes/components/flow/nodes/Invocation/fields/IntegerField/IntegerFieldInput.tsx

@@ -3,23 +3,37 @@ import type { FieldComponentProps } from 'features/nodes/components/flow/nodes/I
 import { useIntegerField } from 'features/nodes/components/flow/nodes/Invocation/fields/IntegerField/useIntegerField';
 import { NO_DRAG_CLASS } from 'features/nodes/types/constants';
 import type { IntegerFieldInputInstance, IntegerFieldInputTemplate } from 'features/nodes/types/field';
+import type { NodeFieldIntegerSettings } from 'features/nodes/types/workflow';
 import { memo } from 'react';
 
 export const IntegerFieldInput = memo(
-  (props: FieldComponentProps<IntegerFieldInputInstance, IntegerFieldInputTemplate>) => {
-    const { defaultValue, onChange, value, min, max, step, fineStep } = useIntegerField(props);
+  (
+    props: FieldComponentProps<
+      IntegerFieldInputInstance,
+      IntegerFieldInputTemplate,
+      { settings?: NodeFieldIntegerSettings }
+    >
+  ) => {
+    const { nodeId, field, fieldTemplate, settings } = props;
+    const { defaultValue, onChange, min, max, step, fineStep, constrainValue } = useIntegerField(
+      nodeId,
+      field.name,
+      fieldTemplate,
+      settings
+    );
 
     return (
       <CompositeNumberInput
         defaultValue={defaultValue}
         onChange={onChange}
-        value={value}
+        value={field.value}
         min={min}
         max={max}
         step={step}
         fineStep={fineStep}
         className={NO_DRAG_CLASS}
         flex="1 1 0"
+        constrainValue={constrainValue}
       />
     );
   }