Release/3.1.0 (#4397)

## What type of PR is this? (check all applicable)

This is the 3.1.0 release candidate. Minor bugfixes will be applied here
during testing and then merged into main upon release.

Author: lstein

installer/create_installer.sh

@@ -46,6 +46,7 @@ if [[ $(python -c 'from importlib.util import find_spec; print(find_spec("build"
     pip install --user build
 fi
 
+rm -r ../build
 python -m build --wheel --outdir dist/ ../.
 
 # ----------------------

invokeai/app/invocations/compel.py

@@ -302,6 +302,29 @@ def invoke(self, context: InvocationContext) -> ConditioningOutput:
 
         add_time_ids = torch.tensor([original_size + crop_coords + target_size])
 
+        # [1, 77, 768], [1, 154, 1280]
+        if c1.shape[1] < c2.shape[1]:
+            c1 = torch.cat(
+                [
+                    c1,
+                    torch.zeros(
+                        (c1.shape[0], c2.shape[1] - c1.shape[1], c1.shape[2]), device=c1.device, dtype=c1.dtype
+                    ),
+                ],
+                dim=1,
+            )
+
+        elif c1.shape[1] > c2.shape[1]:
+            c2 = torch.cat(
+                [
+                    c2,
+                    torch.zeros(
+                        (c2.shape[0], c1.shape[1] - c2.shape[1], c2.shape[2]), device=c2.device, dtype=c2.dtype
+                    ),
+                ],
+                dim=1,
+            )
+
         conditioning_data = ConditioningFieldData(
             conditionings=[
                 SDXLConditioningInfo(

invokeai/app/invocations/latent.py

@@ -445,8 +445,14 @@ def invoke(self, context: InvocationContext) -> LatentsOutput:
                 latents = context.services.latents.get(self.latents.latents_name)
                 if seed is None:
                     seed = self.latents.seed
-            else:
+
+                if noise is not None and noise.shape[1:] != latents.shape[1:]:
+                    raise Exception(f"Incompatable 'noise' and 'latents' shapes: {latents.shape=} {noise.shape=}")
+
+            elif noise is not None:
                 latents = torch.zeros_like(noise)
+            else:
+                raise Exception("'latents' or 'noise' must be provided!")
 
             if seed is None:
                 seed = 0

invokeai/app/invocations/primitives.py

@@ -109,7 +109,7 @@ class IntegerCollectionInvocation(BaseInvocation):
     """A collection of integer primitive values"""
 
     collection: list[int] = InputField(
-        default=0, description="The collection of integer values", ui_type=UIType.IntegerCollection
+        default_factory=list, description="The collection of integer values", ui_type=UIType.IntegerCollection
     )
 
     def invoke(self, context: InvocationContext) -> IntegerCollectionOutput:
@@ -261,7 +261,7 @@ class ImageCollectionInvocation(BaseInvocation):
     """A collection of image primitive values"""
 
     collection: list[ImageField] = InputField(
-        default=0, description="The collection of image values", ui_type=UIType.ImageCollection
+        default_factory=list, description="The collection of image values", ui_type=UIType.ImageCollection
     )
 
     def invoke(self, context: InvocationContext) -> ImageCollectionOutput:
@@ -451,7 +451,9 @@ class ConditioningCollectionInvocation(BaseInvocation):
     """A collection of conditioning tensor primitive values"""
 
     collection: list[ConditioningField] = InputField(
-        default=0, description="The collection of conditioning tensors", ui_type=UIType.ConditioningCollection
+        default_factory=list,
+        description="The collection of conditioning tensors",
+        ui_type=UIType.ConditioningCollection,
     )
 
     def invoke(self, context: InvocationContext) -> ConditioningCollectionOutput:

invokeai/app/services/config/__init__.py

@@ -6,3 +6,4 @@
     InvokeAIAppConfig,
     get_invokeai_config,
 )
+from .base import PagingArgumentParser  # noqa F401

invokeai/backend/install/migrate_to_3.py

@@ -492,10 +492,10 @@ def _parse_legacy_yamlfile(root: Path, initfile: Path) -> ModelPaths:
     loras = paths.get("lora_dir", "loras")
     controlnets = paths.get("controlnet_dir", "controlnets")
     return ModelPaths(
-        models=root / models,
-        embeddings=root / embeddings,
-        loras=root / loras,
-        controlnets=root / controlnets,
+        models=root / models if models else None,
+        embeddings=root / embeddings if embeddings else None,
+        loras=root / loras if loras else None,
+        controlnets=root / controlnets if controlnets else None,
     )
 
 

invokeai/backend/stable_diffusion/diffusion/cross_attention_control.py

@@ -265,7 +265,7 @@ def einsum_op_mps_v1(self, q, k, v):
         if q.shape[1] <= 4096:  # (512x512) max q.shape[1]: 4096
             return self.einsum_lowest_level(q, k, v, None, None, None)
         else:
-            slice_size = math.floor(2**30 / (q.shape[0] * q.shape[1]))
+            slice_size = math.floor(2 ** 30 / (q.shape[0] * q.shape[1]))
             return self.einsum_op_slice_dim1(q, k, v, slice_size)
 
     def einsum_op_mps_v2(self, q, k, v):

invokeai/backend/stable_diffusion/diffusion/shared_invokeai_diffusion.py

@@ -215,10 +215,7 @@ def do_controlnet_step(
                                 dim=0,
                             ),
                         }
-                    (
-                        encoder_hidden_states,
-                        encoder_attention_mask,
-                    ) = self._concat_conditionings_for_batch(
+                    (encoder_hidden_states, encoder_attention_mask,) = self._concat_conditionings_for_batch(
                         conditioning_data.unconditioned_embeddings.embeds,
                         conditioning_data.text_embeddings.embeds,
                     )
@@ -280,32 +277,23 @@ def do_unet_step(
         wants_cross_attention_control = len(cross_attention_control_types_to_do) > 0
 
         if wants_cross_attention_control:
-            (
-                unconditioned_next_x,
-                conditioned_next_x,
-            ) = self._apply_cross_attention_controlled_conditioning(
+            (unconditioned_next_x, conditioned_next_x,) = self._apply_cross_attention_controlled_conditioning(
                 sample,
                 timestep,
                 conditioning_data,
                 cross_attention_control_types_to_do,
                 **kwargs,
             )
         elif self.sequential_guidance:
-            (
-                unconditioned_next_x,
-                conditioned_next_x,
-            ) = self._apply_standard_conditioning_sequentially(
+            (unconditioned_next_x, conditioned_next_x,) = self._apply_standard_conditioning_sequentially(
                 sample,
                 timestep,
                 conditioning_data,
                 **kwargs,
             )
 
         else:
-            (
-                unconditioned_next_x,
-                conditioned_next_x,
-            ) = self._apply_standard_conditioning(
+            (unconditioned_next_x, conditioned_next_x,) = self._apply_standard_conditioning(
                 sample,
                 timestep,
                 conditioning_data,

invokeai/backend/stable_diffusion/image_degradation/bsrgan.py

@@ -395,7 +395,7 @@ def add_Gaussian_noise(img, noise_level1=2, noise_level2=25):
         D = np.diag(np.random.rand(3))
         U = orth(np.random.rand(3, 3))
         conv = np.dot(np.dot(np.transpose(U), D), U)
-        img = img + np.random.multivariate_normal([0, 0, 0], np.abs(L**2 * conv), img.shape[:2]).astype(np.float32)
+        img = img + np.random.multivariate_normal([0, 0, 0], np.abs(L ** 2 * conv), img.shape[:2]).astype(np.float32)
     img = np.clip(img, 0.0, 1.0)
     return img
 
@@ -413,7 +413,7 @@ def add_speckle_noise(img, noise_level1=2, noise_level2=25):
         D = np.diag(np.random.rand(3))
         U = orth(np.random.rand(3, 3))
         conv = np.dot(np.dot(np.transpose(U), D), U)
-        img += img * np.random.multivariate_normal([0, 0, 0], np.abs(L**2 * conv), img.shape[:2]).astype(np.float32)
+        img += img * np.random.multivariate_normal([0, 0, 0], np.abs(L ** 2 * conv), img.shape[:2]).astype(np.float32)
     img = np.clip(img, 0.0, 1.0)
     return img
 

invokeai/backend/stable_diffusion/image_degradation/bsrgan_light.py

@@ -399,7 +399,7 @@ def add_Gaussian_noise(img, noise_level1=2, noise_level2=25):
         D = np.diag(np.random.rand(3))
         U = orth(np.random.rand(3, 3))
         conv = np.dot(np.dot(np.transpose(U), D), U)
-        img = img + np.random.multivariate_normal([0, 0, 0], np.abs(L**2 * conv), img.shape[:2]).astype(np.float32)
+        img = img + np.random.multivariate_normal([0, 0, 0], np.abs(L ** 2 * conv), img.shape[:2]).astype(np.float32)
     img = np.clip(img, 0.0, 1.0)
     return img
 
@@ -417,7 +417,7 @@ def add_speckle_noise(img, noise_level1=2, noise_level2=25):
         D = np.diag(np.random.rand(3))
         U = orth(np.random.rand(3, 3))
         conv = np.dot(np.dot(np.transpose(U), D), U)
-        img += img * np.random.multivariate_normal([0, 0, 0], np.abs(L**2 * conv), img.shape[:2]).astype(np.float32)
+        img += img * np.random.multivariate_normal([0, 0, 0], np.abs(L ** 2 * conv), img.shape[:2]).astype(np.float32)
     img = np.clip(img, 0.0, 1.0)
     return img