diffusers 0.27.0__py3-none-any.whl → 0.32.2__py3-none-any.whl

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py

@@ -13,15 +13,16 @@
 # limitations under the License.
 
 import inspect
-from typing import Callable, Dict, List, Optional, Union
+from typing import Any, Callable, Dict, List, Optional, Union
 
 import numpy as np
 import PIL.Image
 import torch
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
+from ...callbacks import MultiPipelineCallbacks, PipelineCallback
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...loaders import IPAdapterMixin, StableDiffusionLoraLoaderMixin, TextualInversionLoaderMixin
 from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import PIL_INTERPOLATION, deprecate, logging
@@ -73,7 +74,11 @@ def retrieve_latents(
 
 
 class StableDiffusionInstructPix2PixPipeline(
-    DiffusionPipeline, StableDiffusionMixin, TextualInversionLoaderMixin, LoraLoaderMixin, IPAdapterMixin
+    DiffusionPipeline,
+    StableDiffusionMixin,
+    TextualInversionLoaderMixin,
+    StableDiffusionLoraLoaderMixin,
+    IPAdapterMixin,
 ):
     r"""
     Pipeline for pixel-level image editing by following text instructions (based on Stable Diffusion).
@@ -83,8 +88,8 @@ class StableDiffusionInstructPix2PixPipeline(
 
     The pipeline also inherits the following loading methods:
         - [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
-        - [`~loaders.LoraLoaderMixin.load_lora_weights`] for loading LoRA weights
-        - [`~loaders.LoraLoaderMixin.save_lora_weights`] for saving LoRA weights
+        - [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] for loading LoRA weights
+        - [`~loaders.StableDiffusionLoraLoaderMixin.save_lora_weights`] for saving LoRA weights
         - [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
 
     Args:
@@ -168,14 +173,18 @@ class StableDiffusionInstructPix2PixPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
+        ip_adapter_image_embeds: Optional[List[torch.Tensor]] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
+        callback_on_step_end: Optional[
+            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
+        ] = None,
         callback_on_step_end_tensor_inputs: List[str] = ["latents"],
+        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         **kwargs,
     ):
         r"""
@@ -184,7 +193,7 @@ class StableDiffusionInstructPix2PixPipeline(
         Args:
             prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts to guide image generation. If not defined, you need to pass `prompt_embeds`.
-            image (`torch.FloatTensor` `np.ndarray`, `PIL.Image.Image`, `List[torch.FloatTensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`):
+            image (`torch.Tensor` `np.ndarray`, `PIL.Image.Image`, `List[torch.Tensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`):
                 `Image` or tensor representing an image batch to be repainted according to `prompt`. Can also accept
                 image latents as `image`, but if passing latents directly it is not encoded again.
             num_inference_steps (`int`, *optional*, defaults to 100):
@@ -194,7 +203,7 @@ class StableDiffusionInstructPix2PixPipeline(
                 A higher guidance scale value encourages the model to generate images closely linked to the text
                 `prompt` at the expense of lower image quality. Guidance scale is enabled when `guidance_scale > 1`.
             image_guidance_scale (`float`, *optional*, defaults to 1.5):
-                Push the generated image towards the inital `image`. Image guidance scale is enabled by setting
+                Push the generated image towards the initial `image`. Image guidance scale is enabled by setting
                 `image_guidance_scale > 1`. Higher image guidance scale encourages generated images that are closely
                 linked to the source `image`, usually at the expense of lower image quality. This pipeline requires a
                 value of at least `1`.
@@ -209,14 +218,14 @@ class StableDiffusionInstructPix2PixPipeline(
             generator (`torch.Generator`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             ip_adapter_image: (`PipelineImageInput`, *optional*):
@@ -226,15 +235,18 @@ class StableDiffusionInstructPix2PixPipeline(
             return_dict (`bool`, *optional*, defaults to `True`):
                 Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a
                 plain tuple.
-            callback_on_step_end (`Callable`, *optional*):
-                A function that calls at the end of each denoising steps during the inference. The function is called
-                with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
-                callback_kwargs: Dict)`. `callback_kwargs` will include a list of all tensors as specified by
-                `callback_on_step_end_tensor_inputs`.
+            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
+                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
+                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
+                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
+                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
             callback_on_step_end_tensor_inputs (`List`, *optional*):
                 The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
                 will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
                 `._callback_tensor_inputs` attribute of your pipeline class.
+            cross_attention_kwargs (`dict`, *optional*):
+                A kwargs dictionary that if specified is passed along to the [`AttentionProcessor`] as defined in
+                [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
 
         Examples:
 
@@ -289,6 +301,9 @@ class StableDiffusionInstructPix2PixPipeline(
                 "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
             )
 
+        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
+            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
+
         # 0. Check inputs
         self.check_inputs(
             prompt,
@@ -296,6 +311,8 @@ class StableDiffusionInstructPix2PixPipeline(
             negative_prompt,
             prompt_embeds,
             negative_prompt_embeds,
+            ip_adapter_image,
+            ip_adapter_image_embeds,
             callback_on_step_end_tensor_inputs,
         )
         self._guidance_scale = guidance_scale
@@ -303,14 +320,6 @@ class StableDiffusionInstructPix2PixPipeline(
 
         device = self._execution_device
 
-        if ip_adapter_image is not None:
-            output_hidden_state = False if isinstance(self.unet.encoder_hid_proj, ImageProjection) else True
-            image_embeds, negative_image_embeds = self.encode_image(
-                ip_adapter_image, device, num_images_per_prompt, output_hidden_state
-            )
-            if self.do_classifier_free_guidance:
-                image_embeds = torch.cat([image_embeds, negative_image_embeds, negative_image_embeds])
-
         if image is None:
             raise ValueError("`image` input cannot be undefined.")
 
@@ -335,6 +344,14 @@ class StableDiffusionInstructPix2PixPipeline(
             negative_prompt_embeds=negative_prompt_embeds,
         )
 
+        if ip_adapter_image is not None or ip_adapter_image_embeds is not None:
+            image_embeds = self.prepare_ip_adapter_image_embeds(
+                ip_adapter_image,
+                ip_adapter_image_embeds,
+                device,
+                batch_size * num_images_per_prompt,
+                self.do_classifier_free_guidance,
+            )
         # 3. Preprocess image
         image = self.image_processor.preprocess(image)
 
@@ -406,6 +423,7 @@ class StableDiffusionInstructPix2PixPipeline(
                     t,
                     encoder_hidden_states=prompt_embeds,
                     added_cond_kwargs=added_cond_kwargs,
+                    cross_attention_kwargs=cross_attention_kwargs,
                     return_dict=False,
                 )[0]
 
@@ -468,8 +486,8 @@ class StableDiffusionInstructPix2PixPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
@@ -487,10 +505,10 @@ class StableDiffusionInstructPix2PixPipeline(
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -635,6 +653,65 @@ class StableDiffusionInstructPix2PixPipeline(
 
             return image_embeds, uncond_image_embeds
 
+    def prepare_ip_adapter_image_embeds(
+        self, ip_adapter_image, ip_adapter_image_embeds, device, num_images_per_prompt, do_classifier_free_guidance
+    ):
+        if ip_adapter_image_embeds is None:
+            if not isinstance(ip_adapter_image, list):
+                ip_adapter_image = [ip_adapter_image]
+
+            if len(ip_adapter_image) != len(self.unet.encoder_hid_proj.image_projection_layers):
+                raise ValueError(
+                    f"`ip_adapter_image` must have same length as the number of IP Adapters. Got {len(ip_adapter_image)} images and {len(self.unet.encoder_hid_proj.image_projection_layers)} IP Adapters."
+                )
+
+            image_embeds = []
+            for single_ip_adapter_image, image_proj_layer in zip(
+                ip_adapter_image, self.unet.encoder_hid_proj.image_projection_layers
+            ):
+                output_hidden_state = not isinstance(image_proj_layer, ImageProjection)
+                single_image_embeds, single_negative_image_embeds = self.encode_image(
+                    single_ip_adapter_image, device, 1, output_hidden_state
+                )
+                single_image_embeds = torch.stack([single_image_embeds] * num_images_per_prompt, dim=0)
+                single_negative_image_embeds = torch.stack(
+                    [single_negative_image_embeds] * num_images_per_prompt, dim=0
+                )
+
+                if do_classifier_free_guidance:
+                    single_image_embeds = torch.cat(
+                        [single_image_embeds, single_negative_image_embeds, single_negative_image_embeds]
+                    )
+                    single_image_embeds = single_image_embeds.to(device)
+
+                image_embeds.append(single_image_embeds)
+        else:
+            repeat_dims = [1]
+            image_embeds = []
+            for single_image_embeds in ip_adapter_image_embeds:
+                if do_classifier_free_guidance:
+                    (
+                        single_image_embeds,
+                        single_negative_image_embeds,
+                        single_negative_image_embeds,
+                    ) = single_image_embeds.chunk(3)
+                    single_image_embeds = single_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_image_embeds.shape[1:]))
+                    )
+                    single_negative_image_embeds = single_negative_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_negative_image_embeds.shape[1:]))
+                    )
+                    single_image_embeds = torch.cat(
+                        [single_image_embeds, single_negative_image_embeds, single_negative_image_embeds]
+                    )
+                else:
+                    single_image_embeds = single_image_embeds.repeat(
+                        num_images_per_prompt, *(repeat_dims * len(single_image_embeds.shape[1:]))
+                    )
+                image_embeds.append(single_image_embeds)
+
+        return image_embeds
+
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.run_safety_checker
     def run_safety_checker(self, image, device, dtype):
         if self.safety_checker is None:
@@ -687,6 +764,8 @@ class StableDiffusionInstructPix2PixPipeline(
         negative_prompt=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
+        ip_adapter_image=None,
+        ip_adapter_image_embeds=None,
         callback_on_step_end_tensor_inputs=None,
     ):
         if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
@@ -728,9 +807,29 @@ class StableDiffusionInstructPix2PixPipeline(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+        if ip_adapter_image is not None and ip_adapter_image_embeds is not None:
+            raise ValueError(
+                "Provide either `ip_adapter_image` or `ip_adapter_image_embeds`. Cannot leave both `ip_adapter_image` and `ip_adapter_image_embeds` defined."
+            )
+
+        if ip_adapter_image_embeds is not None:
+            if not isinstance(ip_adapter_image_embeds, list):
+                raise ValueError(
+                    f"`ip_adapter_image_embeds` has to be of type `list` but is {type(ip_adapter_image_embeds)}"
+                )
+            elif ip_adapter_image_embeds[0].ndim not in [3, 4]:
+                raise ValueError(
+                    f"`ip_adapter_image_embeds` has to be a list of 3D or 4D tensors but is {ip_adapter_image_embeds[0].ndim}D"
+                )
+
     # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
     def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
-        shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
+        shape = (
+            batch_size,
+            num_channels_latents,
+            int(height) // self.vae_scale_factor,
+            int(width) // self.vae_scale_factor,
+        )
         if isinstance(generator, list) and len(generator) != batch_size:
             raise ValueError(
                 f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"