diffusers 0.23.1__py3-none-any.whl → 0.25.0__py3-none-any.whl

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py

@@ -19,12 +19,13 @@ import numpy as np
 import PIL.Image
 import torch
 from packaging import version
-from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
 from ...configuration_utils import FrozenDict
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
-from ...models import AsymmetricAutoencoderKL, AutoencoderKL, UNet2DConditionModel
+from ...loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AsymmetricAutoencoderKL, AutoencoderKL, ImageProjection, UNet2DConditionModel
+from ...models.attention_processor import FusedAttnProcessor2_0
 from ...models.lora import adjust_lora_scale_text_encoder
 from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
@@ -160,17 +161,66 @@ def prepare_mask_and_masked_image(image, mask, height, width, return_image: bool
 
 
 # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents
-def retrieve_latents(encoder_output, generator):
-    if hasattr(encoder_output, "latent_dist"):
+def retrieve_latents(
+    encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample"
+):
+    if hasattr(encoder_output, "latent_dist") and sample_mode == "sample":
         return encoder_output.latent_dist.sample(generator)
+    elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax":
+        return encoder_output.latent_dist.mode()
     elif hasattr(encoder_output, "latents"):
         return encoder_output.latents
     else:
         raise AttributeError("Could not access latents of provided encoder_output")
 
 
+# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps
+def retrieve_timesteps(
+    scheduler,
+    num_inference_steps: Optional[int] = None,
+    device: Optional[Union[str, torch.device]] = None,
+    timesteps: Optional[List[int]] = None,
+    **kwargs,
+):
+    """
+    Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
+    custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
+
+    Args:
+        scheduler (`SchedulerMixin`):
+            The scheduler to get timesteps from.
+        num_inference_steps (`int`):
+            The number of diffusion steps used when generating samples with a pre-trained model. If used,
+            `timesteps` must be `None`.
+        device (`str` or `torch.device`, *optional*):
+            The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
+        timesteps (`List[int]`, *optional*):
+                Custom timesteps used to support arbitrary spacing between timesteps. If `None`, then the default
+                timestep spacing strategy of the scheduler is used. If `timesteps` is passed, `num_inference_steps`
+                must be `None`.
+
+    Returns:
+        `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the
+        second element is the number of inference steps.
+    """
+    if timesteps is not None:
+        accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
+        if not accepts_timesteps:
+            raise ValueError(
+                f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
+                f" timestep schedules. Please check whether you are using the correct scheduler."
+            )
+        scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs)
+        timesteps = scheduler.timesteps
+        num_inference_steps = len(timesteps)
+    else:
+        scheduler.set_timesteps(num_inference_steps, device=device, **kwargs)
+        timesteps = scheduler.timesteps
+    return timesteps, num_inference_steps
+
+
 class StableDiffusionInpaintPipeline(
-    DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin, FromSingleFileMixin
+    DiffusionPipeline, TextualInversionLoaderMixin, IPAdapterMixin, LoraLoaderMixin, FromSingleFileMixin
 ):
     r"""
     Pipeline for text-guided image inpainting using Stable Diffusion.
@@ -182,6 +232,8 @@ class StableDiffusionInpaintPipeline(
         - [`~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.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
+        - [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
 
     Args:
         vae ([`AutoencoderKL`, `AsymmetricAutoencoderKL`]):
@@ -202,8 +254,9 @@ class StableDiffusionInpaintPipeline(
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
-    model_cpu_offload_seq = "text_encoder->unet->vae"
-    _optional_components = ["safety_checker", "feature_extractor"]
+
+    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds", "mask", "masked_image_latents"]
 
@@ -216,6 +269,7 @@ class StableDiffusionInpaintPipeline(
         scheduler: KarrasDiffusionSchedulers,
         safety_checker: StableDiffusionSafetyChecker,
         feature_extractor: CLIPImageProcessor,
+        image_encoder: CLIPVisionModelWithProjection = None,
         requires_safety_checker: bool = True,
     ):
         super().__init__()
@@ -297,6 +351,7 @@ class StableDiffusionInpaintPipeline(
             scheduler=scheduler,
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
+            image_encoder=image_encoder,
         )
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
@@ -520,6 +575,31 @@ class StableDiffusionInpaintPipeline(
 
         return prompt_embeds, negative_prompt_embeds
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt, output_hidden_states=None):
+        dtype = next(self.image_encoder.parameters()).dtype
+
+        if not isinstance(image, torch.Tensor):
+            image = self.feature_extractor(image, return_tensors="pt").pixel_values
+
+        image = image.to(device=device, dtype=dtype)
+        if output_hidden_states:
+            image_enc_hidden_states = self.image_encoder(image, output_hidden_states=True).hidden_states[-2]
+            image_enc_hidden_states = image_enc_hidden_states.repeat_interleave(num_images_per_prompt, dim=0)
+            uncond_image_enc_hidden_states = self.image_encoder(
+                torch.zeros_like(image), output_hidden_states=True
+            ).hidden_states[-2]
+            uncond_image_enc_hidden_states = uncond_image_enc_hidden_states.repeat_interleave(
+                num_images_per_prompt, dim=0
+            )
+            return image_enc_hidden_states, uncond_image_enc_hidden_states
+        else:
+            image_embeds = self.image_encoder(image).image_embeds
+            image_embeds = image_embeds.repeat_interleave(num_images_per_prompt, dim=0)
+            uncond_image_embeds = torch.zeros_like(image_embeds)
+
+            return image_embeds, uncond_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:
@@ -556,6 +636,8 @@ class StableDiffusionInpaintPipeline(
     def check_inputs(
         self,
         prompt,
+        image,
+        mask_image,
         height,
         width,
         strength,
@@ -564,6 +646,7 @@ class StableDiffusionInpaintPipeline(
         prompt_embeds=None,
         negative_prompt_embeds=None,
         callback_on_step_end_tensor_inputs=None,
+        padding_mask_crop=None,
     ):
         if strength < 0 or strength > 1:
             raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}")
@@ -609,6 +692,21 @@ class StableDiffusionInpaintPipeline(
                     f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
                     f" {negative_prompt_embeds.shape}."
                 )
+        if padding_mask_crop is not None:
+            if self.unet.config.in_channels != 4:
+                raise ValueError(
+                    f"The UNet should have 4 input channels for inpainting mask crop, but has"
+                    f" {self.unet.config.in_channels} input channels."
+                )
+            if not isinstance(image, PIL.Image.Image):
+                raise ValueError(
+                    f"The image should be a PIL image when inpainting mask crop, but is of type" f" {type(image)}."
+                )
+            if not isinstance(mask_image, PIL.Image.Image):
+                raise ValueError(
+                    f"The mask image should be a PIL image when inpainting mask crop, but is of type"
+                    f" {type(mask_image)}."
+                )
 
     def prepare_latents(
         self,
@@ -765,6 +863,67 @@ class StableDiffusionInpaintPipeline(
         """Disables the FreeU mechanism if enabled."""
         self.unet.disable_freeu()
 
+    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.fuse_qkv_projections
+    def fuse_qkv_projections(self, unet: bool = True, vae: bool = True):
+        """
+        Enables fused QKV projections. For self-attention modules, all projection matrices (i.e., query,
+        key, value) are fused. For cross-attention modules, key and value projection matrices are fused.
+
+        <Tip warning={true}>
+
+        This API is 🧪 experimental.
+
+        </Tip>
+
+        Args:
+            unet (`bool`, defaults to `True`): To apply fusion on the UNet.
+            vae (`bool`, defaults to `True`): To apply fusion on the VAE.
+        """
+        self.fusing_unet = False
+        self.fusing_vae = False
+
+        if unet:
+            self.fusing_unet = True
+            self.unet.fuse_qkv_projections()
+            self.unet.set_attn_processor(FusedAttnProcessor2_0())
+
+        if vae:
+            if not isinstance(self.vae, AutoencoderKL):
+                raise ValueError("`fuse_qkv_projections()` is only supported for the VAE of type `AutoencoderKL`.")
+
+            self.fusing_vae = True
+            self.vae.fuse_qkv_projections()
+            self.vae.set_attn_processor(FusedAttnProcessor2_0())
+
+    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.unfuse_qkv_projections
+    def unfuse_qkv_projections(self, unet: bool = True, vae: bool = True):
+        """Disable QKV projection fusion if enabled.
+
+        <Tip warning={true}>
+
+        This API is 🧪 experimental.
+
+        </Tip>
+
+        Args:
+            unet (`bool`, defaults to `True`): To apply fusion on the UNet.
+            vae (`bool`, defaults to `True`): To apply fusion on the VAE.
+
+        """
+        if unet:
+            if not self.fusing_unet:
+                logger.warning("The UNet was not initially fused for QKV projections. Doing nothing.")
+            else:
+                self.unet.unfuse_qkv_projections()
+                self.fusing_unet = False
+
+        if vae:
+            if not self.fusing_vae:
+                logger.warning("The VAE was not initially fused for QKV projections. Doing nothing.")
+            else:
+                self.vae.unfuse_qkv_projections()
+                self.fusing_vae = False
+
     # Copied from diffusers.pipelines.latent_consistency_models.pipeline_latent_consistency_text2img.LatentConsistencyModelPipeline.get_guidance_scale_embedding
     def get_guidance_scale_embedding(self, w, embedding_dim=512, dtype=torch.float32):
         """
@@ -817,6 +976,10 @@ class StableDiffusionInpaintPipeline(
     def num_timesteps(self):
         return self._num_timesteps
 
+    @property
+    def interrupt(self):
+        return self._interrupt
+
     @torch.no_grad()
     def __call__(
         self,
@@ -826,8 +989,10 @@ class StableDiffusionInpaintPipeline(
         masked_image_latents: torch.FloatTensor = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
+        padding_mask_crop: Optional[int] = None,
         strength: float = 1.0,
         num_inference_steps: int = 50,
+        timesteps: List[int] = None,
         guidance_scale: float = 7.5,
         negative_prompt: Optional[Union[str, List[str]]] = None,
         num_images_per_prompt: Optional[int] = 1,
@@ -836,6 +1001,7 @@ class StableDiffusionInpaintPipeline(
         latents: Optional[torch.FloatTensor] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        ip_adapter_image: Optional[PipelineImageInput] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -868,6 +1034,12 @@ class StableDiffusionInpaintPipeline(
                 The height in pixels of the generated image.
             width (`int`, *optional*, defaults to `self.unet.config.sample_size * self.vae_scale_factor`):
                 The width in pixels of the generated image.
+            padding_mask_crop (`int`, *optional*, defaults to `None`):
+                The size of margin in the crop to be applied to the image and masking. If `None`, no crop is applied to image and mask_image. If
+                `padding_mask_crop` is not `None`, it will first find a rectangular region with the same aspect ration of the image and
+                contains all masked area, and then expand that area based on `padding_mask_crop`. The image and mask_image will then be cropped based on
+                the expanded area before resizing to the original image size for inpainting. This is useful when the masked area is small while the image is large
+                and contain information inreleant for inpainging, such as background.
             strength (`float`, *optional*, defaults to 1.0):
                 Indicates extent to transform the reference `image`. Must be between 0 and 1. `image` is used as a
                 starting point and more noise is added the higher the `strength`. The number of denoising steps depends
@@ -877,6 +1049,10 @@ class StableDiffusionInpaintPipeline(
             num_inference_steps (`int`, *optional*, defaults to 50):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference. This parameter is modulated by `strength`.
+            timesteps (`List[int]`, *optional*):
+                Custom timesteps to use for the denoising process with schedulers which support a `timesteps` argument
+                in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
+                passed will be used. Must be in descending order.
             guidance_scale (`float`, *optional*, defaults to 7.5):
                 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`.
@@ -901,6 +1077,7 @@ class StableDiffusionInpaintPipeline(
             negative_prompt_embeds (`torch.FloatTensor`, *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*): Optional image input to work with IP Adapters.
             output_type (`str`, *optional*, defaults to `"pil"`):
                 The output format of the generated image. Choose between `PIL.Image` or `np.array`.
             return_dict (`bool`, *optional*, defaults to `True`):
@@ -920,7 +1097,7 @@ class StableDiffusionInpaintPipeline(
             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 pipeine class.
+                `._callback_tensor_inputs` attribute of your pipeline class.
         Examples:
 
         ```py
@@ -983,6 +1160,8 @@ class StableDiffusionInpaintPipeline(
         # 1. Check inputs
         self.check_inputs(
             prompt,
+            image,
+            mask_image,
             height,
             width,
             strength,
@@ -991,11 +1170,13 @@ class StableDiffusionInpaintPipeline(
             prompt_embeds,
             negative_prompt_embeds,
             callback_on_step_end_tensor_inputs,
+            padding_mask_crop,
         )
 
         self._guidance_scale = guidance_scale
         self._clip_skip = clip_skip
         self._cross_attention_kwargs = cross_attention_kwargs
+        self._interrupt = False
 
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
@@ -1028,8 +1209,16 @@ class StableDiffusionInpaintPipeline(
         if self.do_classifier_free_guidance:
             prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])
 
+        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([negative_image_embeds, image_embeds])
+
         # 4. set timesteps
-        self.scheduler.set_timesteps(num_inference_steps, device=device)
+        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
         timesteps, num_inference_steps = self.get_timesteps(
             num_inference_steps=num_inference_steps, strength=strength, device=device
         )
@@ -1046,7 +1235,17 @@ class StableDiffusionInpaintPipeline(
 
         # 5. Preprocess mask and image
 
-        init_image = self.image_processor.preprocess(image, height=height, width=width)
+        if padding_mask_crop is not None:
+            crops_coords = self.mask_processor.get_crop_region(mask_image, width, height, pad=padding_mask_crop)
+            resize_mode = "fill"
+        else:
+            crops_coords = None
+            resize_mode = "default"
+
+        original_image = image
+        init_image = self.image_processor.preprocess(
+            image, height=height, width=width, crops_coords=crops_coords, resize_mode=resize_mode
+        )
         init_image = init_image.to(dtype=torch.float32)
 
         # 6. Prepare latent variables
@@ -1076,7 +1275,9 @@ class StableDiffusionInpaintPipeline(
             latents, noise = latents_outputs
 
         # 7. Prepare mask latent variables
-        mask_condition = self.mask_processor.preprocess(mask_image, height=height, width=width)
+        mask_condition = self.mask_processor.preprocess(
+            mask_image, height=height, width=width, resize_mode=resize_mode, crops_coords=crops_coords
+        )
 
         if masked_image_latents is None:
             masked_image = init_image * (mask_condition < 0.5)
@@ -1116,7 +1317,10 @@ class StableDiffusionInpaintPipeline(
         # 9. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
         extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
 
-        # 9.5 Optionally get Guidance Scale Embedding
+        # 9.1 Add image embeds for IP-Adapter
+        added_cond_kwargs = {"image_embeds": image_embeds} if ip_adapter_image is not None else None
+
+        # 9.2 Optionally get Guidance Scale Embedding
         timestep_cond = None
         if self.unet.config.time_cond_proj_dim is not None:
             guidance_scale_tensor = torch.tensor(self.guidance_scale - 1).repeat(batch_size * num_images_per_prompt)
@@ -1129,6 +1333,9 @@ class StableDiffusionInpaintPipeline(
         self._num_timesteps = len(timesteps)
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
+                if self.interrupt:
+                    continue
+
                 # expand the latents if we are doing classifier free guidance
                 latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
 
@@ -1145,6 +1352,7 @@ class StableDiffusionInpaintPipeline(
                     encoder_hidden_states=prompt_embeds,
                     timestep_cond=timestep_cond,
                     cross_attention_kwargs=self.cross_attention_kwargs,
+                    added_cond_kwargs=added_cond_kwargs,
                     return_dict=False,
                 )[0]
 
@@ -1212,6 +1420,9 @@ class StableDiffusionInpaintPipeline(
 
         image = self.image_processor.postprocess(image, output_type=output_type, do_denormalize=do_denormalize)
 
+        if padding_mask_crop is not None:
+            image = [self.image_processor.apply_overlay(mask_image, original_image, i, crops_coords) for i in image]
+
         # Offload all models
         self.maybe_free_model_hooks()
 

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py

@@ -18,11 +18,11 @@ from typing import Callable, Dict, List, Optional, Union
 import numpy as np
 import PIL.Image
 import torch
-from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
 
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
-from ...models import AutoencoderKL, UNet2DConditionModel
+from ...loaders import IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
 from ...schedulers import KarrasDiffusionSchedulers
 from ...utils import PIL_INTERPOLATION, deprecate, logging
 from ...utils.torch_utils import randn_tensor
@@ -58,7 +58,23 @@ def preprocess(image):
     return image
 
 
-class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin):
+# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents
+def retrieve_latents(
+    encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample"
+):
+    if hasattr(encoder_output, "latent_dist") and sample_mode == "sample":
+        return encoder_output.latent_dist.sample(generator)
+    elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax":
+        return encoder_output.latent_dist.mode()
+    elif hasattr(encoder_output, "latents"):
+        return encoder_output.latents
+    else:
+        raise AttributeError("Could not access latents of provided encoder_output")
+
+
+class StableDiffusionInstructPix2PixPipeline(
+    DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin, IPAdapterMixin
+):
     r"""
     Pipeline for pixel-level image editing by following text instructions (based on Stable Diffusion).
 
@@ -69,6 +85,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         - [`~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.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
 
     Args:
         vae ([`AutoencoderKL`]):
@@ -89,8 +106,9 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         feature_extractor ([`~transformers.CLIPImageProcessor`]):
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
+
     model_cpu_offload_seq = "text_encoder->unet->vae"
-    _optional_components = ["safety_checker", "feature_extractor"]
+    _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "image_latents"]
 
@@ -103,6 +121,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         scheduler: KarrasDiffusionSchedulers,
         safety_checker: StableDiffusionSafetyChecker,
         feature_extractor: CLIPImageProcessor,
+        image_encoder: Optional[CLIPVisionModelWithProjection] = None,
         requires_safety_checker: bool = True,
     ):
         super().__init__()
@@ -131,6 +150,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             scheduler=scheduler,
             safety_checker=safety_checker,
             feature_extractor=feature_extractor,
+            image_encoder=image_encoder,
         )
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
@@ -151,6 +171,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         latents: Optional[torch.FloatTensor] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        ip_adapter_image: Optional[PipelineImageInput] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
@@ -198,6 +219,8 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             negative_prompt_embeds (`torch.FloatTensor`, *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*):
+                Optional image input to work with IP Adapters.
             output_type (`str`, *optional*, defaults to `"pil"`):
                 The output format of the generated image. Choose between `PIL.Image` or `np.array`.
             return_dict (`bool`, *optional*, defaults to `True`):
@@ -211,7 +234,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             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 pipeine class.
+                `._callback_tensor_inputs` attribute of your pipeline class.
 
         Examples:
 
@@ -278,6 +301,16 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         self._guidance_scale = guidance_scale
         self._image_guidance_scale = image_guidance_scale
 
+        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.")
 
@@ -319,7 +352,6 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             prompt_embeds.dtype,
             device,
             self.do_classifier_free_guidance,
-            generator,
         )
 
         height, width = image_latents.shape[-2:]
@@ -353,6 +385,9 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         # 8. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
         extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
 
+        # 8.1 Add image embeds for IP-Adapter
+        added_cond_kwargs = {"image_embeds": image_embeds} if ip_adapter_image is not None else None
+
         # 9. Denoising loop
         num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
         self._num_timesteps = len(timesteps)
@@ -369,7 +404,11 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
 
                 # predict the noise residual
                 noise_pred = self.unet(
-                    scaled_latent_model_input, t, encoder_hidden_states=prompt_embeds, return_dict=False
+                    scaled_latent_model_input,
+                    t,
+                    encoder_hidden_states=prompt_embeds,
+                    added_cond_kwargs=added_cond_kwargs,
+                    return_dict=False,
                 )[0]
 
                 # Hack:
@@ -584,11 +623,36 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
             # For classifier free guidance, we need to do two forward passes.
             # Here we concatenate the unconditional and text embeddings into a single batch
             # to avoid doing two forward passes
-            # pix2pix has two  negative embeddings, and unlike in other pipelines latents are ordered [prompt_embeds, negative_prompt_embeds, negative_prompt_embeds]
+            # pix2pix has two negative embeddings, and unlike in other pipelines latents are ordered [prompt_embeds, negative_prompt_embeds, negative_prompt_embeds]
             prompt_embeds = torch.cat([prompt_embeds, negative_prompt_embeds, negative_prompt_embeds])
 
         return prompt_embeds
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt, output_hidden_states=None):
+        dtype = next(self.image_encoder.parameters()).dtype
+
+        if not isinstance(image, torch.Tensor):
+            image = self.feature_extractor(image, return_tensors="pt").pixel_values
+
+        image = image.to(device=device, dtype=dtype)
+        if output_hidden_states:
+            image_enc_hidden_states = self.image_encoder(image, output_hidden_states=True).hidden_states[-2]
+            image_enc_hidden_states = image_enc_hidden_states.repeat_interleave(num_images_per_prompt, dim=0)
+            uncond_image_enc_hidden_states = self.image_encoder(
+                torch.zeros_like(image), output_hidden_states=True
+            ).hidden_states[-2]
+            uncond_image_enc_hidden_states = uncond_image_enc_hidden_states.repeat_interleave(
+                num_images_per_prompt, dim=0
+            )
+            return image_enc_hidden_states, uncond_image_enc_hidden_states
+        else:
+            image_embeds = self.image_encoder(image).image_embeds
+            image_embeds = image_embeds.repeat_interleave(num_images_per_prompt, dim=0)
+            uncond_image_embeds = torch.zeros_like(image_embeds)
+
+            return image_embeds, uncond_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:
@@ -715,17 +779,7 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
         if image.shape[1] == 4:
             image_latents = image
         else:
-            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"
-                    f" size of {batch_size}. Make sure the batch size matches the length of the generators."
-                )
-
-            if isinstance(generator, list):
-                image_latents = [self.vae.encode(image[i : i + 1]).latent_dist.mode() for i in range(batch_size)]
-                image_latents = torch.cat(image_latents, dim=0)
-            else:
-                image_latents = self.vae.encode(image).latent_dist.mode()
+            image_latents = retrieve_latents(self.vae.encode(image), sample_mode="argmax")
 
         if batch_size > image_latents.shape[0] and batch_size % image_latents.shape[0] == 0:
             # expand image_latents for batch_size

diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py

@@ -67,6 +67,9 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, FromSingleFileMixi
     This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
     implemented for all pipelines (downloading, saving, running on a particular device, etc.).
 
+    The pipeline also inherits the following loading methods:
+        - [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
+
     Args:
         vae ([`AutoencoderKL`]):
             Variational Auto-Encoder (VAE) model to encode and decode images to and from latent representations.
@@ -79,6 +82,7 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline, FromSingleFileMixi
         scheduler ([`SchedulerMixin`]):
             A [`EulerDiscreteScheduler`] to be used in combination with `unet` to denoise the encoded image latents.
     """
+
     model_cpu_offload_seq = "text_encoder->unet->vae"
 
     def __init__(