diffusers 0.23.0__py3-none-any.whl → 0.24.0__py3-none-any.whl

diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py

@@ -16,11 +16,18 @@ import inspect
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import torch
-from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
+from transformers import (
+    CLIPImageProcessor,
+    CLIPTextModel,
+    CLIPTextModelWithProjection,
+    CLIPTokenizer,
+    CLIPVisionModelWithProjection,
+)
 
-from ...image_processor import VaeImageProcessor
+from ...image_processor import PipelineImageInput, VaeImageProcessor
 from ...loaders import (
     FromSingleFileMixin,
+    IPAdapterMixin,
     StableDiffusionXLLoraLoaderMixin,
     TextualInversionLoaderMixin,
 )
@@ -93,8 +100,57 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
     return noise_cfg
 
 
+# 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 StableDiffusionXLPipeline(
-    DiffusionPipeline, FromSingleFileMixin, StableDiffusionXLLoraLoaderMixin, TextualInversionLoaderMixin
+    DiffusionPipeline,
+    FromSingleFileMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    TextualInversionLoaderMixin,
+    IPAdapterMixin,
 ):
     r"""
     Pipeline for text-to-image generation using Stable Diffusion XL.
@@ -140,8 +196,16 @@ class StableDiffusionXLPipeline(
             watermark output images. If not defined, it will default to True if the package is installed, otherwise no
             watermarker will be used.
     """
+
     model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
-    _optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
+    _optional_components = [
+        "tokenizer",
+        "tokenizer_2",
+        "text_encoder",
+        "text_encoder_2",
+        "image_encoder",
+        "feature_extractor",
+    ]
     _callback_tensor_inputs = [
         "latents",
         "prompt_embeds",
@@ -161,6 +225,8 @@ class StableDiffusionXLPipeline(
         tokenizer_2: CLIPTokenizer,
         unet: UNet2DConditionModel,
         scheduler: KarrasDiffusionSchedulers,
+        image_encoder: CLIPVisionModelWithProjection = None,
+        feature_extractor: CLIPImageProcessor = None,
         force_zeros_for_empty_prompt: bool = True,
         add_watermarker: Optional[bool] = None,
     ):
@@ -174,6 +240,8 @@ class StableDiffusionXLPipeline(
             tokenizer_2=tokenizer_2,
             unet=unet,
             scheduler=scheduler,
+            image_encoder=image_encoder,
+            feature_extractor=feature_extractor,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
@@ -455,6 +523,20 @@ class StableDiffusionXLPipeline(
 
         return prompt_embeds, negative_prompt_embeds, pooled_prompt_embeds, negative_pooled_prompt_embeds
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt):
+        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)
+        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.prepare_extra_step_kwargs
     def prepare_extra_step_kwargs(self, generator, eta):
         # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
@@ -705,6 +787,7 @@ class StableDiffusionXLPipeline(
         height: Optional[int] = None,
         width: Optional[int] = None,
         num_inference_steps: int = 50,
+        timesteps: List[int] = None,
         denoising_end: Optional[float] = None,
         guidance_scale: float = 5.0,
         negative_prompt: Optional[Union[str, List[str]]] = None,
@@ -717,6 +800,7 @@ class StableDiffusionXLPipeline(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_pooled_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,
@@ -755,6 +839,10 @@ class StableDiffusionXLPipeline(
             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.
+            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.
             denoising_end (`float`, *optional*):
                 When specified, determines the fraction (between 0.0 and 1.0) of the total denoising process to be
                 completed before it is intentionally prematurely terminated. As a result, the returned sample will
@@ -801,6 +889,7 @@ class StableDiffusionXLPipeline(
                 Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, pooled negative_prompt_embeds will be generated from `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 generate image. Choose between
                 [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
@@ -853,7 +942,7 @@ class StableDiffusionXLPipeline(
             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:
 
@@ -945,9 +1034,7 @@ class StableDiffusionXLPipeline(
         )
 
         # 4. Prepare timesteps
-        self.scheduler.set_timesteps(num_inference_steps, device=device)
-
-        timesteps = self.scheduler.timesteps
+        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
 
         # 5. Prepare latent variables
         num_channels_latents = self.unet.config.in_channels
@@ -999,6 +1086,12 @@ class StableDiffusionXLPipeline(
         add_text_embeds = add_text_embeds.to(device)
         add_time_ids = add_time_ids.to(device).repeat(batch_size * num_images_per_prompt, 1)
 
+        if ip_adapter_image is not None:
+            image_embeds, negative_image_embeds = self.encode_image(ip_adapter_image, device, num_images_per_prompt)
+            if self.do_classifier_free_guidance:
+                image_embeds = torch.cat([negative_image_embeds, image_embeds])
+                image_embeds = image_embeds.to(device)
+
         # 8. Denoising loop
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
 
@@ -1036,6 +1129,8 @@ class StableDiffusionXLPipeline(
 
                 # predict the noise residual
                 added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
+                if ip_adapter_image is not None:
+                    added_cond_kwargs["image_embeds"] = image_embeds
                 noise_pred = self.unet(
                     latent_model_input,
                     t,

diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py

@@ -17,10 +17,21 @@ from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import PIL.Image
 import torch
-from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
+from transformers import (
+    CLIPImageProcessor,
+    CLIPTextModel,
+    CLIPTextModelWithProjection,
+    CLIPTokenizer,
+    CLIPVisionModelWithProjection,
+)
 
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import FromSingleFileMixin, StableDiffusionXLLoraLoaderMixin, TextualInversionLoaderMixin
+from ...loaders import (
+    FromSingleFileMixin,
+    IPAdapterMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    TextualInversionLoaderMixin,
+)
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...models.attention_processor import (
     AttnProcessor2_0,
@@ -94,17 +105,70 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
 
 
 # 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 StableDiffusionXLImg2ImgPipeline(
-    DiffusionPipeline, TextualInversionLoaderMixin, FromSingleFileMixin, StableDiffusionXLLoraLoaderMixin
+    DiffusionPipeline,
+    TextualInversionLoaderMixin,
+    FromSingleFileMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    IPAdapterMixin,
 ):
     r"""
     Pipeline for text-to-image generation using Stable Diffusion XL.
@@ -153,8 +217,16 @@ class StableDiffusionXLImg2ImgPipeline(
             watermark output images. If not defined, it will default to True if the package is installed, otherwise no
             watermarker will be used.
     """
+
     model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
-    _optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
+    _optional_components = [
+        "tokenizer",
+        "tokenizer_2",
+        "text_encoder",
+        "text_encoder_2",
+        "image_encoder",
+        "feature_extractor",
+    ]
     _callback_tensor_inputs = [
         "latents",
         "prompt_embeds",
@@ -174,6 +246,8 @@ class StableDiffusionXLImg2ImgPipeline(
         tokenizer_2: CLIPTokenizer,
         unet: UNet2DConditionModel,
         scheduler: KarrasDiffusionSchedulers,
+        image_encoder: CLIPVisionModelWithProjection = None,
+        feature_extractor: CLIPImageProcessor = None,
         requires_aesthetics_score: bool = False,
         force_zeros_for_empty_prompt: bool = True,
         add_watermarker: Optional[bool] = None,
@@ -187,6 +261,8 @@ class StableDiffusionXLImg2ImgPipeline(
             tokenizer=tokenizer,
             tokenizer_2=tokenizer_2,
             unet=unet,
+            image_encoder=image_encoder,
+            feature_extractor=feature_extractor,
             scheduler=scheduler,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
@@ -664,6 +740,20 @@ class StableDiffusionXLImg2ImgPipeline(
 
         return latents
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt):
+        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)
+        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
+
     def _get_add_time_ids(
         self,
         original_size,
@@ -836,6 +926,7 @@ class StableDiffusionXLImg2ImgPipeline(
         image: PipelineImageInput = None,
         strength: float = 0.3,
         num_inference_steps: int = 50,
+        timesteps: List[int] = None,
         denoising_start: Optional[float] = None,
         denoising_end: Optional[float] = None,
         guidance_scale: float = 5.0,
@@ -849,6 +940,7 @@ class StableDiffusionXLImg2ImgPipeline(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_pooled_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,
@@ -888,6 +980,10 @@ class StableDiffusionXLImg2ImgPipeline(
             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.
+            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.
             denoising_start (`float`, *optional*):
                 When specified, indicates the fraction (between 0.0 and 1.0) of the total denoising process to be
                 bypassed before it is initiated. Consequently, the initial part of the denoising process is skipped and
@@ -942,6 +1038,7 @@ class StableDiffusionXLImg2ImgPipeline(
                 Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, pooled negative_prompt_embeds will be generated from `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 generate image. Choose between
                 [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
@@ -1005,7 +1102,7 @@ class StableDiffusionXLImg2ImgPipeline(
             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:
 
@@ -1094,7 +1191,7 @@ class StableDiffusionXLImg2ImgPipeline(
         def denoising_value_valid(dnv):
             return isinstance(self.denoising_end, float) and 0 < dnv < 1
 
-        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,
             strength,
@@ -1161,6 +1258,12 @@ class StableDiffusionXLImg2ImgPipeline(
         add_text_embeds = add_text_embeds.to(device)
         add_time_ids = add_time_ids.to(device)
 
+        if ip_adapter_image is not None:
+            image_embeds, negative_image_embeds = self.encode_image(ip_adapter_image, device, num_images_per_prompt)
+            if self.do_classifier_free_guidance:
+                image_embeds = torch.cat([negative_image_embeds, image_embeds])
+                image_embeds = image_embeds.to(device)
+
         # 9. Denoising loop
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
 
@@ -1204,6 +1307,8 @@ class StableDiffusionXLImg2ImgPipeline(
 
                 # predict the noise residual
                 added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
+                if ip_adapter_image is not None:
+                    added_cond_kwargs["image_embeds"] = image_embeds
                 noise_pred = self.unet(
                     latent_model_input,
                     t,

diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py

@@ -18,10 +18,21 @@ from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 import numpy as np
 import PIL.Image
 import torch
-from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
+from transformers import (
+    CLIPImageProcessor,
+    CLIPTextModel,
+    CLIPTextModelWithProjection,
+    CLIPTokenizer,
+    CLIPVisionModelWithProjection,
+)
 
 from ...image_processor import PipelineImageInput, VaeImageProcessor
-from ...loaders import FromSingleFileMixin, StableDiffusionXLLoraLoaderMixin, TextualInversionLoaderMixin
+from ...loaders import (
+    FromSingleFileMixin,
+    IPAdapterMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    TextualInversionLoaderMixin,
+)
 from ...models import AutoencoderKL, UNet2DConditionModel
 from ...models.attention_processor import (
     AttnProcessor2_0,
@@ -239,17 +250,70 @@ 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 StableDiffusionXLInpaintPipeline(
-    DiffusionPipeline, TextualInversionLoaderMixin, StableDiffusionXLLoraLoaderMixin, FromSingleFileMixin
+    DiffusionPipeline,
+    TextualInversionLoaderMixin,
+    StableDiffusionXLLoraLoaderMixin,
+    FromSingleFileMixin,
+    IPAdapterMixin,
 ):
     r"""
     Pipeline for text-to-image generation using Stable Diffusion XL.
@@ -298,9 +362,17 @@ class StableDiffusionXLInpaintPipeline(
             watermark output images. If not defined, it will default to True if the package is installed, otherwise no
             watermarker will be used.
     """
+
     model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
 
-    _optional_components = ["tokenizer", "tokenizer_2", "text_encoder", "text_encoder_2"]
+    _optional_components = [
+        "tokenizer",
+        "tokenizer_2",
+        "text_encoder",
+        "text_encoder_2",
+        "image_encoder",
+        "feature_extractor",
+    ]
     _callback_tensor_inputs = [
         "latents",
         "prompt_embeds",
@@ -322,6 +394,8 @@ class StableDiffusionXLInpaintPipeline(
         tokenizer_2: CLIPTokenizer,
         unet: UNet2DConditionModel,
         scheduler: KarrasDiffusionSchedulers,
+        image_encoder: CLIPVisionModelWithProjection = None,
+        feature_extractor: CLIPImageProcessor = None,
         requires_aesthetics_score: bool = False,
         force_zeros_for_empty_prompt: bool = True,
         add_watermarker: Optional[bool] = None,
@@ -335,6 +409,8 @@ class StableDiffusionXLInpaintPipeline(
             tokenizer=tokenizer,
             tokenizer_2=tokenizer_2,
             unet=unet,
+            image_encoder=image_encoder,
+            feature_extractor=feature_extractor,
             scheduler=scheduler,
         )
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
@@ -385,6 +461,20 @@ class StableDiffusionXLInpaintPipeline(
         """
         self.vae.disable_tiling()
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_image
+    def encode_image(self, image, device, num_images_per_prompt):
+        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)
+        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_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.encode_prompt
     def encode_prompt(
         self,
@@ -741,10 +831,11 @@ class StableDiffusionXLInpaintPipeline(
 
         if image.shape[1] == 4:
             image_latents = image.to(device=device, dtype=dtype)
+            image_latents = image_latents.repeat(batch_size // image_latents.shape[0], 1, 1, 1)
         elif return_image_latents or (latents is None and not is_strength_max):
             image = image.to(device=device, dtype=dtype)
             image_latents = self._encode_vae_image(image=image, generator=generator)
-        image_latents = image_latents.repeat(batch_size // image_latents.shape[0], 1, 1, 1)
+            image_latents = image_latents.repeat(batch_size // image_latents.shape[0], 1, 1, 1)
 
         if latents is None and add_noise:
             noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
@@ -1059,6 +1150,7 @@ class StableDiffusionXLInpaintPipeline(
         width: Optional[int] = None,
         strength: float = 0.9999,
         num_inference_steps: int = 50,
+        timesteps: List[int] = None,
         denoising_start: Optional[float] = None,
         denoising_end: Optional[float] = None,
         guidance_scale: float = 7.5,
@@ -1072,6 +1164,7 @@ class StableDiffusionXLInpaintPipeline(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_pooled_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,
@@ -1128,6 +1221,10 @@ class StableDiffusionXLInpaintPipeline(
             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.
+            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.
             denoising_start (`float`, *optional*):
                 When specified, indicates the fraction (between 0.0 and 1.0) of the total denoising process to be
                 bypassed before it is initiated. Consequently, the initial part of the denoising process is skipped and
@@ -1170,6 +1267,7 @@ class StableDiffusionXLInpaintPipeline(
                 Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_prompt`
                 input argument.
+            ip_adapter_image: (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.
             num_images_per_prompt (`int`, *optional*, defaults to 1):
                 The number of images to generate per prompt.
             eta (`float`, *optional*, defaults to 0.0):
@@ -1240,7 +1338,7 @@ class StableDiffusionXLInpaintPipeline(
             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:
 
@@ -1332,7 +1430,7 @@ class StableDiffusionXLInpaintPipeline(
         def denoising_value_valid(dnv):
             return isinstance(self.denoising_end, float) and 0 < dnv < 1
 
-        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,
             strength,
@@ -1469,6 +1567,12 @@ class StableDiffusionXLInpaintPipeline(
         add_text_embeds = add_text_embeds.to(device)
         add_time_ids = add_time_ids.to(device)
 
+        if ip_adapter_image is not None:
+            image_embeds, negative_image_embeds = self.encode_image(ip_adapter_image, device, num_images_per_prompt)
+            if self.do_classifier_free_guidance:
+                image_embeds = torch.cat([negative_image_embeds, image_embeds])
+                image_embeds = image_embeds.to(device)
+
         # 11. Denoising loop
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
 
@@ -1515,6 +1619,8 @@ class StableDiffusionXLInpaintPipeline(
 
                 # predict the noise residual
                 added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
+                if ip_adapter_image is not None:
+                    added_cond_kwargs["image_embeds"] = image_embeds
                 noise_pred = self.unet(
                     latent_model_input,
                     t,