diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion.py

import inspect
from typing import Callable, List, Optional, Union

import PIL.Image
import torch
from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer, CLIPVisionModel

from ...models import AutoencoderKL, UNet2DConditionModel
from ...schedulers import KarrasDiffusionSchedulers
from ...utils import logging
from ..pipeline_utils import DiffusionPipeline
from .pipeline_versatile_diffusion_dual_guided import VersatileDiffusionDualGuidedPipeline
from .pipeline_versatile_diffusion_image_variation import VersatileDiffusionImageVariationPipeline
from .pipeline_versatile_diffusion_text_to_image import VersatileDiffusionTextToImagePipeline


logger = logging.get_logger(__name__)  # pylint: disable=invalid-name


class VersatileDiffusionPipeline(DiffusionPipeline):
    r"""
    Pipeline for text-to-image generation using Stable Diffusion.

    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the
    library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)

    Args:
        vae ([`AutoencoderKL`]):
            Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.
        text_encoder ([`CLIPTextModel`]):
            Frozen text-encoder. Stable Diffusion uses the text portion of
            [CLIP](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), specifically
            the [clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14) variant.
        tokenizer (`CLIPTokenizer`):
            Tokenizer of class
            [CLIPTokenizer](https://huggingface.co/docs/transformers/v4.21.0/en/model_doc/clip#transformers.CLIPTokenizer).
        unet ([`UNet2DConditionModel`]): Conditional U-Net architecture to denoise the encoded image latents.
        scheduler ([`SchedulerMixin`]):
            A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
            [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
        safety_checker ([`StableDiffusionMegaSafetyChecker`]):
            Classification module that estimates whether generated images could be considered offensive or harmful.
            Please, refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for details.
        feature_extractor ([`CLIPFeatureExtractor`]):
            Model that extracts features from generated images to be used as inputs for the `safety_checker`.
    """

    tokenizer: CLIPTokenizer
    image_feature_extractor: CLIPFeatureExtractor
    text_encoder: CLIPTextModel
    image_encoder: CLIPVisionModel
    image_unet: UNet2DConditionModel
    text_unet: UNet2DConditionModel
    vae: AutoencoderKL
    scheduler: KarrasDiffusionSchedulers

    def __init__(
        self,
        tokenizer: CLIPTokenizer,
        image_feature_extractor: CLIPFeatureExtractor,
        text_encoder: CLIPTextModel,
        image_encoder: CLIPVisionModel,
        image_unet: UNet2DConditionModel,
        text_unet: UNet2DConditionModel,
        vae: AutoencoderKL,
        scheduler: KarrasDiffusionSchedulers,
    ):
        super().__init__()

        self.register_modules(
            tokenizer=tokenizer,
            image_feature_extractor=image_feature_extractor,
            text_encoder=text_encoder,
            image_encoder=image_encoder,
            image_unet=image_unet,
            text_unet=text_unet,
            vae=vae,
            scheduler=scheduler,
        )
        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)

    @torch.no_grad()
    def image_variation(
        self,
        image: Union[torch.FloatTensor, PIL.Image.Image],
        height: Optional[int] = None,
        width: Optional[int] = None,
        num_inference_steps: int = 50,
        guidance_scale: float = 7.5,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        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,
        output_type: Optional[str] = "pil",
        return_dict: bool = True,
        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
        callback_steps: int = 1,
    ):
        r"""
        Function invoked when calling the pipeline for generation.

        Args:
            image (`PIL.Image.Image`, `List[PIL.Image.Image]` or `torch.Tensor`):
                The image prompt or prompts to guide the image generation.
            height (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The width in pixels of the generated image.
            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.
            guidance_scale (`float`, *optional*, defaults to 7.5):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                if `guidance_scale` is less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`torch.Generator`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            latents (`torch.FloatTensor`, *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 will ge generated by sampling using the supplied random `generator`.
            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`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a
                plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.

        Examples:

        ```py
        >>> from diffusers import VersatileDiffusionPipeline
        >>> import torch
        >>> import requests
        >>> from io import BytesIO
        >>> from PIL import Image

        >>> # let's download an initial image
        >>> url = "https://huggingface.co/datasets/diffusers/images/resolve/main/benz.jpg"

        >>> response = requests.get(url)
        >>> image = Image.open(BytesIO(response.content)).convert("RGB")

        >>> pipe = VersatileDiffusionPipeline.from_pretrained(
        ...     "shi-labs/versatile-diffusion", torch_dtype=torch.float16
        ... )
        >>> pipe = pipe.to("cuda")

        >>> generator = torch.Generator(device="cuda").manual_seed(0)
        >>> image = pipe.image_variation(image, generator=generator).images[0]
        >>> image.save("./car_variation.png")
        ```

        Returns:
            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] if `return_dict` is True, otherwise a `tuple.
            When returning a tuple, the first element is a list with the generated images, and the second element is a
            list of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work"
            (nsfw) content, according to the `safety_checker`.
        """
        expected_components = inspect.signature(VersatileDiffusionImageVariationPipeline.__init__).parameters.keys()
        components = {name: component for name, component in self.components.items() if name in expected_components}
        return VersatileDiffusionImageVariationPipeline(**components)(
            image=image,
            height=height,
            width=width,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
            negative_prompt=negative_prompt,
            num_images_per_prompt=num_images_per_prompt,
            eta=eta,
            generator=generator,
            latents=latents,
            output_type=output_type,
            return_dict=return_dict,
            callback=callback,
            callback_steps=callback_steps,
        )

    @torch.no_grad()
    def text_to_image(
        self,
        prompt: Union[str, List[str]],
        height: Optional[int] = None,
        width: Optional[int] = None,
        num_inference_steps: int = 50,
        guidance_scale: float = 7.5,
        negative_prompt: Optional[Union[str, List[str]]] = None,
        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,
        output_type: Optional[str] = "pil",
        return_dict: bool = True,
        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
        callback_steps: int = 1,
    ):
        r"""
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`):
                The prompt or prompts to guide the image generation.
            height (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The width in pixels of the generated image.
            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.
            guidance_scale (`float`, *optional*, defaults to 7.5):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                if `guidance_scale` is less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`torch.Generator`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            latents (`torch.FloatTensor`, *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 will ge generated by sampling using the supplied random `generator`.
            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`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a
                plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.

        Examples:

        ```py
        >>> from diffusers import VersatileDiffusionPipeline
        >>> import torch

        >>> pipe = VersatileDiffusionPipeline.from_pretrained(
        ...     "shi-labs/versatile-diffusion", torch_dtype=torch.float16
        ... )
        >>> pipe = pipe.to("cuda")

        >>> generator = torch.Generator(device="cuda").manual_seed(0)
        >>> image = pipe.text_to_image("an astronaut riding on a horse on mars", generator=generator).images[0]
        >>> image.save("./astronaut.png")
        ```

        Returns:
            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] if `return_dict` is True, otherwise a `tuple.
            When returning a tuple, the first element is a list with the generated images, and the second element is a
            list of `bool`s denoting whether the corresponding generated image likely represents "not-safe-for-work"
            (nsfw) content, according to the `safety_checker`.
        """
        expected_components = inspect.signature(VersatileDiffusionTextToImagePipeline.__init__).parameters.keys()
        components = {name: component for name, component in self.components.items() if name in expected_components}
        temp_pipeline = VersatileDiffusionTextToImagePipeline(**components)
        output = temp_pipeline(
            prompt=prompt,
            height=height,
            width=width,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
            negative_prompt=negative_prompt,
            num_images_per_prompt=num_images_per_prompt,
            eta=eta,
            generator=generator,
            latents=latents,
            output_type=output_type,
            return_dict=return_dict,
            callback=callback,
            callback_steps=callback_steps,
        )
        # swap the attention blocks back to the original state
        temp_pipeline._swap_unet_attention_blocks()

        return output

    @torch.no_grad()
    def dual_guided(
        self,
        prompt: Union[PIL.Image.Image, List[PIL.Image.Image]],
        image: Union[str, List[str]],
        text_to_image_strength: float = 0.5,
        height: Optional[int] = None,
        width: Optional[int] = None,
        num_inference_steps: int = 50,
        guidance_scale: float = 7.5,
        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,
        output_type: Optional[str] = "pil",
        return_dict: bool = True,
        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
        callback_steps: int = 1,
    ):
        r"""
        Function invoked when calling the pipeline for generation.

        Args:
            prompt (`str` or `List[str]`):
                The prompt or prompts to guide the image generation.
            height (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The height in pixels of the generated image.
            width (`int`, *optional*, defaults to self.image_unet.config.sample_size * self.vae_scale_factor):
                The width in pixels of the generated image.
            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.
            guidance_scale (`float`, *optional*, defaults to 7.5):
                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                `guidance_scale` is defined as `w` of equation 2. of [Imagen
                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
                usually at the expense of lower image quality.
            negative_prompt (`str` or `List[str]`, *optional*):
                The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                if `guidance_scale` is less than `1`).
            num_images_per_prompt (`int`, *optional*, defaults to 1):
                The number of images to generate per prompt.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
                [`schedulers.DDIMScheduler`], will be ignored for others.
            generator (`torch.Generator`, *optional*):
                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                to make generation deterministic.
            latents (`torch.FloatTensor`, *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 will ge generated by sampling using the supplied random `generator`.
            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`.
            return_dict (`bool`, *optional*, defaults to `True`):
                Whether or not to return a [`~pipelines.stable_diffusion.StableDiffusionPipelineOutput`] instead of a
                plain tuple.
            callback (`Callable`, *optional*):
                A function that will be called every `callback_steps` steps during inference. The function will be
                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
            callback_steps (`int`, *optional*, defaults to 1):
                The frequency at which the `callback` function will be called. If not specified, the callback will be
                called at every step.

        Examples:

        ```py
        >>> from diffusers import VersatileDiffusionPipeline
        >>> import torch
        >>> import requests
        >>> from io import BytesIO
        >>> from PIL import Image

        >>> # let's download an initial image
        >>> url = "https://huggingface.co/datasets/diffusers/images/resolve/main/benz.jpg"

        >>> response = requests.get(url)
        >>> image = Image.open(BytesIO(response.content)).convert("RGB")
        >>> text = "a red car in the sun"

        >>> pipe = VersatileDiffusionPipeline.from_pretrained(
        ...     "shi-labs/versatile-diffusion", torch_dtype=torch.float16
        ... )
        >>> pipe = pipe.to("cuda")

        >>> generator = torch.Generator(device="cuda").manual_seed(0)
        >>> text_to_image_strength = 0.75

        >>> image = pipe.dual_guided(
        ...     prompt=text, image=image, text_to_image_strength=text_to_image_strength, generator=generator
        ... ).images[0]
        >>> image.save("./car_variation.png")
        ```

        Returns:
            [`~pipelines.stable_diffusion.ImagePipelineOutput`] or `tuple`:
            [`~pipelines.stable_diffusion.ImagePipelineOutput`] if `return_dict` is True, otherwise a `tuple. When
            returning a tuple, the first element is a list with the generated images.
        """

        expected_components = inspect.signature(VersatileDiffusionDualGuidedPipeline.__init__).parameters.keys()
        components = {name: component for name, component in self.components.items() if name in expected_components}
        temp_pipeline = VersatileDiffusionDualGuidedPipeline(**components)
        output = temp_pipeline(
            prompt=prompt,
            image=image,
            text_to_image_strength=text_to_image_strength,
            height=height,
            width=width,
            num_inference_steps=num_inference_steps,
            guidance_scale=guidance_scale,
            num_images_per_prompt=num_images_per_prompt,
            eta=eta,
            generator=generator,
            latents=latents,
            output_type=output_type,
            return_dict=return_dict,
            callback=callback,
            callback_steps=callback_steps,
        )
        temp_pipeline._revert_dual_attention()

        return output