File size: 8,061 Bytes
62c110b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
import inspect
from typing import List, Optional, Tuple, Union

import numpy as np
import PIL.Image
import torch
import torch.utils.checkpoint

from ...models import UNet2DModel, VQModel
from ...schedulers import (
    DDIMScheduler,
    DPMSolverMultistepScheduler,
    EulerAncestralDiscreteScheduler,
    EulerDiscreteScheduler,
    LMSDiscreteScheduler,
    PNDMScheduler,
)
from ...utils import PIL_INTERPOLATION
from ...utils.torch_utils import randn_tensor
from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput


def preprocess(image):
    w, h = image.size
    w, h = (x - x % 32 for x in (w, h))  # resize to integer multiple of 32
    image = image.resize((w, h), resample=PIL_INTERPOLATION["lanczos"])
    image = np.array(image).astype(np.float32) / 255.0
    image = image[None].transpose(0, 3, 1, 2)
    image = torch.from_numpy(image)
    return 2.0 * image - 1.0


class LDMSuperResolutionPipeline(DiffusionPipeline):
    r"""
    A pipeline for image super-resolution using latent diffusion.

    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.).

    Parameters:
        vqvae ([`VQModel`]):
            Vector-quantized (VQ) model to encode and decode images to and from latent representations.
        unet ([`UNet2DModel`]):
            A `UNet2DModel` to denoise the encoded image.
        scheduler ([`SchedulerMixin`]):
            A scheduler to be used in combination with `unet` to denoise the encoded image latens. Can be one of
            [`DDIMScheduler`], [`LMSDiscreteScheduler`], [`EulerDiscreteScheduler`],
            [`EulerAncestralDiscreteScheduler`], [`DPMSolverMultistepScheduler`], or [`PNDMScheduler`].
    """

    def __init__(
        self,
        vqvae: VQModel,
        unet: UNet2DModel,
        scheduler: Union[
            DDIMScheduler,
            PNDMScheduler,
            LMSDiscreteScheduler,
            EulerDiscreteScheduler,
            EulerAncestralDiscreteScheduler,
            DPMSolverMultistepScheduler,
        ],
    ):
        super().__init__()
        self.register_modules(vqvae=vqvae, unet=unet, scheduler=scheduler)

    @torch.no_grad()
    def __call__(
        self,
        image: Union[torch.Tensor, PIL.Image.Image] = None,
        batch_size: Optional[int] = 1,
        num_inference_steps: Optional[int] = 100,
        eta: Optional[float] = 0.0,
        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
        output_type: Optional[str] = "pil",
        return_dict: bool = True,
    ) -> Union[Tuple, ImagePipelineOutput]:
        r"""
        The call function to the pipeline for generation.

        Args:
            image (`torch.Tensor` or `PIL.Image.Image`):
                `Image` or tensor representing an image batch to be used as the starting point for the process.
            batch_size (`int`, *optional*, defaults to 1):
                Number of images to generate.
            num_inference_steps (`int`, *optional*, defaults to 100):
                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                expense of slower inference.
            eta (`float`, *optional*, defaults to 0.0):
                Corresponds to parameter eta (η) from the [DDIM](https://arxiv.org/abs/2010.02502) paper. Only applies
                to the [`~schedulers.DDIMScheduler`], and is ignored in other schedulers.
            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                generation deterministic.
            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`):
                Whether or not to return a [`ImagePipelineOutput`] instead of a plain tuple.

        Example:

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

        >>> # load model and scheduler
        >>> pipeline = LDMSuperResolutionPipeline.from_pretrained("CompVis/ldm-super-resolution-4x-openimages")
        >>> pipeline = pipeline.to("cuda")

        >>> # let's download an  image
        >>> url = (
        ...     "https://user-images.githubusercontent.com/38061659/199705896-b48e17b8-b231-47cd-a270-4ffa5a93fa3e.png"
        ... )
        >>> response = requests.get(url)
        >>> low_res_img = Image.open(BytesIO(response.content)).convert("RGB")
        >>> low_res_img = low_res_img.resize((128, 128))

        >>> # run pipeline in inference (sample random noise and denoise)
        >>> upscaled_image = pipeline(low_res_img, num_inference_steps=100, eta=1).images[0]
        >>> # save image
        >>> upscaled_image.save("ldm_generated_image.png")
        ```

        Returns:
            [`~pipelines.ImagePipelineOutput`] or `tuple`:
                If `return_dict` is `True`, [`~pipelines.ImagePipelineOutput`] is returned, otherwise a `tuple` is
                returned where the first element is a list with the generated images
        """
        if isinstance(image, PIL.Image.Image):
            batch_size = 1
        elif isinstance(image, torch.Tensor):
            batch_size = image.shape[0]
        else:
            raise ValueError(f"`image` has to be of type `PIL.Image.Image` or `torch.Tensor` but is {type(image)}")

        if isinstance(image, PIL.Image.Image):
            image = preprocess(image)

        height, width = image.shape[-2:]

        # in_channels should be 6: 3 for latents, 3 for low resolution image
        latents_shape = (batch_size, self.unet.config.in_channels // 2, height, width)
        latents_dtype = next(self.unet.parameters()).dtype

        latents = randn_tensor(latents_shape, generator=generator, device=self.device, dtype=latents_dtype)

        image = image.to(device=self.device, dtype=latents_dtype)

        # set timesteps and move to the correct device
        self.scheduler.set_timesteps(num_inference_steps, device=self.device)
        timesteps_tensor = self.scheduler.timesteps

        # scale the initial noise by the standard deviation required by the scheduler
        latents = latents * self.scheduler.init_noise_sigma

        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature.
        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
        # and should be between [0, 1]
        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
        extra_kwargs = {}
        if accepts_eta:
            extra_kwargs["eta"] = eta

        for t in self.progress_bar(timesteps_tensor):
            # concat latents and low resolution image in the channel dimension.
            latents_input = torch.cat([latents, image], dim=1)
            latents_input = self.scheduler.scale_model_input(latents_input, t)
            # predict the noise residual
            noise_pred = self.unet(latents_input, t).sample
            # compute the previous noisy sample x_t -> x_t-1
            latents = self.scheduler.step(noise_pred, t, latents, **extra_kwargs).prev_sample

        # decode the image latents with the VQVAE
        image = self.vqvae.decode(latents).sample
        image = torch.clamp(image, -1.0, 1.0)
        image = image / 2 + 0.5
        image = image.cpu().permute(0, 2, 3, 1).numpy()

        if output_type == "pil":
            image = self.numpy_to_pil(image)

        if not return_dict:
            return (image,)

        return ImagePipelineOutput(images=image)