File size: 7,720 Bytes
1a2c8b5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from transformers import CLIPTextModel, CLIPTokenizer, logging
from diffusers import AutoencoderKL, UNet2DConditionModel, DDIMScheduler
# suppress partial model loading warning
logging.set_verbosity_error()

import os
from tqdm import tqdm, trange
import torch
import torch.nn as nn
import argparse
from torchvision.io import write_video
from pathlib import Path
from util import *
import torchvision.transforms as T


def get_timesteps(scheduler, num_inference_steps, strength, device):
    # get the original timestep using init_timestep
    init_timestep = min(int(num_inference_steps * strength), num_inference_steps)

    t_start = max(num_inference_steps - init_timestep, 0)
    timesteps = scheduler.timesteps[t_start:]

    return timesteps, num_inference_steps - t_start
    
@torch.no_grad()
def decode_latents(pipe, latents):
    decoded = []
    batch_size = 8
    for b in range(0, latents.shape[0], batch_size):
            latents_batch = 1 / 0.18215 * latents[b:b + batch_size]
            imgs = pipe.vae.decode(latents_batch).sample
            imgs = (imgs / 2 + 0.5).clamp(0, 1)
            decoded.append(imgs)
    return torch.cat(decoded)

@torch.no_grad()
def ddim_inversion(pipe, cond, latent_frames,  batch_size, save_latents=True, timesteps_to_save=None):
    
    timesteps = reversed(pipe.scheduler.timesteps)
    timesteps_to_save = timesteps_to_save if timesteps_to_save is not None else timesteps
    for i, t in enumerate(tqdm(timesteps)):
        for b in range(0, latent_frames.shape[0], batch_size):
            x_batch = latent_frames[b:b + batch_size]
            model_input = x_batch
            cond_batch = cond.repeat(x_batch.shape[0], 1, 1)
            #remove comment from commented block to support controlnet
            # if self.sd_version == 'depth':
            #     depth_maps = torch.cat([self.depth_maps[b: b + batch_size]])
            #     model_input = torch.cat([x_batch, depth_maps],dim=1)

            alpha_prod_t = pipe.scheduler.alphas_cumprod[t]
            alpha_prod_t_prev = (
                pipe.scheduler.alphas_cumprod[timesteps[i - 1]]
                if i > 0 else pipe.scheduler.final_alpha_cumprod
            )

            mu = alpha_prod_t ** 0.5
            mu_prev = alpha_prod_t_prev ** 0.5
            sigma = (1 - alpha_prod_t) ** 0.5
            sigma_prev = (1 - alpha_prod_t_prev) ** 0.5

            
            #remove line below and replace with commented block to support controlnet
            eps = pipe.unet(model_input, t, encoder_hidden_states=cond_batch).sample
            # if self.sd_version != 'ControlNet':
            #     eps = pipe.unet(model_input, t, encoder_hidden_states=cond_batch).sample
            # else:
            #     eps = self.controlnet_pred(x_batch, t, cond_batch, torch.cat([self.canny_cond[b: b + batch_size]]))
            
            pred_x0 = (x_batch - sigma_prev * eps) / mu_prev
            latent_frames[b:b + batch_size] = mu * pred_x0 + sigma * eps

    #     if save_latents and t in timesteps_to_save:
    #         torch.save(latent_frames, os.path.join(save_path, 'latents', f'noisy_latents_{t}.pt'))
    # torch.save(latent_frames, os.path.join(save_path, 'latents', f'noisy_latents_{t}.pt'))
    return latent_frames    
    
@torch.no_grad()
def ddim_sample(pipe, x, cond, batch_size):
    timesteps = pipe.scheduler.timesteps
    for i, t in enumerate(tqdm(timesteps)):
        for b in range(0, x.shape[0], batch_size):
            x_batch = x[b:b + batch_size]
            model_input = x_batch
            cond_batch = cond.repeat(x_batch.shape[0], 1, 1)
            
            #remove comment from commented block to support controlnet
            # if self.sd_version == 'depth':
            #     depth_maps = torch.cat([self.depth_maps[b: b + batch_size]])
            #     model_input = torch.cat([x_batch, depth_maps],dim=1)

            alpha_prod_t = pipe.scheduler.alphas_cumprod[t]
            alpha_prod_t_prev = (
                pipe.scheduler.alphas_cumprod[timesteps[i + 1]]
                if i < len(timesteps) - 1
                else pipe.scheduler.final_alpha_cumprod
            )
            mu = alpha_prod_t ** 0.5
            sigma = (1 - alpha_prod_t) ** 0.5
            mu_prev = alpha_prod_t_prev ** 0.5
            sigma_prev = (1 - alpha_prod_t_prev) ** 0.5

            #remove line below and replace with commented block to support controlnet
            eps = pipe.unet(model_input, t, encoder_hidden_states=cond_batch).sample
            # if self.sd_version != 'ControlNet':
            #     eps = pipe.unet(model_input, t, encoder_hidden_states=cond_batch).sample
            # else:
            #     eps = self.controlnet_pred(x_batch, t, cond_batch, torch.cat([self.canny_cond[b: b + batch_size]]))

            pred_x0 = (x_batch - sigma * eps) / mu
            x[b:b + batch_size] = mu_prev * pred_x0 + sigma_prev * eps
    return x


@torch.no_grad()
def get_text_embeds(pipe, prompt, negative_prompt, batch_size=1, device="cuda"):
    # Tokenize text and get embeddings
    text_input = pipe.tokenizer(prompt, padding='max_length', max_length=pipe.tokenizer.model_max_length,
                                truncation=True, return_tensors='pt')
    text_embeddings = pipe.text_encoder(text_input.input_ids.to(pipe.device))[0]

    # Do the same for unconditional embeddings
    uncond_input = pipe.tokenizer(negative_prompt, padding='max_length', max_length=pipe.tokenizer.model_max_length,
                                  return_tensors='pt')

    uncond_embeddings = pipe.text_encoder(uncond_input.input_ids.to(pipe.device))[0]

    # Cat for final embeddings
    text_embeddings = torch.cat([uncond_embeddings] * batch_size + [text_embeddings] * batch_size)
    return text_embeddings

@torch.no_grad()
def extract_latents(pipe,
                    num_steps,
                    latent_frames,
                    batch_size,
                    timesteps_to_save,
                    inversion_prompt=''):
    pipe.scheduler.set_timesteps(num_steps)
    cond = get_text_embeds(pipe, inversion_prompt, "", device=pipe.device)[1].unsqueeze(0)
    # latent_frames = self.latents

    inverted_latents = ddim_inversion(pipe, cond,
                                latent_frames,
                                batch_size=batch_size,
                                save_latents=False,
                                timesteps_to_save=timesteps_to_save)
    
    # latent_reconstruction = ddim_sample(pipe, inverted_latents, cond, batch_size=batch_size)

#     rgb_reconstruction = decode_latents(pipe, latent_reconstruction)

#     return rgb_reconstruction
    return inverted_latents
    
@torch.no_grad()
def encode_imgs(pipe, imgs, batch_size=10, deterministic=True):
    imgs = 2 * imgs - 1
    latents = []
    for i in range(0, len(imgs), batch_size):
        posterior = pipe.vae.encode(imgs[i:i + batch_size]).latent_dist
        latent = posterior.mean if deterministic else posterior.sample()
        latents.append(latent * 0.18215)
    latents = torch.cat(latents)
    return latents
    
def get_data(pipe, frames, n_frames):
    """
    converts frames to tensors, saves to device and encodes to obtain latents
    """
    frames = frames[:n_frames]
    if frames[0].size[0] == frames[0].size[1]:
        frames = [frame.convert("RGB").resize((512, 512), resample=Image.Resampling.LANCZOS) for frame in frames]
    stacked_tensor_frames = torch.stack([T.ToTensor()(frame) for frame in frames]).to(torch.float16).to(pipe.device)
    # encode to latents
    latents = encode_imgs(pipe, stacked_tensor_frames, deterministic=True).to(torch.float16).to(pipe.device)
    return stacked_tensor_frames, latents