| import os | |
| import torch | |
| import numpy as np | |
| from torchvision.utils import save_image | |
| import imageio | |
| from einops import rearrange | |
| def save_images(save_img_root, img_tensors): | |
| ''' | |
| tensors are normalized into (-1, 1) | |
| ''' | |
| os.makedirs(save_img_root, exist_ok=True) | |
| for i in range(img_tensors.shape[0]): | |
| save_image(img_tensors[i], f"{save_img_root}/{str(i).zfill(8)}.png", | |
| normalize=True, value_range=(-1, 1)) | |
| def save_video(save_video_path, img_tensors,fps=16): | |
| # os.makedirs(os.path.basename(save_video_path), exist_ok=True) | |
| img_tensors = (img_tensors / 2 + 0.5).clamp(0, 1) * 255 | |
| img_tensors = rearrange(img_tensors, 't c h w -> t h w c').contiguous() | |
| img_tensors = img_tensors.cpu().numpy().astype(np.uint8) | |
| imageio.mimwrite(save_video_path, img_tensors, fps=fps, quality=4, output_params=["-loglevel", "error"]) # Highest quality is 10, lowest is 0 |