Spyderball

It's barely possible without using DDP

I tried to extract several key points from the dwpose and perform affine transformation to obtain a matrix. Then, I processed each frame of the pose video and achieved better...

same problem

you can change pose2vid.py: video = pipe( ref_image_pil, pose_list, width, height, len(pose_list), args.steps, args.cfg, generator=generator, ).videos save_videos_grid( video, f"{save_dir}/{ref_name}_{pose_name}_{args.H}x{args.W}_{int(args.cfg)}_{time_str}_animate.mp4", n_rows=1, fps=src_fps if args.fps is None else args.fps, )