Tune-A-Video - Redshift

Model Description

• Base model: nitrosocke/redshift-diffusion
• Training prompt: a man is skiing.

Samples

Test prompt: (redshift style) [spider man/black widow/batman/hulk] is skiing.

Usage

Clone the github repo

git clone https://github.com/showlab/Tune-A-Video.git

Run inference code

from tuneavideo.pipelines.pipeline_tuneavideo import TuneAVideoPipeline
from tuneavideo.models.unet import UNet3DConditionModel
from tuneavideo.util import save_videos_grid
import torch

pretrained_model_path = "nitrosocke/redshift-diffusion"
unet_model_path = "Tune-A-Video-library/redshift-man-skiing"
unet = UNet3DConditionModel.from_pretrained(unet_model_path, subfolder='unet', torch_dtype=torch.float16).to('cuda')
pipe = TuneAVideoPipeline.from_pretrained(pretrained_model_path, unet=unet, torch_dtype=torch.float16).to("cuda")
pipe.enable_xformers_memory_efficient_attention()

prompt = "(redshift style) spider man is skiing"
video = pipe(prompt, video_length=8, height=512, width=512, num_inference_steps=50, guidance_scale=7.5).videos

save_videos_grid(video, f"./{prompt}.gif")

Related Papers:

• Tune-A-Video: One-Shot Tuning of Image Diffusion Models for Text-to-Video Generation
• Stable Diffusion: High-Resolution Image Synthesis with Latent Diffusion Models