Redshift Man Skiing

🚀 Tune-A-Video - Redshift

Tune-A-Video - Redshift 是一個基於特定基礎模型訓練的項目，可用於文本到視頻的生成，通過特定的訓練提示，能生成如人物滑雪等風格的視頻。

🚀 快速開始

克隆倉庫

首先，你需要克隆 github 倉庫。

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

運行推理代碼

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")

✨ 主要特性

• 基於 nitrosocke/redshift-diffusion 基礎模型進行訓練。
• 通過特定的訓練提示 “a man is skiing” 進行訓練，可生成具有特定風格的視頻。

📦 安裝指南

克隆項目所需的倉庫：

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

💻 使用示例

基礎用法

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")

📚 詳細文檔

模型描述

示例展示

此為訓練示例。

測試提示：(redshift style) [spider man/black widow/batman/hulk] is skiing。

相關論文

• 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

📄 許可證

本項目使用 creativeml-openrail-m 許可證。