全球首款无限时长电影生成模型SkyReels-V2本地部署教程：视频时长无限制！_skyreels本地部署教程
一、简介
SkyReels-V2 模型集成了多模态大语言模型（MLLM）、多阶段预训练、强化学习以及创新的扩散强迫（Diffusion-forcing）框架，实现了在提示词遵循、视觉质量、运动动态以及视频时长等方面的全面突破。通过扩散强迫框架和 多阶段优化技术 ，首次实现了单镜头 30 秒、40 秒的流畅输出，并通过“ Extend ”无限延伸，彻底打破了时长枷锁。
二、环境部署
基础环境要求:
从算家云基础镜像开始创建：
三、SkyReels-V2安装
首先查看环境信息：
从github 克隆代码，然后 pip 安装依赖包。
# clone the repository.git clone https://github.com/SkyworkAI/SkyReels-V2cd SkyReels-V2pip install -r requirements.txt若遇到 flash_attn 一起在安装的问题：可以真的到Releases · Dao-AILab/flash-attention 下载它的编译好的 whl 包，一定要选择对应的版本，
wget https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.5cxx11abiFALSE-cp312-cp312-linux_x86_64.whlpip install flash_attn-2.7.4.post1+cu12torch2.5cxx11abiFALSE-cp312-cp312-linux_x86_64.whl然后在执行 pip install -r requirements.txt
四、下载模型文件
目前开源的模型如下：
从ModelScope 下载 Diffusion Forcing1.3B-540P 的模型：
git lfs installgit clone https://www.modelscope.cn/Skywork/SkyReels-V2-DF-1.3B-540P.git五、运行测试
下载完模型后，将下列代码写入：test.sh
model_id=Skywork/SkyReels-V2-DF-1.3B-540P# synchronous inferencepython generate_video_df.py \\ --model_id ${model_id} \\ --resolution 540P \\ --ar_step 0 \\ --base_num_frames 97 \\ --num_frames 257 \\ --overlap_history 17 \\ --prompt \"A graceful white swan with a curved neck and delicate feathers swimming in a serene lake at dawn, its reflection perfectly mirrored in the still water as mist rises from the surface, with the swan occasionally dipping its head into the water to feed.\" \\ --addnoise_condition 20 \\ --offload运行 sh test.sh 完成后结果保存在：root/SkyReels-V2/result/diffusion_forcing
- 若出现在问题：
Error opening output file: File name too long是由于导出视频的文件名太长所导致的，可以修改generate_video_df.py的代码 ：找到下列代码，将[:100]修改为[:10]
if local_rank == 0: current_time = time.strftime(\"%Y-%m-%d_%H-%M-%S\", time.localtime()) video_out_file = f\"{args.prompt[:10].replace(\'/\',\'\')}_{args.seed}_{current_time}.mp4\" output_path = os.path.join(save_dir, video_out_file) imageio.mimwrite(output_path, video_frames, fps=fps, quality=8, output_params=[\"-loglevel\", \"error\"])- 无法使用14B模型，若要使用需要至少一张大于60G显存的显卡
六、使用webui界面
通过 python app.py 运行webui
模型可以放在 SkyReels-V2/Skywork 文件夹下，这个镜像中的模型是使用 ln 创建的链接
# app.pyimport osimport gradio as grimport subprocessimport timeimport globimport torch.distributed as distdef get_model_list(base_dir=\"./Skywork/\"): \"\"\"扫描Skywork目录获取可用的模型列表\"\"\" try: # 确保目录存在 if not os.path.exists(base_dir): os.makedirs(base_dir) return [\"请选择模型\"] # 返回默认列表 # 扫描目录下的所有子目录 model_dirs = glob.glob(os.path.join(base_dir, \"*\")) model_ids = [] for dir_path in model_dirs: if os.path.isdir(dir_path): # 将目录路径转换为model_id格式 model_name = os.path.basename(dir_path) model_id = f\"{base_dir}/{model_name}\" model_ids.append(model_id) # 如果没有找到模型，返回默认列表 return model_ids if model_ids else [\"请选择模型\"] except Exception as e: print(f\"扫描模型目录出错: {str(e)}\") return [\"扫描模型目录出错\"] # 发生错误时返回默认列表def enhance_prompt(prompt): \"\"\"使用提示词增强器处理提示词\"\"\" try: cmd = [ \"python\", \"skyreels_v2_infer/pipelines/prompt_enhancer.py\", \"--prompt\", str(prompt) ] # 执行提示词增强 process = subprocess.Popen( cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1, universal_newlines=True ) output = [] enhanced_prompt = None for line in process.stdout: print(line, end=\'\') output.append(line) if \"Enhanced prompt:\" in line: enhanced_prompt = line.split(\"Enhanced prompt:\")[-1].strip() process.wait() if process.returncode != 0: raise RuntimeError(\"提示词增强失败\") return enhanced_prompt if enhanced_prompt else prompt except Exception as e: print(f\"提示词增强失败: {str(e)}\") return promptdef run_video_generation( prompt, model_id=\"选择模型\", resolution=\"540P\", ar_step=4, base_num_frames=97, num_frames=257, overlap_history=17, addnoise_condition=20, use_usp=True, offload=True, seed=42, gpu_count=2, image=None, inference_steps=50, fps=24, shift=8.0, guidance_scale=6.0, causal_block_size=5, outdir=\"./video_out\", prompt_enhancer=False, # 添加prompt_enhancer参数 teacache=False # 添加teacache参数): \"\"\"执行视频生成命令\"\"\" # 记录开始时间 start_time = time.time() # 确保输出目录存在并获取绝对路径 outdir = os.path.abspath(outdir) os.makedirs(outdir, exist_ok=True) # 根据条件处理提示词增强 if prompt_enhancer: try: enhanced = enhance_prompt(prompt) if enhanced != prompt: print(f\"原始提示词: {prompt}\") print(f\"增强后的提示词: {enhanced}\") prompt = enhanced except Exception as e: print(f\"提示词增强失败，将使用原始提示词: {str(e)}\") # 自动关闭提示词增强功能 prompt_enhancer = False # 构建基础命令参数，确保参数顺序正确 base_args = [ \"--prompt\", str(prompt), \"--model_id\", str(model_id), \"--resolution\", str(resolution), \"--ar_step\", str(ar_step), \"--base_num_frames\", str(base_num_frames), \"--num_frames\", str(num_frames), \"--overlap_history\", str(overlap_history), \"--addnoise_condition\", str(addnoise_condition), \"--inference_steps\", str(inference_steps), \"--fps\", str(fps), \"--shift\", str(shift), \"--guidance_scale\", str(guidance_scale), \"--causal_block_size\", str(causal_block_size), \"--outdir\", str(outdir) # 确保outdir在最后 ] if teacache: base_args.append(\"--teacache\") if image: base_args.extend([\"--image\", str(image)]) if offload: base_args.append(\"--offload\") if seed != -1: base_args.append(\"--seed\") base_args.append(str(seed)) if prompt_enhancer: base_args.append(\"--prompt_enhancer\") # 根据是否使用USP构建完整命令 if use_usp: cmd = [ \"torchrun\", f\"--nproc_per_node={gpu_count}\", f\"generate_video_df.py\" ] + base_args + [\"--use_usp\"] else: cmd = [\"python\", \"generate_video_df.py\"] + base_args # 打印执行命令 print(\"执行命令:\", \" \".join(cmd)) # 确保输出目录存在并获取绝对路径 outdir = os.path.abspath(outdir) os.makedirs(outdir, exist_ok=True) # 执行命令并实时显示输出 try: process = subprocess.Popen( cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, bufsize=1, universal_newlines=True ) output = [] video_path = None # 实时读取并显示输出 for line in process.stdout: print(line, end=\'\') # 实时显示 output.append(line) # 保存输出 # 检查是否包含视频路径信息并转换为绝对路径 if \"Saved video to\" in line: video_path = line.split(\"Saved video to\")[-1].strip() video_path = os.path.abspath(video_path) # 确保是绝对路径 print(f\"视频保存路径: {video_path}\") process.wait() if process.returncode != 0: raise RuntimeError(\"生成视频失败\") # 计算并添加耗时信息 end_time = time.time() elapsed_time = end_time - start_time time_info = f\"\\n总耗时: {elapsed_time:.2f} 秒\" output.append(time_info) return \"\".join(output), video_path except Exception as e: raise RuntimeError(f\"执行命令失败: {str(e)}\")def get_available_gpus(): \"\"\"获取可用的GPU数量和设备ID\"\"\" try: import torch gpu_count = torch.cuda.device_count() if gpu_count == 0: return 0, [] return gpu_count, list(range(gpu_count)) except: return 0, []def scan_video_directory(directory=\"./video_out\"): \"\"\"扫描视频目录并返回所有视频文件\"\"\" if not os.path.exists(directory): return [] video_files = [] for ext in [\'.mp4\', \'.avi\', \'.mov\', \'.mkv\']: files = glob.glob(os.path.join(directory, f\'*{ext}\')) # 返回(显示名称, 文件路径)的元组 for f in files: basename = os.path.basename(f) video_files.append((basename, f)) # 按修改时间排序 return sorted(video_files, key=lambda x: os.path.getmtime(x[1]), reverse=True)# 在Gradio界面中使用时：video_list = gr.Dropdown( label=\"选择视频播放\", choices=scan_video_directory(), value=None, interactive=True)def update_video_list(): \"\"\"刷新视频列表\"\"\" return gr.Dropdown(choices=scan_video_directory())def on_video_select(video_path): \"\"\"当选择视频时\"\"\" if video_path: # video_path现在是完整路径 return video_path return Nonedef on_generate_click(*args): \"\"\"生成视频并更新列表\"\"\" try: # 获取参数 prompt, model_id, resolution, ar_step, base_num_frames, num_frames, \\ overlap_history, addnoise_condition, use_usp, offload, seed, gpu_count, \\ image, inference_steps, fps, shift, guidance_scale, \\ causal_block_size, outdir, prompt_enhancer, teacache = args # 添加teacache # 如果ar_step为0(同步模式)，强制设置causal_block_size为1 if ar_step == 0: causal_block_size = 1 # 如果同时启用了use_usp和prompt_enhancer，先进行提示词增强 if use_usp and prompt_enhancer: try: enhanced = enhance_prompt(prompt) if enhanced != prompt: print(f\"原始提示词: {prompt}\") print(f\"增强后的提示词: {enhanced}\") prompt = enhanced except Exception as e: print(f\"提示词增强失败，将使用原始提示词: {str(e)}\") # 禁用prompt_enhancer参数，因为提示词已经增强过了 prompt_enhancer = False # 执行视频生成 output, video_path = run_video_generation( prompt, model_id, resolution, ar_step, base_num_frames, num_frames, overlap_history, addnoise_condition, use_usp, offload, seed, gpu_count, image, inference_steps, fps, shift, guidance_scale, causal_block_size, outdir, prompt_enhancer,teacache ) # 更新视频列表 new_choices = scan_video_directory() if video_path and os.path.exists(video_path): return output, video_path, gr.Dropdown(choices=new_choices, value=video_path) return output, None, gr.Dropdown(choices=new_choices) except Exception as e: error_msg = f\"生成失败: {str(e)}\" return error_msg, None, gr.Dropdown(choices=scan_video_directory())# 创建Gradio界面with gr.Blocks(title=\"SkyReels-V2 视频生成器\") as demo: gr.Markdown(\"# SkyReels-V2 视频生成器\") with gr.Row(): with gr.Column(): prompt = gr.Textbox( label=\"提示词 (prompt)\", lines=3, value=\"A graceful white swan with a curved neck and delicate feathers swimming in a serene lake at dawn\" ) # 图片输入 image = gr.Image( label=\"输入图片 (仅用于图像到视频生成)\", type=\"filepath\", visible=True ) # In the Gradio interface definition, update the model_id dropdown: model_id = gr.Dropdown( choices=get_model_list(), label=\"模型ID (model_id)\", value=\"选择模型\",# Changed to match the available model allow_custom_value=True ) resolution = gr.Dropdown( choices=[\"540P\", \"720P\"], label=\"分辨率 (resolution)\", value=\"540P\" ) with gr.Row(): num_frames = gr.Number( label=\"生成帧数 (num_frames)\", value=97, precision=0 ) base_num_frames = gr.Number( label=\"基础帧数 (base_num_frames)\", value=97, precision=0 ) with gr.Row(): inference_steps = gr.Number( label=\"推理步骤数 (inference_steps)\", value=50, precision=0 ) fps = gr.Number( label=\"输出视频帧率 (fps)\", value=24, precision=0 ) with gr.Row(): shift = gr.Number( label=\"流匹配参数 (shift)\", value=8.0, info=\"T2V模型推荐8.0，I2V模型推荐5.0\" ) guidance_scale = gr.Number( label=\"文本匹配强度 (guidance_scale)\", value=6.0, info=\"T2V模型推荐6.0，I2V模型推荐5.0\" ) with gr.Row(): ar_step = gr.Number( label=\"AR步数 (ar_step)\", value=4, info=\"0表示同步模式\" ) causal_block_size = gr.Number( label=\"因果块大小 (causal_block_size)\", value=4, info=\"异步推理时推荐使用(ar_step>0)\" ) with gr.Row(): overlap_history = gr.Number( label=\"重叠历史帧数 (overlap_history)\", value=17 ) addnoise_condition = gr.Number( label=\"噪声条件 (addnoise_condition)\", value=20 ) with gr.Row(): use_usp = gr.Checkbox( label=\"启用use_usp(使用多GPU并行推理)\", value=True ) offload = gr.Checkbox( label=\"启用offload(使用显存优化)\", value=True ) prompt_enhancer = gr.Checkbox( label=\"启用prompt_enhancer(提示词扩展)\", value=True, ) teacache = gr.Checkbox( label=\"启用teacache(使用缓存加速)\", value=False, ) with gr.Row(): seed = gr.Number( label=\"随机种子 (-1为随机)\", value=42 ) gpu_count = gr.Number( label=\"使用GPU数量\", value=2, precision=0 ) outdir = gr.Textbox( label=\"输出目录 (outdir)\", value=\"./video_out\" ) # 添加说明 gr.Markdown(\"\"\" ### 使用说明 1. 单GPU模式：取消勾选\"启用USP\"选项 2. 多GPU模式： - 勾选\"启用USP\"选项 - 设置要使用的GPU数量 - 需要固定的随机种子 - 多GPU无法使用teacache 3. 无法使用14B模型，若要使用需要单张显存大于60G ### 更多信息 访问 [GitHub仓库](https://github.com/SkyworkAI/SkyReels-V2) 获取详细文档和更新。 \"\"\") # 生成按钮和输出信息 generate_btn = gr.Button(\"生成视频\") output_text = gr.Textbox(label=\"输出信息\", visible=True, lines=3) # 视频预览部分 gr.Markdown(\"### 视频预览\") with gr.Row(): with gr.Column(scale=4): video_player = gr.Video( label=\"视频播放\", format=\"mp4\", autoplay=False, # 自动播放选中的视频 interactive=True # 允许用户控制播放 ) with gr.Column(scale=1): video_list = gr.Dropdown( label=\"选择视频\", choices=scan_video_directory(), value=None, interactive=True ) refresh_btn = gr.Button(\"🔄 刷新视频列表\") # 更新事件绑定 generate_btn.click( on_generate_click, inputs=[prompt, model_id, resolution, ar_step, base_num_frames, num_frames, overlap_history, addnoise_condition, use_usp, offload, seed, gpu_count, image, inference_steps, fps, shift, guidance_scale, causal_block_size, outdir, prompt_enhancer, teacache], # 添加teacache outputs=[output_text, video_player, video_list] ) video_list.change( on_video_select, inputs=[video_list], outputs=[video_player] ) refresh_btn.click( update_video_list, outputs=[video_list] )if __name__ == \"__main__\": try: # 设置CUDA架构列表 os.environ[\"TORCH_CUDA_ARCH_LIST\"] = \"8.6;8.9\" # 30系和40系的架构 # 设置CUDA内存分配策略 os.environ[\"PYTORCH_CUDA_ALLOC_CONF\"] = \"expandable_segments:True\" # 检测GPU gpu_count, gpu_ids = get_available_gpus() # 设置环境变量以消除tokenizers警告 os.environ[\"TOKENIZERS_PARALLELISM\"] = \"false\" if gpu_count == 0: print(\"警告：未检测到可用的GPU，系统将使用CPU运行（不推荐）\") else: print(f\"检测到{gpu_count}个GPU {gpu_ids}\") os.environ[\"CUDA_VISIBLE_DEVICES\"] = \",\".join(map(str, gpu_ids)) # 启动Gradio界面，移除不支持的参数 demo.launch( server_name=\"0.0.0.0\", server_port=8080, share=False, ) finally: # 确保在程序退出时销毁进程组 if dist.is_initialized(): dist.destroy_process_group()
