Total runs:
87.8K
Run Growth:
-3.3K
Growth Rate:
-3.77%
Sponsored by
AdsCreator.com
alibaba-pai / EasyAnimateV5-12b-zh
Model's Last Updated: November 22 2024
Introduction of EasyAnimateV5-12b-zh
Model Details of EasyAnimateV5-12b-zh
EasyAnimate | 高分辨率长视频生成的端到端解决方案
😊 EasyAnimate是一个用于生成高分辨率和长视频的端到端解决方案。我们可以训练基于转换器的扩散生成器，训练用于处理长视频的VAE，以及预处理元数据。
😊 我们基于DIT，使用transformer进行作为扩散器进行视频与图片生成。
😊 Welcome!
English | 简体中文
目录
简介
EasyAnimate是一个基于transformer结构的pipeline，可用于生成AI图片与视频、训练Diffusion Transformer的基线模型与Lora模型，我们支持从已经训练好的EasyAnimate模型直接进行预测，生成不同分辨率，6秒左右、fps8的视频（EasyAnimateV5，1 ~ 49帧），也支持用户训练自己的基线模型与Lora模型，进行一定的风格变换。
我们会逐渐支持从不同平台快速启动，请参阅 快速启动 。
新特性：
- 更新到v5版本，最大支持1024x1024，49帧, 6s, 8fps视频生成，拓展模型规模到12B，应用MMDIT结构，支持不同输入的控制模型，支持中文与英文双语预测。[ 2024.11.08 ]
- 更新到v4版本，最大支持1024x1024，144帧, 6s, 24fps视频生成，支持文、图、视频生视频，单个模型可支持512到1280任意分辨率，支持中文与英文双语预测。[ 2024.08.15 ]
- 更新到v3版本，最大支持960x960，144帧，6s, 24fps视频生成，支持文与图生视频模型。[ 2024.07.01 ]
- ModelScope-Sora“数据导演”创意竞速——第三届Data-Juicer大模型数据挑战赛已经正式启动！其使用EasyAnimate作为基础模型，探究数据处理对于模型训练的作用。立即访问 竞赛官网 ，了解赛事详情。[ 2024.06.17 ]
- 更新到v2版本，最大支持768x768，144帧，6s, 24fps视频生成。[ 2024.05.26 ]
- 创建代码！现在支持 Windows 和 Linux。[ 2024.04.12 ]
功能概览：
我们的ui界面如下:
快速启动
1. 云使用: AliyunDSW/Docker
a. 通过阿里云 DSW
DSW 有免费 GPU 时间，用户可申请一次，申请后3个月内有效。
阿里云在 Freetier 提供免费GPU时间，获取并在阿里云PAI-DSW中使用，5分钟内即可启动EasyAnimate
b. 通过ComfyUI
我们的ComfyUI界面如下，具体查看
ComfyUI README
。
c. 通过docker
使用docker的情况下，请保证机器中已经正确安装显卡驱动与CUDA环境，然后以此执行以下命令：
# pull image
docker pull mybigpai-public-registry.cn-beijing.cr.aliyuncs.com/easycv/torch_cuda:easyanimate
# enter image
docker run -it -p 7860:7860 --network host --gpus all --security-opt seccomp:unconfined --shm-size 200g mybigpai-public-registry.cn-beijing.cr.aliyuncs.com/easycv/torch_cuda:easyanimate
# clone code
git clone https://github.com/aigc-apps/EasyAnimate.git
# enter EasyAnimate's dir
cd EasyAnimate
# download weights
mkdir models/Diffusion_Transformer
mkdir models/Motion_Module
mkdir models/Personalized_Model
# Please use the hugginface link or modelscope link to download the EasyAnimateV5 model.
# I2V models
# https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh-InP
# https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh-InP
# T2V models
# https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh
# https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh
2. 本地安装: 环境检查/下载/安装
a. 环境检查
我们已验证EasyAnimate可在以下环境中执行：
Windows 的详细信息：
- 操作系统 Windows 10
- python: python3.10 & python3.11
- pytorch: torch2.2.0
- CUDA: 11.8 & 12.1
- CUDNN: 8+
- GPU： Nvidia-3060 12G
Linux 的详细信息：
- 操作系统 Ubuntu 20.04, CentOS
- python: python3.10 & python3.11
- pytorch: torch2.2.0
- CUDA: 11.8 & 12.1
- CUDNN: 8+
- GPU：Nvidia-V100 16G & Nvidia-A10 24G & Nvidia-A100 40G & Nvidia-A100 80G
我们需要大约 60GB 的可用磁盘空间，请检查！
b. 权重放置
我们最好将 权重 按照指定路径进行放置：
EasyAnimateV5:
📦 models/
├── 📂 Diffusion_Transformer/
│ ├── 📂 EasyAnimateV5-12b-zh-InP/
│ └── 📂 EasyAnimateV5-12b-zh/
├── 📂 Personalized_Model/
│ └── your trained trainformer model / your trained lora model (for UI load)
视频作品
所展示的结果都是图生视频获得。
EasyAnimateV5-12b-zh-InP
Resolution-1024
Resolution-768
Resolution-512
EasyAnimateV5-12b-zh-Control
如何使用
1. 生成
a、运行python文件
- 步骤1：下载对应 权重 放入models文件夹。
- 步骤2：在predict_t2v.py文件中修改prompt、neg_prompt、guidance_scale和seed。
- 步骤3：运行predict_t2v.py文件，等待生成结果，结果保存在samples/easyanimate-videos文件夹中。
- 步骤4：如果想结合自己训练的其他backbone与Lora，则看情况修改predict_t2v.py中的predict_t2v.py和lora_path。
b、通过ui界面
- 步骤1：下载对应 权重 放入models文件夹。
- 步骤2：运行app.py文件，进入gradio页面。
- 步骤3：根据页面选择生成模型，填入prompt、neg_prompt、guidance_scale和seed等，点击生成，等待生成结果，结果保存在sample文件夹中。
c、通过comfyui
具体查看 ComfyUI README 。
d、显存节省方案
由于EasyAnimateV5的参数非常大，我们需要考虑显存节省方案，以节省显存适应消费级显卡。我们给每个预测文件都提供了GPU_memory_mode，可以在model_cpu_offload，model_cpu_offload_and_qfloat8，sequential_cpu_offload中进行选择。
- model_cpu_offload代表整个模型在使用后会进入cpu，可以节省部分显存。
- model_cpu_offload_and_qfloat8代表整个模型在使用后会进入cpu，并且对transformer模型进行了float8的量化，可以节省更多的显存。
- sequential_cpu_offload代表模型的每一层在使用后会进入cpu，速度较慢，节省大量显存。
qfloat8会降低模型的性能，但可以节省更多的显存。如果显存足够，推荐使用model_cpu_offload。
2. 模型训练
一个完整的EasyAnimate训练链路应该包括数据预处理、Video VAE训练、Video DiT训练。其中Video VAE训练是一个可选项，因为我们已经提供了训练好的Video VAE。
a.数据预处理
我们给出了一个简单的demo通过图片数据训练lora模型，详情可以查看 wiki 。
一个完整的长视频切分、清洗、描述的数据预处理链路可以参考video caption部分的 README 进行。
如果期望训练一个文生图视频的生成模型，您需要以这种格式排列数据集。
📦 project/
├── 📂 datasets/
│ ├── 📂 internal_datasets/
│ ├── 📂 train/
│ │ ├── 📄 00000001.mp4
│ │ ├── 📄 00000002.jpg
│ │ └── 📄 .....
│ └── 📄 json_of_internal_datasets.json
json_of_internal_datasets.json是一个标准的json文件。json中的file_path可以被设置为相对路径，如下所示：
[
{
"file_path": "train/00000001.mp4",
"text": "A group of young men in suits and sunglasses are walking down a city street.",
"type": "video"
},
{
"file_path": "train/00000002.jpg",
"text": "A group of young men in suits and sunglasses are walking down a city street.",
"type": "image"
},
.....
]
你也可以将路径设置为绝对路径：
[
{
"file_path": "/mnt/data/videos/00000001.mp4",
"text": "A group of young men in suits and sunglasses are walking down a city street.",
"type": "video"
},
{
"file_path": "/mnt/data/train/00000001.jpg",
"text": "A group of young men in suits and sunglasses are walking down a city street.",
"type": "image"
},
.....
]
b. Video VAE训练 （可选）
Video VAE训练是一个可选项，因为我们已经提供了训练好的Video VAE。如果想要进行训练，可以参考video vae部分的 README 进行。
c. Video DiT训练
如果数据预处理时，数据的格式为相对路径，则进入scripts/train.sh进行如下设置。
export DATASET_NAME="datasets/internal_datasets/"
export DATASET_META_NAME="datasets/internal_datasets/json_of_internal_datasets.json"
...
train_data_format="normal"
如果数据的格式为绝对路径，则进入scripts/train.sh进行如下设置。
export DATASET_NAME=""
export DATASET_META_NAME="/mnt/data/json_of_internal_datasets.json"
最后运行scripts/train.sh。
sh scripts/train.sh
关于一些参数的设置细节，可以查看 Readme Train 与 Readme Lora
(Obsolete) EasyAnimateV1:
如果你想训练EasyAnimateV1。请切换到git分支v1。模型地址
EasyAnimateV5:
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|---|---|---|---|---|---|
| EasyAnimateV5-12b-zh-InP | EasyAnimateV5 | 34 GB | 🤗Link | 😄Link | 官方的图生视频权重。支持多分辨率（512，768，1024）的视频预测，支持多分辨率（512，768，1024）的视频预测，以49帧、每秒8帧进行训练，支持中文与英文双语预测 |
| EasyAnimateV5-12b-zh-Control | EasyAnimateV5 | 34 GB | 🤗Link | 😄Link | 官方的视频控制权重，支持不同的控制条件，如Canny、Depth、Pose、MLSD等。支持多分辨率（512，768，1024）的视频预测，支持多分辨率（512，768，1024）的视频预测，以49帧、每秒8帧进行训练，支持中文与英文双语预测 |
| EasyAnimateV5-12b-zh | EasyAnimateV5 | 34 GB | 🤗Link | 😄Link | 官方的文生视频权重。可用于进行下游任务的fientune。支持多分辨率（512，768，1024）的视频预测，支持多分辨率（512，768，1024）的视频预测，以49帧、每秒8帧进行训练，支持中文与英文双语预测 |
(Obsolete) EasyAnimateV4:
(Obsolete) EasyAnimateV3:
| 名称 | 种类 | 存储空间 | 下载地址 | Hugging Face | 描述 |
|---|---|---|---|---|---|
| EasyAnimateV3-XL-2-InP-512x512.tar | EasyAnimateV3 | 18.2GB | Download | 🤗Link | 官方的512x512分辨率的图生视频权重。以144帧、每秒24帧进行训练 |
| EasyAnimateV3-XL-2-InP-768x768.tar | EasyAnimateV3 | 18.2GB | Download | 🤗Link | 官方的768x768分辨率的图生视频权重。以144帧、每秒24帧进行训练 |
| EasyAnimateV3-XL-2-InP-960x960.tar | EasyAnimateV3 | 18.2GB | Download | 🤗Link | 官方的960x960（720P）分辨率的图生视频权重。以144帧、每秒24帧进行训练 |
(Obsolete) EasyAnimateV2:
| 名称 | 种类 | 存储空间 | 下载地址 | Hugging Face | 描述 |
|---|---|---|---|---|---|
| EasyAnimateV2-XL-2-512x512.tar | EasyAnimateV2 | 16.2GB | Download | 🤗Link | 官方的512x512分辨率的重量。以144帧、每秒24帧进行训练 |
| EasyAnimateV2-XL-2-768x768.tar | EasyAnimateV2 | 16.2GB | Download | 🤗Link | 官方的768x768分辨率的重量。以144帧、每秒24帧进行训练 |
| easyanimatev2_minimalism_lora.safetensors | Lora of Pixart | 485.1MB | Download | - | 使用特定类型的图像进行lora训练的结果。图片可从这里 下载 . |
(Obsolete) EasyAnimateV1:
1、运动权重
| 名称 | 种类 | 存储空间 | 下载地址 | 描述 |
|---|---|---|---|---|
| easyanimate_v1_mm.safetensors | Motion Module | 4.1GB | download | Training with 80 frames and fps 12 |
2、其他权重
| 名称 | 种类 | 存储空间 | 下载地址 | 描述 |
|---|---|---|---|---|
| PixArt-XL-2-512x512.tar | Pixart | 11.4GB | download | Pixart-Alpha official weights |
| easyanimate_portrait.safetensors | Checkpoint of Pixart | 2.3GB | download | Training with internal portrait datasets |
| easyanimate_portrait_lora.safetensors | Lora of Pixart | 654.0MB | download | Training with internal portrait datasets |
未来计划
- 支持更大规模参数量的文视频生成模型。
联系我们
- 扫描下方二维码或搜索群号：77450006752 来加入钉钉群。
-
扫描下方二维码来加入微信群（如果二维码失效，可扫描最右边同学的微信，邀请您入群）
参考文献
- CogVideo: https://github.com/THUDM/CogVideo/
- magvit: https://github.com/google-research/magvit
- PixArt: https://github.com/PixArt-alpha/PixArt-alpha
- Open-Sora-Plan: https://github.com/PKU-YuanGroup/Open-Sora-Plan
- Open-Sora: https://github.com/hpcaitech/Open-Sora
- Animatediff: https://github.com/guoyww/AnimateDiff
- ComfyUI-EasyAnimateWrapper: https://github.com/kijai/ComfyUI-EasyAnimateWrapper
- HunYuan DiT: https://github.com/tencent/HunyuanDiT
许可证
本项目采用 Apache License (Version 2.0) .
Runs of alibaba-pai EasyAnimateV5-12b-zh on huggingface.co
12
Total runs
-3
24-hour runs
-7
3-day runs
-8
7-day runs
5
30-day runs
More Information About EasyAnimateV5-12b-zh huggingface.co Model
Url of EasyAnimateV5-12b-zh
EasyAnimateV5-12b-zh huggingface.co Url
Provider of EasyAnimateV5-12b-zh huggingface.co
Other API from alibaba-pai
Updated:February 26 2026
Total runs:
62.6K
Run Growth:
5.4K
Growth Rate:
8.63%
Total runs:
50.3K
Run Growth:
-31.1K
Growth Rate:
-61.83%
Total runs:
48.1K
Run Growth:
42.4K
Growth Rate:
88.19%
Total runs:
42.4K
Run Growth:
41.5K
Growth Rate:
98.01%
Total runs:
37.0K
Run Growth:
37.0K
Growth Rate:
100.00%
Total runs:
21.1K
Run Growth:
21.1K
Growth Rate:
100.00%
Total runs:
20.7K
Run Growth:
-2.7K
Growth Rate:
-13.18%
Total runs:
17.1K
Run Growth:
17.0K
Growth Rate:
99.05%
Total runs:
15.1K
Run Growth:
5.2K
Growth Rate:
34.45%
Total runs:
14.5K
Run Growth:
9.2K
Growth Rate:
63.48%
Total runs:
13.4K
Run Growth:
-118
Growth Rate:
-0.88%
Total runs:
12.8K
Run Growth:
8.6K
Growth Rate:
67.41%
Total runs:
12.0K
Run Growth:
11.2K
Growth Rate:
93.20%
Total runs:
11.7K
Run Growth:
32
Growth Rate:
0.27%
Total runs:
11.3K
Run Growth:
7.8K
Growth Rate:
69.52%
Total runs:
10.5K
Run Growth:
-917
Growth Rate:
-8.74%
Total runs:
7.5K
Run Growth:
353
Growth Rate:
4.68%
Total runs:
7.2K
Run Growth:
7.1K
Growth Rate:
98.19%
Total runs:
5.2K
Run Growth:
315
Growth Rate:
6.06%
Total runs:
4.9K
Run Growth:
3.4K
Growth Rate:
69.53%
Total runs:
4.1K
Run Growth:
-908
Growth Rate:
-22.22%
Total runs:
2.8K
Run Growth:
2.1K
Growth Rate:
75.37%
Total runs:
1.8K
Run Growth:
900
Growth Rate:
50.42%
Total runs:
1.8K
Run Growth:
-597
Growth Rate:
-33.71%
Total runs:
1.8K
Run Growth:
1.0K
Growth Rate:
57.08%
Total runs:
1.3K
Run Growth:
109
Growth Rate:
8.48%
Total runs:
1.2K
Run Growth:
1.1K
Growth Rate:
92.04%
Total runs:
1.0K
Run Growth:
622
Growth Rate:
60.10%
Total runs:
583
Run Growth:
348
Growth Rate:
59.69%
Total runs:
581
Run Growth:
304
Growth Rate:
52.32%
Total runs:
422
Run Growth:
189
Growth Rate:
44.79%
Total runs:
399
Run Growth:
191
Growth Rate:
47.87%
Total runs:
388
Run Growth:
4
Growth Rate:
1.03%
Total runs:
340
Run Growth:
10
Growth Rate:
2.94%
Total runs:
265
Run Growth:
15
Growth Rate:
5.66%
Total runs:
262
Run Growth:
198
Growth Rate:
75.57%
Total runs:
252
Run Growth:
-391
Growth Rate:
-155.16%
Total runs:
234
Run Growth:
-36
Growth Rate:
-15.38%
Total runs:
206
Run Growth:
126
Growth Rate:
61.17%
Total runs:
201
Run Growth:
130
Growth Rate:
64.68%
Total runs:
198
Run Growth:
139
Growth Rate:
70.20%
Total runs:
185
Run Growth:
-51
Growth Rate:
-27.57%
Total runs:
170
Run Growth:
-35
Growth Rate:
-20.59%
Total runs:
137
Run Growth:
-96
Growth Rate:
-70.07%
Total runs:
137
Run Growth:
7
Growth Rate:
5.11%
Total runs:
133
Run Growth:
-43
Growth Rate:
-32.33%
Total runs:
122
Run Growth:
-95
Growth Rate:
-77.87%
Total runs:
121
Run Growth:
46
Growth Rate:
38.02%
Total runs:
116
Run Growth:
72
Growth Rate:
62.07%
Total runs:
111
Run Growth:
25
Growth Rate:
22.52%
Total runs:
103
Run Growth:
3
Growth Rate:
2.91%
Total runs:
93
Run Growth:
80
Growth Rate:
86.02%
Total runs:
83
Run Growth:
-11
Growth Rate:
-13.25%
Total runs:
75
Run Growth:
-52
Growth Rate:
-69.33%
Total runs:
66
Run Growth:
-15
Growth Rate:
-22.73%
Total runs:
64
Run Growth:
26
Growth Rate:
40.63%
Total runs:
59
Run Growth:
12
Growth Rate:
20.34%
Total runs:
57
Run Growth:
18
Growth Rate:
31.58%
Total runs:
55
Run Growth:
28
Growth Rate:
50.91%
Total runs:
54
Run Growth:
24
Growth Rate:
44.44%
Total runs:
45
Run Growth:
26
Growth Rate:
57.78%
Total runs:
36
Run Growth:
32
Growth Rate:
88.89%
Total runs:
29
Run Growth:
-71
Growth Rate:
-244.83%
Total runs:
28
Run Growth:
8
Growth Rate:
28.57%
Total runs:
26
Run Growth:
2
Growth Rate:
7.69%
Total runs:
25
Run Growth:
-14
Growth Rate:
-56.00%
Total runs:
25
Run Growth:
14
Growth Rate:
56.00%
Total runs:
23
Run Growth:
14
Growth Rate:
60.87%
Total runs:
21
Run Growth:
14
Growth Rate:
66.67%
Total runs:
20
Run Growth:
4
Growth Rate:
20.00%
Total runs:
19
Run Growth:
3
Growth Rate:
15.79%
Total runs:
18
Run Growth:
6
Growth Rate:
33.33%
Total runs:
17
Run Growth:
0
Growth Rate:
0.00%
Total runs:
17
Run Growth:
-8
Growth Rate:
-47.06%
Total runs:
17
Run Growth:
-3
Growth Rate:
-17.65%
Total runs:
17
Run Growth:
1
Growth Rate:
5.88%
Total runs:
16
Run Growth:
-2
Growth Rate:
-12.50%
Total runs:
16
Run Growth:
-11
Growth Rate:
-68.75%
Total runs:
15
Run Growth:
1
Growth Rate:
6.67%
Total runs:
15
Run Growth:
8
Growth Rate:
53.33%
Total runs:
15
Run Growth:
3
Growth Rate:
20.00%
Total runs:
15
Run Growth:
1
Growth Rate:
6.67%
Total runs:
14
Run Growth:
8
Growth Rate:
57.14%
Total runs:
14
Run Growth:
7
Growth Rate:
50.00%
Total runs:
14
Run Growth:
8
Growth Rate:
57.14%
Total runs:
13
Run Growth:
9
Growth Rate:
69.23%
Total runs:
13
Run Growth:
7
Growth Rate:
53.85%
Total runs:
12
Run Growth:
1
Growth Rate:
8.33%
Total runs:
12
Run Growth:
7
Growth Rate:
58.33%
Total runs:
12
Run Growth:
0
Growth Rate:
0.00%
Total runs:
12
Run Growth:
1
Growth Rate:
8.33%
Total runs:
12
Run Growth:
2
Growth Rate:
16.67%
Total runs:
11
Run Growth:
4
Growth Rate:
36.36%
Total runs:
11
Run Growth:
9
Growth Rate:
81.82%
Total runs:
10
Run Growth:
1
Growth Rate:
10.00%
Total runs:
10
Run Growth:
-1
Growth Rate:
-10.00%
Total runs:
10
Run Growth:
4
Growth Rate:
40.00%
Total runs:
10
Run Growth:
4
Growth Rate:
40.00%
Total runs:
9
Run Growth:
1
Growth Rate:
11.11%