bubbliiiing
commited on
Commit
•
5dfb363
1
Parent(s):
f625293
Update Readme
Browse files- README.md +88 -32
- README_en.md +93 -18
README.md
CHANGED
@@ -30,21 +30,6 @@ tasks:
|
|
30 |
#- vllm
|
31 |
---
|
32 |
|
33 |
-
# EasyAnimate | 高分辨率长视频生成的端到端解决方案
|
34 |
-
😊 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。
|
35 |
-
|
36 |
-
😊 我们基于DIT,使用transformer进行作为扩散器进行视频与图片生成。
|
37 |
-
|
38 |
-
😊 Welcome!
|
39 |
-
|
40 |
-
[![Arxiv Page](https://img.shields.io/badge/Arxiv-Page-red)](https://arxiv.org/abs/2405.18991)
|
41 |
-
[![Project Page](https://img.shields.io/badge/Project-Website-green)](https://easyanimate.github.io/)
|
42 |
-
[![Modelscope Studio](https://img.shields.io/badge/Modelscope-Studio-blue)](https://modelscope.cn/studios/PAI/EasyAnimate/summary)
|
43 |
-
[![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-yellow)](https://huggingface.co/spaces/alibaba-pai/EasyAnimate)
|
44 |
-
[![Discord Page](https://img.shields.io/badge/Discord-Page-blue)](https://discord.gg/UzkpB4Bn)
|
45 |
-
|
46 |
-
[English](./README.md) | 简体中文
|
47 |
-
|
48 |
# 目录
|
49 |
- [目录](#目录)
|
50 |
- [简介](#简介)
|
@@ -143,6 +128,39 @@ Linux 的详细信息:
|
|
143 |
|
144 |
我们需要大约 60GB 的可用磁盘空间,请检查!
|
145 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
146 |
#### b. 权重放置
|
147 |
我们最好将[权重](#model-zoo)按照指定路径进行放置:
|
148 |
|
@@ -161,8 +179,7 @@ EasyAnimateV5:
|
|
161 |
|
162 |
### EasyAnimateV5-12b-zh-InP
|
163 |
|
164 |
-
|
165 |
-
|
166 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
167 |
<tr>
|
168 |
<td>
|
@@ -181,8 +198,6 @@ Resolution-1024
|
|
181 |
</table>
|
182 |
|
183 |
|
184 |
-
Resolution-768
|
185 |
-
|
186 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
187 |
<tr>
|
188 |
<td>
|
@@ -200,8 +215,6 @@ Resolution-768
|
|
200 |
</tr>
|
201 |
</table>
|
202 |
|
203 |
-
Resolution-512
|
204 |
-
|
205 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
206 |
<tr>
|
207 |
<td>
|
@@ -219,6 +232,41 @@ Resolution-512
|
|
219 |
</tr>
|
220 |
</table>
|
221 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
222 |
### EasyAnimateV5-12b-zh-Control
|
223 |
|
224 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
@@ -364,6 +412,13 @@ sh scripts/train.sh
|
|
364 |
# 模型地址
|
365 |
EasyAnimateV5:
|
366 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
367 |
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|
368 |
|--|--|--|--|--|--|
|
369 |
| EasyAnimateV5-12b-zh-InP | EasyAnimateV5 | 34 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh-InP)| 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 |
|
@@ -373,29 +428,29 @@ EasyAnimateV5:
|
|
373 |
<details>
|
374 |
<summary>(Obsolete) EasyAnimateV4:</summary>
|
375 |
|
376 |
-
| 名称 | 种类 | 存储空间 |
|
377 |
|--|--|--|--|--|--|
|
378 |
-
| EasyAnimateV4-XL-2-InP.tar.gz | EasyAnimateV4 | 解压前 8.9 GB / 解压后 14.0 GB | [
|
379 |
</details>
|
380 |
|
381 |
<details>
|
382 |
<summary>(Obsolete) EasyAnimateV3:</summary>
|
383 |
|
384 |
-
| 名称 | 种类 | 存储空间 |
|
385 |
|--|--|--|--|--|--|
|
386 |
-
| EasyAnimateV3-XL-2-InP-512x512.tar | EasyAnimateV3 | 18.2GB
|
387 |
-
| EasyAnimateV3-XL-2-InP-768x768.tar | EasyAnimateV3 | 18.2GB | [
|
388 |
-
| EasyAnimateV3-XL-2-InP-960x960.tar | EasyAnimateV3 | 18.2GB | [
|
389 |
</details>
|
390 |
|
391 |
<details>
|
392 |
<summary>(Obsolete) EasyAnimateV2:</summary>
|
393 |
|
394 |
-
| 名称 | 种类 | 存储空间 | 下载地址 | Hugging Face | 描述 |
|
395 |
-
|
396 |
-
| EasyAnimateV2-XL-2-512x512.tar | EasyAnimateV2 | 16.2GB | [
|
397 |
-
| EasyAnimateV2-XL-2-768x768.tar | EasyAnimateV2 | 16.2GB | [
|
398 |
-
| easyanimatev2_minimalism_lora.safetensors | Lora of Pixart | 485.1MB | [Download](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/easyanimate/Personalized_Model/easyanimatev2_minimalism_lora.safetensors)| - | 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](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/webui/Minimalism.zip). |
|
399 |
</details>
|
400 |
|
401 |
<details>
|
@@ -426,6 +481,7 @@ EasyAnimateV5:
|
|
426 |
|
427 |
# 参考文献
|
428 |
- CogVideo: https://github.com/THUDM/CogVideo/
|
|
|
429 |
- magvit: https://github.com/google-research/magvit
|
430 |
- PixArt: https://github.com/PixArt-alpha/PixArt-alpha
|
431 |
- Open-Sora-Plan: https://github.com/PKU-YuanGroup/Open-Sora-Plan
|
|
|
30 |
#- vllm
|
31 |
---
|
32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
# 目录
|
34 |
- [目录](#目录)
|
35 |
- [简介](#简介)
|
|
|
128 |
|
129 |
我们需要大约 60GB 的可用磁盘空间,请检查!
|
130 |
|
131 |
+
EasyAnimateV5-12B的视频大小可以由不同的GPU Memory生成,包括:
|
132 |
+
| GPU memory |384x672x72|384x672x49|576x1008x25|576x1008x49|768x1344x25|768x1344x49|
|
133 |
+
|----------|----------|----------|----------|----------|----------|----------|
|
134 |
+
| 16GB | 🧡 | 🧡 | ❌ | ❌ | ❌ | ❌ |
|
135 |
+
| 24GB | 🧡 | 🧡 | 🧡 | 🧡 | ❌ | ❌ |
|
136 |
+
| 40GB | ✅ | ✅ | ✅ | ✅ | ❌ | ❌ |
|
137 |
+
| 80GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
138 |
+
|
139 |
+
✅ 表示它可以在"model_cpu_offload"的情况下运行,🧡代表它可以在"model_cpu_offload_and_qfloat8"的情况下运行,⭕️ 表示它可以在"sequential_cpu_offload"的情况下运行,❌ 表示它无法运行。请注意,使用sequential_cpu_offload运行会更慢。
|
140 |
+
|
141 |
+
有一些不支持torch.bfloat16的卡型,如2080ti、V100,需要将app.py、predict文件中的weight_dtype修改为torch.float16才可以运行。
|
142 |
+
|
143 |
+
EasyAnimateV5-12B使用不同GPU在25个steps中的生成时间如下:
|
144 |
+
| GPU |384x672x72|384x672x49|576x1008x25|576x1008x49|768x1344x25|768x1344x49|
|
145 |
+
|----------|----------|----------|----------|----------|----------|----------|
|
146 |
+
| A10 24GB |约120秒 (4.8s/it)|约240秒 (9.6s/it)|约320秒 (12.7s/it)| 约750秒 (29.8s/it)| ❌ | ❌ |
|
147 |
+
| A100 80GB |约45秒 (1.75s/it)|约90秒 (3.7s/it)|约120秒 (4.7s/it)|约300秒 (11.4s/it)|约265秒 (10.6s/it)| 约710秒 (28.3s/it)|
|
148 |
+
|
149 |
+
(⭕️) 表示它可以在low_gpu_memory_mode=True的情况下运行,但速度较慢,同时❌ 表示它无法运行。
|
150 |
+
|
151 |
+
<details>
|
152 |
+
<summary>(Obsolete) EasyAnimateV3:</summary>
|
153 |
+
|
154 |
+
EasyAnimateV3的视频大小可以由不同的GPU Memory生成,包括:
|
155 |
+
| GPU memory | 384x672x72 | 384x672x144 | 576x1008x72 | 576x1008x144 | 720x1280x72 | 720x1280x144 |
|
156 |
+
|----------|----------|----------|----------|----------|----------|----------|
|
157 |
+
| 12GB | ⭕️ | ⭕️ | ⭕️ | ⭕️ | ❌ | ❌ |
|
158 |
+
| 16GB | ✅ | ✅ | ⭕️ | ⭕️ | ⭕️ | ❌ |
|
159 |
+
| 24GB | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ |
|
160 |
+
| 40GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
161 |
+
| 80GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
162 |
+
</details>
|
163 |
+
|
164 |
#### b. 权重放置
|
165 |
我们最好将[权重](#model-zoo)按照指定路径进行放置:
|
166 |
|
|
|
179 |
|
180 |
### EasyAnimateV5-12b-zh-InP
|
181 |
|
182 |
+
#### I2V
|
|
|
183 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
184 |
<tr>
|
185 |
<td>
|
|
|
198 |
</table>
|
199 |
|
200 |
|
|
|
|
|
201 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
202 |
<tr>
|
203 |
<td>
|
|
|
215 |
</tr>
|
216 |
</table>
|
217 |
|
|
|
|
|
218 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
219 |
<tr>
|
220 |
<td>
|
|
|
232 |
</tr>
|
233 |
</table>
|
234 |
|
235 |
+
#### T2V
|
236 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
237 |
+
<tr>
|
238 |
+
<td>
|
239 |
+
<video src="https://github.com/user-attachments/assets/eccb0797-4feb-48e9-91d3-5769ce30142b" width="100%" controls autoplay loop></video>
|
240 |
+
</td>
|
241 |
+
<td>
|
242 |
+
<video src="https://github.com/user-attachments/assets/76b3db64-9c7a-4d38-8854-dba940240ceb" width="100%" controls autoplay loop></video>
|
243 |
+
</td>
|
244 |
+
<td>
|
245 |
+
<video src="https://github.com/user-attachments/assets/0b8fab66-8de7-44ff-bd43-8f701bad6bb7" width="100%" controls autoplay loop></video>
|
246 |
+
</td>
|
247 |
+
<td>
|
248 |
+
<video src="https://github.com/user-attachments/assets/9fbddf5f-7fcd-4cc6-9d7c-3bdf1d4ce59e" width="100%" controls autoplay loop></video>
|
249 |
+
</td>
|
250 |
+
</tr>
|
251 |
+
</table>
|
252 |
+
|
253 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
254 |
+
<tr>
|
255 |
+
<td>
|
256 |
+
<video src="https://github.com/user-attachments/assets/19c1742b-e417-45ac-97d6-8bf3a80d8e13" width="100%" controls autoplay loop></video>
|
257 |
+
</td>
|
258 |
+
<td>
|
259 |
+
<video src="https://github.com/user-attachments/assets/641e56c8-a3d9-489d-a3a6-42c50a9aeca1" width="100%" controls autoplay loop></video>
|
260 |
+
</td>
|
261 |
+
<td>
|
262 |
+
<video src="https://github.com/user-attachments/assets/2b16be76-518b-44c6-a69b-5c49d76df365" width="100%" controls autoplay loop></video>
|
263 |
+
</td>
|
264 |
+
<td>
|
265 |
+
<video src="https://github.com/user-attachments/assets/e7d9c0fc-136f-405c-9fab-629389e196be" width="100%" controls autoplay loop></video>
|
266 |
+
</td>
|
267 |
+
</tr>
|
268 |
+
</table>
|
269 |
+
|
270 |
### EasyAnimateV5-12b-zh-Control
|
271 |
|
272 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
|
|
412 |
# 模型地址
|
413 |
EasyAnimateV5:
|
414 |
|
415 |
+
7B:
|
416 |
+
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|
417 |
+
|--|--|--|--|--|--|
|
418 |
+
| EasyAnimateV5-7b-zh-InP | EasyAnimateV5 | 22 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-7b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-7b-zh-InP)| 官方的7B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 |
|
419 |
+
| EasyAnimateV5-7b-zh | EasyAnimateV5 | 22 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-7b-zh) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh)| 官方的7B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 |
|
420 |
+
|
421 |
+
12B:
|
422 |
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|
423 |
|--|--|--|--|--|--|
|
424 |
| EasyAnimateV5-12b-zh-InP | EasyAnimateV5 | 34 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh-InP)| 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 |
|
|
|
428 |
<details>
|
429 |
<summary>(Obsolete) EasyAnimateV4:</summary>
|
430 |
|
431 |
+
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|
432 |
|--|--|--|--|--|--|
|
433 |
+
| EasyAnimateV4-XL-2-InP.tar.gz | EasyAnimateV4 | 解压前 8.9 GB / 解压后 14.0 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV4-XL-2-InP)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV4-XL-2-InP)| 官方的图生视频权重。支持多分辨率(512,768,1024,1280)的视频预测,以144帧、每秒24帧进行训练 |
|
434 |
</details>
|
435 |
|
436 |
<details>
|
437 |
<summary>(Obsolete) EasyAnimateV3:</summary>
|
438 |
|
439 |
+
| 名称 | 种类 | 存储空间 | Hugging Face | Model Scope | 描述 |
|
440 |
|--|--|--|--|--|--|
|
441 |
+
| EasyAnimateV3-XL-2-InP-512x512.tar | EasyAnimateV3 | 18.2GB| [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-512x512)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-512x512)| 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 |
|
442 |
+
| EasyAnimateV3-XL-2-InP-768x768.tar | EasyAnimateV3 | 18.2GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-768x768) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-768x768)| 官方的768x768分辨��的图生视频权重。以144帧、每秒24帧进行训练 |
|
443 |
+
| EasyAnimateV3-XL-2-InP-960x960.tar | EasyAnimateV3 | 18.2GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-960x960) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-960x960)| 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 |
|
444 |
</details>
|
445 |
|
446 |
<details>
|
447 |
<summary>(Obsolete) EasyAnimateV2:</summary>
|
448 |
|
449 |
+
| 名称 | 种类 | 存储空间 | 下载地址 | Hugging Face | Model Scope | 描述 |
|
450 |
+
|--|--|--|--|--|--|--|
|
451 |
+
| EasyAnimateV2-XL-2-512x512.tar | EasyAnimateV2 | 16.2GB | - | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV2-XL-2-512x512)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV2-XL-2-512x512)| 官方的512x512分辨率的重量。以144帧、每秒24帧进行训练 |
|
452 |
+
| EasyAnimateV2-XL-2-768x768.tar | EasyAnimateV2 | 16.2GB | - | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV2-XL-2-768x768) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV2-XL-2-768x768)| 官方的768x768分辨率的重量。以144帧、每秒24帧进行训练 |
|
453 |
+
| easyanimatev2_minimalism_lora.safetensors | Lora of Pixart | 485.1MB | [Download](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/easyanimate/Personalized_Model/easyanimatev2_minimalism_lora.safetensors)| - | - | 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](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/webui/Minimalism.zip). |
|
454 |
</details>
|
455 |
|
456 |
<details>
|
|
|
481 |
|
482 |
# 参考文献
|
483 |
- CogVideo: https://github.com/THUDM/CogVideo/
|
484 |
+
- Flux: https://github.com/black-forest-labs/flux
|
485 |
- magvit: https://github.com/google-research/magvit
|
486 |
- PixArt: https://github.com/PixArt-alpha/PixArt-alpha
|
487 |
- Open-Sora-Plan: https://github.com/PKU-YuanGroup/Open-Sora-Plan
|
README_en.md
CHANGED
@@ -112,6 +112,41 @@ The detailed of Linux:
|
|
112 |
- GPU:Nvidia-V100 16G & Nvidia-A10 24G & Nvidia-A100 40G & Nvidia-A100 80G
|
113 |
|
114 |
We need about 60GB available on disk (for saving weights), please check!
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
|
116 |
#### b. Weights
|
117 |
We'd better place the [weights](#model-zoo) along the specified path:
|
@@ -131,8 +166,7 @@ The results displayed are all based on image.
|
|
131 |
|
132 |
### EasyAnimateV5-12b-zh-InP
|
133 |
|
134 |
-
|
135 |
-
|
136 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
137 |
<tr>
|
138 |
<td>
|
@@ -151,8 +185,6 @@ Resolution-1024
|
|
151 |
</table>
|
152 |
|
153 |
|
154 |
-
Resolution-768
|
155 |
-
|
156 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
157 |
<tr>
|
158 |
<td>
|
@@ -170,8 +202,6 @@ Resolution-768
|
|
170 |
</tr>
|
171 |
</table>
|
172 |
|
173 |
-
Resolution-512
|
174 |
-
|
175 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
176 |
<tr>
|
177 |
<td>
|
@@ -189,6 +219,41 @@ Resolution-512
|
|
189 |
</tr>
|
190 |
</table>
|
191 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
192 |
### EasyAnimateV5-12b-zh-Control
|
193 |
|
194 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
@@ -335,6 +400,13 @@ For details on setting some parameters, please refer to [Readme Train](scripts/R
|
|
335 |
|
336 |
EasyAnimateV5:
|
337 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
338 |
| Name | Type | Storage Space | Hugging Face | Model Scope | Description |
|
339 |
|--|--|--|--|--|--|
|
340 |
| EasyAnimateV5-12b-zh-InP | EasyAnimateV5 | 34 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh-InP) | Official image-to-video weights. Supports video prediction at multiple resolutions (512, 768, 1024), trained with 49 frames at 8 frames per second, and supports bilingual prediction in Chinese and English. |
|
@@ -344,28 +416,29 @@ EasyAnimateV5:
|
|
344 |
<details>
|
345 |
<summary>(Obsolete) EasyAnimateV4:</summary>
|
346 |
|
347 |
-
| Name | Type | Storage Space |
|
348 |
|--|--|--|--|--|--|
|
349 |
-
| EasyAnimateV4-XL-2-InP.tar.gz | EasyAnimateV4 | Before extraction: 8.9 GB \/ After extraction: 14.0 GB |
|
350 |
</details>
|
351 |
|
352 |
<details>
|
353 |
<summary>(Obsolete) EasyAnimateV3:</summary>
|
354 |
|
355 |
-
| Name | Type | Storage Space |
|
356 |
|--|--|--|--|--|--|
|
357 |
-
| EasyAnimateV3-XL-2-InP-512x512.tar | EasyAnimateV3 | 18.2GB | [
|
358 |
-
| EasyAnimateV3-XL-2-InP-768x768.tar | EasyAnimateV3 | 18.2GB | [
|
359 |
-
| EasyAnimateV3-XL-2-InP-960x960.tar | EasyAnimateV3 | 18.2GB | [
|
360 |
</details>
|
361 |
|
362 |
<details>
|
363 |
<summary>(Obsolete) EasyAnimateV2:</summary>
|
364 |
-
|
365 |
-
|
366 |
-
|
367 |
-
| EasyAnimateV2-XL-2-
|
368 |
-
|
|
|
|
369 |
</details>
|
370 |
|
371 |
<details>
|
@@ -397,6 +470,8 @@ EasyAnimateV5:
|
|
397 |
|
398 |
|
399 |
# Reference
|
|
|
|
|
400 |
- magvit: https://github.com/google-research/magvit
|
401 |
- PixArt: https://github.com/PixArt-alpha/PixArt-alpha
|
402 |
- Open-Sora-Plan: https://github.com/PKU-YuanGroup/Open-Sora-Plan
|
@@ -406,4 +481,4 @@ EasyAnimateV5:
|
|
406 |
- HunYuan DiT: https://github.com/tencent/HunyuanDiT
|
407 |
|
408 |
# License
|
409 |
-
This project is licensed under the [Apache License (Version 2.0)](https://github.com/modelscope/modelscope/blob/master/LICENSE).
|
|
|
112 |
- GPU:Nvidia-V100 16G & Nvidia-A10 24G & Nvidia-A100 40G & Nvidia-A100 80G
|
113 |
|
114 |
We need about 60GB available on disk (for saving weights), please check!
|
115 |
+
The video size for EasyAnimateV5-12B can be generated by different GPU Memory, including:
|
116 |
+
|
117 |
+
| GPU memory | 384x672x72 | 384x672x49 | 576x1008x25 | 576x1008x49 | 768x1344x25 | 768x1344x49 |
|
118 |
+
|------------|------------|------------|------------|------------|------------|------------|
|
119 |
+
| 16GB | 🧡 | 🧡 | ❌ | ❌ | ❌ | ❌ |
|
120 |
+
| 24GB | 🧡 | 🧡 | 🧡 | 🧡 | ❌ | ❌ |
|
121 |
+
| 40GB | ✅ | ✅ | ✅ | ✅ | ❌ | ❌ |
|
122 |
+
| 80GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
123 |
+
|
124 |
+
✅ indicates it can run under "model_cpu_offload", 🧡 represents it can run under "model_cpu_offload_and_qfloat8", ⭕️ indicates it can run under "sequential_cpu_offload", ❌ means it can't run. Please note that running with sequential_cpu_offload will be slower.
|
125 |
+
|
126 |
+
Some GPUs that do not support torch.bfloat16, such as 2080ti and V100, require changing the weight_dtype in app.py and predict files to torch.float16 in order to run.
|
127 |
+
|
128 |
+
The generation time for EasyAnimateV5-12B using different GPUs over 25 steps is as follows:
|
129 |
+
|
130 |
+
| GPU | 384x672x72 | 384x672x49 | 576x1008x25 | 576x1008x49 | 768x1344x25 | 768x1344x49 |
|
131 |
+
|-----------|------------------|------------------|------------------|------------------|------------------|-----------------|
|
132 |
+
| A10 24GB | ~120s (4.8s/it) | ~240s (9.6s/it) | ~320s (12.7s/it) | ~750s (29.8s/it) | ❌ | ❌ |
|
133 |
+
| A100 80GB | ~45s (1.75s/it) | ~90s (3.7s/it) | ~120s (4.7s/it) | ~300s (11.4s/it) | ~265s (10.6s/it) | ~710s (28.3s/it) |
|
134 |
+
|
135 |
+
(⭕️) indicates it can run with low_gpu_memory_mode=True, but at a slower speed, and ❌ means it can't run.
|
136 |
+
|
137 |
+
<details>
|
138 |
+
<summary>(Obsolete) EasyAnimateV3:</summary>
|
139 |
+
|
140 |
+
The video size for EasyAnimateV3 can be generated by different GPU Memory, including:
|
141 |
+
|
142 |
+
| GPU memory | 384x672x72 | 384x672x144 | 576x1008x72 | 576x1008x144 | 720x1280x72 | 720x1280x144 |
|
143 |
+
|------------|------------|-------------|-------------|--------------|-------------|--------------|
|
144 |
+
| 12GB | ⭕️ | ⭕️ | ⭕️ | ⭕️ | ❌ | ❌ |
|
145 |
+
| 16GB | ✅ | ✅ | ⭕️ | ⭕️ | ⭕️ | ❌ |
|
146 |
+
| 24GB | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ |
|
147 |
+
| 40GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
148 |
+
| 80GB | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
|
149 |
+
</details>
|
150 |
|
151 |
#### b. Weights
|
152 |
We'd better place the [weights](#model-zoo) along the specified path:
|
|
|
166 |
|
167 |
### EasyAnimateV5-12b-zh-InP
|
168 |
|
169 |
+
#### I2V
|
|
|
170 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
171 |
<tr>
|
172 |
<td>
|
|
|
185 |
</table>
|
186 |
|
187 |
|
|
|
|
|
188 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
189 |
<tr>
|
190 |
<td>
|
|
|
202 |
</tr>
|
203 |
</table>
|
204 |
|
|
|
|
|
205 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
206 |
<tr>
|
207 |
<td>
|
|
|
219 |
</tr>
|
220 |
</table>
|
221 |
|
222 |
+
#### T2V
|
223 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
224 |
+
<tr>
|
225 |
+
<td>
|
226 |
+
<video src="https://github.com/user-attachments/assets/eccb0797-4feb-48e9-91d3-5769ce30142b" width="100%" controls autoplay loop></video>
|
227 |
+
</td>
|
228 |
+
<td>
|
229 |
+
<video src="https://github.com/user-attachments/assets/76b3db64-9c7a-4d38-8854-dba940240ceb" width="100%" controls autoplay loop></video>
|
230 |
+
</td>
|
231 |
+
<td>
|
232 |
+
<video src="https://github.com/user-attachments/assets/0b8fab66-8de7-44ff-bd43-8f701bad6bb7" width="100%" controls autoplay loop></video>
|
233 |
+
</td>
|
234 |
+
<td>
|
235 |
+
<video src="https://github.com/user-attachments/assets/9fbddf5f-7fcd-4cc6-9d7c-3bdf1d4ce59e" width="100%" controls autoplay loop></video>
|
236 |
+
</td>
|
237 |
+
</tr>
|
238 |
+
</table>
|
239 |
+
|
240 |
+
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
241 |
+
<tr>
|
242 |
+
<td>
|
243 |
+
<video src="https://github.com/user-attachments/assets/19c1742b-e417-45ac-97d6-8bf3a80d8e13" width="100%" controls autoplay loop></video>
|
244 |
+
</td>
|
245 |
+
<td>
|
246 |
+
<video src="https://github.com/user-attachments/assets/641e56c8-a3d9-489d-a3a6-42c50a9aeca1" width="100%" controls autoplay loop></video>
|
247 |
+
</td>
|
248 |
+
<td>
|
249 |
+
<video src="https://github.com/user-attachments/assets/2b16be76-518b-44c6-a69b-5c49d76df365" width="100%" controls autoplay loop></video>
|
250 |
+
</td>
|
251 |
+
<td>
|
252 |
+
<video src="https://github.com/user-attachments/assets/e7d9c0fc-136f-405c-9fab-629389e196be" width="100%" controls autoplay loop></video>
|
253 |
+
</td>
|
254 |
+
</tr>
|
255 |
+
</table>
|
256 |
+
|
257 |
### EasyAnimateV5-12b-zh-Control
|
258 |
|
259 |
<table border="0" style="width: 100%; text-align: left; margin-top: 20px;">
|
|
|
400 |
|
401 |
EasyAnimateV5:
|
402 |
|
403 |
+
7B:
|
404 |
+
| Name | Type | Storage Space | Hugging Face | Model Scope | Description |
|
405 |
+
|--|--|--|--|--|--|
|
406 |
+
| EasyAnimateV5-7b-zh-InP | EasyAnimateV5 | 22 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-7b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-7b-zh-InP) | Official 7B image-to-video weights. Supports video prediction at multiple resolutions (512, 768, 1024), trained with 49 frames at 8 frames per second, and supports bilingual prediction in Chinese and English. |
|
407 |
+
| EasyAnimateV5-7b-zh | EasyAnimateV5 | 22 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-7b-zh) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-7b-zh) | Official 7B text-to-video weights. Supports video prediction at multiple resolutions (512, 768, 1024), trained with 49 frames at 8 frames per second, and supports bilingual prediction in Chinese and English. |
|
408 |
+
|
409 |
+
12B:
|
410 |
| Name | Type | Storage Space | Hugging Face | Model Scope | Description |
|
411 |
|--|--|--|--|--|--|
|
412 |
| EasyAnimateV5-12b-zh-InP | EasyAnimateV5 | 34 GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV5-12b-zh-InP) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV5-12b-zh-InP) | Official image-to-video weights. Supports video prediction at multiple resolutions (512, 768, 1024), trained with 49 frames at 8 frames per second, and supports bilingual prediction in Chinese and English. |
|
|
|
416 |
<details>
|
417 |
<summary>(Obsolete) EasyAnimateV4:</summary>
|
418 |
|
419 |
+
| Name | Type | Storage Space | Hugging Face | Model Scope | Description |
|
420 |
|--|--|--|--|--|--|
|
421 |
+
| EasyAnimateV4-XL-2-InP.tar.gz | EasyAnimateV4 | Before extraction: 8.9 GB \/ After extraction: 14.0 GB |[🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV4-XL-2-InP)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV4-XL-2-InP)| | Our official graph-generated video model is capable of predicting videos at multiple resolutions (512, 768, 1024, 1280) and has been trained on 144 frames at a rate of 24 frames per second. |
|
422 |
</details>
|
423 |
|
424 |
<details>
|
425 |
<summary>(Obsolete) EasyAnimateV3:</summary>
|
426 |
|
427 |
+
| Name | Type | Storage Space | Hugging Face | Model Scope | Description |
|
428 |
|--|--|--|--|--|--|
|
429 |
+
| EasyAnimateV3-XL-2-InP-512x512.tar | EasyAnimateV3 | 18.2GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-512x512)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-512x512) | EasyAnimateV3 official weights for 512x512 text and image to video resolution. Training with 144 frames and fps 24 |
|
430 |
+
| EasyAnimateV3-XL-2-InP-768x768.tar | EasyAnimateV3 | 18.2GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-768x768) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-768x768) | EasyAnimateV3 official weights for 768x768 text and image to video resolution. Training with 144 frames and fps 24 |
|
431 |
+
| EasyAnimateV3-XL-2-InP-960x960.tar | EasyAnimateV3 | 18.2GB | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV3-XL-2-InP-960x960) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV3-XL-2-InP-960x960) | EasyAnimateV3 official weights for 960x960 text and image to video resolution. Training with 144 frames and fps 24 |
|
432 |
</details>
|
433 |
|
434 |
<details>
|
435 |
<summary>(Obsolete) EasyAnimateV2:</summary>
|
436 |
+
|
437 |
+
| Name | Type | Storage Space | Url | Hugging Face | Model Scope | Description |
|
438 |
+
|--|--|--|--|--|--|--|
|
439 |
+
| EasyAnimateV2-XL-2-512x512.tar | EasyAnimateV2 | 16.2GB | - | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV2-XL-2-512x512)| [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV2-XL-2-512x512)| EasyAnimateV2 official weights for 512x512 resolution. Training with 144 frames and fps 24 |
|
440 |
+
| EasyAnimateV2-XL-2-768x768.tar | EasyAnimateV2 | 16.2GB | - | [🤗Link](https://huggingface.co/alibaba-pai/EasyAnimateV2-XL-2-768x768) | [😄Link](https://modelscope.cn/models/PAI/EasyAnimateV2-XL-2-768x768)| EasyAnimateV2 official weights for 768x768 resolution. Training with 144 frames and fps 24 |
|
441 |
+
| easyanimatev2_minimalism_lora.safetensors | Lora of Pixart | 485.1MB | [Download](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/easyanimate/Personalized_Model/easyanimatev2_minimalism_lora.safetensors)| - | - | A lora training with a specifial type images. Images can be downloaded from [Url](https://pai-aigc-photog.oss-cn-hangzhou.aliyuncs.com/easyanimate/asset/v2/Minimalism.zip). |
|
442 |
</details>
|
443 |
|
444 |
<details>
|
|
|
470 |
|
471 |
|
472 |
# Reference
|
473 |
+
- CogVideo: https://github.com/THUDM/CogVideo/
|
474 |
+
- Flux: https://github.com/black-forest-labs/flux
|
475 |
- magvit: https://github.com/google-research/magvit
|
476 |
- PixArt: https://github.com/PixArt-alpha/PixArt-alpha
|
477 |
- Open-Sora-Plan: https://github.com/PKU-YuanGroup/Open-Sora-Plan
|
|
|
481 |
- HunYuan DiT: https://github.com/tencent/HunyuanDiT
|
482 |
|
483 |
# License
|
484 |
+
This project is licensed under the [Apache License (Version 2.0)](https://github.com/modelscope/modelscope/blob/master/LICENSE).
|