Hugging Face's logo

Phr00t
/
WAN2.2-14B-Rapid-AllInOne

Image-to-Video
Wan2.2
wan
accelerator
Model card Files
xet
Community
181
WAN2.2-14B-Rapid-AllInOne
File size: 1,468 Bytes 
55f0f39
 
 
 
 
 
 
 
 
0e68051
55f0f39
 
 
 
 
 
 
835b25d
 
e38fdbe
55f0f39
835b25d
 
1381258
 
 
d90d9c8
 
b648659
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
 
---
base_model:
- Wan-AI/Wan2.2-I2V-A14B
- Wan-AI/Wan2.2-T2V-A14B
tags:
- wan
- wan2.2
- accelerator
---
This is an "all in one" merge of the WAN 2.2 models into 1. Models also include umt5xxl CLIP and WAN 2.1 VAE. FP8 precision.

Basically, I mixed in "high" + "low" WAN 2.2, WAN 2.1 models, and finally added Lightx2v and PUSA loras for speed.

I'm going for a "jack of all trades" solution here: speed, WAN 2.2 features, WAN 2.1 lora compatibility and simplicity.

You only need to use the "Load Checkpoint" ComfyUI node and can pull the CLIP and VAE alongside the Model.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/631be8402ea8535ea48abbc6/t_SxUFP9oyNz0C8dj6jze.png)

4 steps, 1 cfg, sa_solver sampler and beta scheduler highly recommended. If you don't like the higher contrast look, you can try euler_a or other samplers. YMMV.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/631be8402ea8535ea48abbc6/GNDAWnRHAjt8vPY0wXNTq.png)

Initial testing shows this is pretty neat and is compatible with WAN 2.1 loras. Give it a try!

![image/png](https://cdn-uploads.huggingface.co/production/uploads/631be8402ea8535ea48abbc6/F3tB7EhHMS1Gn-7iplmV8.png)

![image/png](https://cdn-uploads.huggingface.co/production/uploads/631be8402ea8535ea48abbc6/70X-8YUbn5hPogrG5V8Kv.png)

Seems to work even on 8GB VRAM:

![image/png](https://cdn-uploads.huggingface.co/production/uploads/631be8402ea8535ea48abbc6/i4NRFi7FX_j7FUZyvmImw.png)