📂Files :
Recommendation :
>24 gb Vram: base or Q8_0
16 gb Vram: Q5_K_S
<12 gb Vram: Q4_K_S
For base version
VACE Model: wan2.1_vace_14B_fp8_e4m3fn.safetensors or wan2.1_vace_1.3B_fp16.safetensors
In models/diffusion_models
CLIP: umt5_xxl_fp8_e4m3fn_scaled.safetensors
in models/clip
For GGUF version
VACE Quant Model: Wan2.1-VACE-14B-QX_0.gguf
In models/diffusion_models
Quant CLIP: umt5-xxl-encoder-QX.gguf
in models/clip
VAE: wan_2.1_vae.safetensors
in models/vae
ANY upscale model (depreciated):
Realistic : RealESRGAN_x4plus.pth
Anime : RealESRGAN_x4plus_anime_6B.pth
in models/upscale_models
📦Custom Nodes :
Description
Completely redone version with VACE, and the optimizations present in my new workflow
FAQ
Comments (9)
thanks for this, but how do you install magcache? can't be downloaded from manager and it's not on your nodes auto installer.
i have the same problem
Only working with VACE? Or it can work with the basic Wan 2.1 too?
v3.0 use VACE, older use normal model but worst result
The base version threw up strange errors about not recognizing the model, and I tried them all. I don't understand that one. But the GGUF version is working. FFLF is awesome, perfect for bridging the gap between frames of clips that are not close enough for the usual editing tricks. Looking forward to playing around with this some more. The workflow does have a ton of nodes collapsed that really shouldn't be. Pretty for the wf image, I suppose. There be much expanding to do when you start.
Is it possible to add a background remover?
Have been having an odd issue with this one. The preview shown will have the full completed clip but the saved output will have the first second or so cut off.
actually nvmd they all skip the first .5-1 seconds with just the first frame in there.
I have this error :
SamplerCustomAdvanced
mat1 and mat2 shapes cannot be multiplied (154x768 and 4096x5120)
What can i do ?