Sign In

VIDEO EXTENSION simple workflow WAN2.1 | GGUF | LoRA | UPSCALE | TeaCache

51
688
16
Type
Workflows
Stats
345
0
Reviews
Published
Apr 2, 2025
Base Model
Wan Video
Hash
AutoV2
73AA4FF619
Comfy Workflow Badge
UmeAiRT's Avatar
UmeAiRT

Due to numerous compatibility issues, the "layer skip" node is disabled by default. It allows for better video quality but causes frequent crashes.

This workflow allows you to generate the continuation from a video using WAN2.1 and export the whole as a new video file

The easiest way to install all the files needed for workflow use is to use one of my installation scripts.
To only install the missing nodes, I created another script.

To save a lot of time on video generation, you can install SageAttention. I've written a guide on the subject: link

Resources you need:

Files :

Recommendation :
24 gb Vram: Q8_0
16 gb Vram: Q5_K_S
<12 gb Vram: Q4_K_S

I2V Quant Model: city96/Wan2.1-I2V-14B-480P-gguf at main or city96/Wan2.1-I2V-14B-720P-gguf at main
In models/diffusion_models

Attention: if you're using a 720p model, you'll also need to change the "Apply Tea Cache" node settings.

CLIP : split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors · Comfy-Org/Wan_2.1_ComfyUI_repackaged at main
in models/clip

CLIP-VISION: split_files/clip_vision/clip_vision_h.safetensors · Comfy-Org/Wan_2.1_ComfyUI_repackaged at main
in models/clip_vision

VAE: split_files/vae/wan_2.1_vae.safetensors · Comfy-Org/Wan_2.1_ComfyUI_repackaged at main
in models/vae

ANY UPSCALE MODEL:

in models/upscale_models

Custom Nodes :