Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
Drag an initial image into the image node, adjust the prompt, and press queue. The rest should fit.
Runtime profile
Source description
Howto
Drag an initial image into the image node, adjust the prompt, and press queue. The rest should fit.
Description
A Wan 2.2 GGUF Image to Video workflow with the new Lightning Lora to speed up the generation time
Note that upscaling still better happens in an extra step. You can find example upscaling workflows in my article about video upscaling in comfyui: https://www.tomgoodnoise.de/index.php/video-upscaling-in-comfyui/
This workflow comes in two versions. One with all nodes expanded. This one is a good start to download and install all the models and loras. And one that uses the new subgraph feature to make the workflow super compact by hiding away everything not needed.
AI-generated commentary
AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.
Use Wan 2.2 Image to Video GGUF 14B Lightning to turn an initial image and an adjusted prompt into a video: place the image in the image node, edit the prompt, and press queue.
Choose the expanded version to see all nodes, or the compact subgraph version to hide the details.
Upscaling is described as an extra step after video generation.
An example output is a 1024 × 576 video using 14B GGUF models and the Lightning Lora.
The base model is Wan Video 14B i2v 480p.
The requirements call for at least 12 GB VRAM and 32 GB system RAM; the workflow was generated with 16 GB VRAM.
Arunderan is listed as the publisher, and the expanded-node version is described as a starting point for installing the models and LoRAs.
The workflow lists these model files: umt5_xxl_fp8_e4m3fn_scaled.safetensors, wan_2.1_vae.safetensors, wan2.2_i2v_high_noise_14B_Q4_K_M.gguf, and wan2.2_i2v_low_noise_14B_Q4_K_M.gguf.
The workflow lists these node packs: comfyui-frame-interpolation, comfyui-gguf, comfyui-videohelpersuite, and rgthree-comfy.
It uses Wan 2.2 GGUF image-to-video processing with a Lightning Lora described as speeding generation time.
An example 1024 × 576 video with the 14B GGUF models and Lightning Lora is reported at 8 minutes; do not treat that figure as a runtime promise.
Suggestion · not verified
Before use, verify that these model files are available: umt5_xxl_fp8_e4m3fn_scaled.safetensors, wan_2.1_vae.safetensors, wan2.2_i2v_high_noise_14B_Q4_K_M.gguf, and wan2.2_i2v_low_noise_14B_Q4_K_M.gguf.
Suggestion · not verified
Before setup, verify which comfyui-gguf node pack is intended; the listed name alone is not enough to choose one.
Suggestion · not verified
Does this need to be edited?
Sign in to send an edit request.
Sources
1 sourceSource excerpts
1 excerptSource context: 438 downloads · Type Workflows · Base model Wan Video 14B i2v 480p
Time
The example video in 1024 x 576 pixels with the 14b GGUF models and the Lightning Lora took 8 minutes.
Requirements
This workflow was generated with 16 gb vram. Minimum requirement is 12 gb vram. And you should not have fewer than 32 gb system ram.
Estimated VRAM requirement
Estimate unavailable
18.0 GB across 2 of 4 model files. Model file total + 25% loading overhead + 2 GB execution buffer, rounded up.
Requirements
8 requirementsumt5_xxl_fp8_e4m3fn_scaled.safetensors
File unverifiedText encoder · SAFETENSORS · Unknown
VAE · Unknown
Checkpoint · 8.99 GB · GGUF · Unknown
Checkpoint · 8.99 GB · GGUF · Unknown
Node pack · Registry
comfyui-gguf
PossibleNode pack · Registry
Node pack · Registry
Node pack · Registry
