Character Workflow: Chroma1-HD + Flux.2 Dev + Wan 2.2 + LTX 2.3

Character Workflow graph

This is an end-to-end character workflow for ComfyUI that lets you create professional quality images and videos while ensuring total facial and vocal fidelity for your character. To get started, all you need is an image of your character and a short audio clip of your character.

Link to workflow: https://huggingface.co/ussaaron/workflows/blob/main/character-workflow.json

Character Workflow uses 4 models that each serve a crucial purpose:

1. Chroma1-HD (arguably the best fully flexible open-source image model).
2. Flux.2 Dev (hands down the best character transfer open-source image model).
3. Wan 2.2 (the most mature video-only open source video model).
4. LTX 2.3 (the best audio-video open source video model).

Character Workflow is a 4-step solution.

1. Generate a base photograph with Chroma1-HD 2.
2. Transfer your character image into the Chroma1-HD gen with Flux.2 Dev.
3. Animate the Flux.2 Dev gen with Wan 2.2.
4. Extend the Wan 2.2 gen with foley, lip-sync, character dialog, and more action with LTX 2.3.

Running the default setup for Character Workflow will take approximately 12 minutes and produce one Chroma1-HD image at 1080p, one Flux.2 Dev image at 1080p, one 3 second Wan 2.2 video at 720p, one 12 second LTX video at 720p.

Here are the results from my one shot run with the default setup for Character Workflow.

Crystal Sparkle character base image

First I generated a text-to-image shot with Chroma1-HD to capture full model creativity.

Chroma1-HD output

Then I did a hyper-targeted update to transfer Crystal into the Chroma gen.

Flux.2 Dev output

Next I animated the Flux gen with Wan 2.2 to have Crystal shooting the blaster off-screen.

Wan 2.2 output

Finally I add foley for the gun, dialog for Crystal, and extend the shot with walk away from camera.

LTX 2.3 output \(trimmed last 4 secs for Reddit bug\)

Character Workflow combines two other workflows I made which you can find here:

Chroma + Flux character transfer: https://huggingface.co/ussaaron/workflows/blob/main/chroma\_flux\_character\_transfer.json

There's also a light version (Chroma + Klein 9b): https://huggingface.co/ussaaron/workflows/blob/main/chroma\_klein\_character\_transfer.json

Wan + LTX video extension: https://huggingface.co/ussaaron/workflows/blob/main/wan2\_2\_i2v-with-ltx-id-lora.json

Let me know if you have any questions!

https://redd.it/1tdc3gy
@rStableDiffusion

Wan 2.2 Remix is the best for uncensored video or is there something better ?
https://youtu.be/s4w14gWc58I

https://redd.it/1tdhm3j
@rStableDiffusion

BEGONE PLASTIC FLUX SKIN! - Better Skin v2

https://redd.it/1tdkvb2
@rStableDiffusion