ABERCROM-ME v3 — IC-LoRA for LTX-2.3

Transform a photo of yourself (or a friend, or a couple) into a scene straight out of a vintage 2000s-era A&F Quarterly catalog — moody monochrome, sun-drenched beach or locker-room energy, and a full styling makeover to match. This is an image-conditioned LoRA (IC-LoRA): you feed it a reference photo and it restyles the subject, clothing, and background to match the aesthetic while preserving identity and pose.

This is version 3 — versions 1 and 2 never made it past my own testing, they just weren't good enough to put in front of anyone else. This one finally is.

How to use it

Prompt with a simple trigger word:

ABERCROM-ME

Or add more detail to steer the specific scene, setting, or action — the trigger sets the style, the rest of your prompt guides the moment.

Recommended LoRA strength: 0.8–1.0. Both work well; 0.8 tends to keep clothing/textures a touch more coherent, 1.0 pushes the styling a little harder.

Workflow

TL DR I realize but there's a lot going on here and took weeks figuring it all out.

This uses the same IC-LoRA pattern as other reference-image LoRAs you may already use (Ingredients, MSR, etc.) — reference image in, styled result out.

If you're running a 2-phase (base + upscale) pipeline: apply this LoRA only in the base generation stage. Do not apply it during the stage-2 upscale pass — that's the one thing that'll trip people up coming from a standard workflow.

My ComfyUI workflow is attached in the required files section. The only thing you really need to do is upload a starting image and set your model paths. Check the note nodes on the far left for the custom nodes it depends on (Pixaroma nodes and my new CoachBate node pack - neither are required and could be replaced with other nodes if you like, but these are pretty nice).

The workflow has 3 stages, with stage 2 using the LTX spatial upscale model, and stage 3 using the LTX temporal upscale model that doubles your framerate. There is no other ComfyUI workflow for using the ltx-2.3-temporal-upscaler-x2-1.0.safetensors temporal upscale model out there, so let me know how this works for you or if mine can be improved any. You can just skip stage 3 if it gives you problems and use RIFE or FILM interpolation but it looks great at 24 fps even. It uses Ollama to generate a prompt or you can add more specifics yourself and have it enhance it, but it will use the trigger word ABERCROM-ME at the beginning. The workflow says to make something exciting happen in the 8 seconds (or however long you set your duration) and looks at the reference image for inspiration. Of course you can bypass that node and write your own. I feel that you do need the 2 stages to get the results I intended on producing but you can try running it with the second stage off also in my User Settings section.

Feedback welcome

This is a first public release, and I'd genuinely appreciate feedback on where it holds up and where it breaks — odd clothing artifacts, identity drift, background coherence, that kind of thing. Helps shape where v4 goes.