Support my work by buying me a coffee! Tossing even just a couple dollars my way will help me fund more Qwen Image Edit LoRAs. Thank uuuu ~~ <33
Have you ever wanted to see what a film or scene's animation layout would look like?
The animation layout phase lets animators animate their characters in an environment where it's easy to recognize clear shapes and form, with basic materials set for the environment and the characters that inhabit it. This is essentially the entire scene except without things like lighting (raytraced/path traced), volumetric fog, ambient occlusion, global illumination, or realistic shadows; just basic colors and forms with a matcap applied.
This allows animators to animate their characters efficiently, since computer resources aren't being used to update calculations of the lighting constantly as the animators position the characters' rigs.
This LoRA for Qwen Image Edit (non-plus, ie. original model) allows you to peer into what a scene might look like in it's animation layout phase. This can be useful as a learning opportunity, since this can give you clues on how the scene was actually animated, giving you insight on how it could have been done.
Use this at strength 1, at 20 steps, CFG 4, with just the following text:
Turn the image into an animation layout model renderThis LoRA works on art, people, animals, objects, among other subjects. This LoRA's effect isn't as strong with the 4-step lightning LoRA, so I recommend the 8-step one since it's much, much more accurate (though the best accuracy is using the model without any accelerators).
Why the Non-Plus version of Qwen Image Edit? I chose the Non-Plus version because the new Plus version of Qwen Image Edit has a really hard time at style transfer. The Non-Plus version of Qwen Image Edit is SO much better at style transfer it's not even funny.
Description
Initial version.
FAQ
Comments (7)
this series of yours is super cool. great look. all you're missing is a wireframe render and then you have an entire cg making of
I've been thinking of creating a wireframe render "style", but Qwen Image Edit can do wireframes reasonably okay, so I'm not sure if it's worth the training time. Unless of course you mean "x-ray", in which Qwen Edit isn't really that good at that. Thanks for the support :3
@lilylilith qwen edit can do the clay render with wireframe out of the box? thats pretty cool. havent tried that. I'll check it out
not talking x-ray. just a wireframe topology render
Woah! Would love the other way around :D <3
Yes, the turning an animation layout render into a full render is coming up next. However, I am currently training another LoRA that might end up eclipsing that one, to the point that it might not even be worthwhile to train for (hopefully all goes well!). It's my most ambitious LoRA yet, and I really hope that Qwen Edit can parse together the concepts. Thanks for the support :3
Why 'Plus' and 'Non-Plus'? Isn't that a bit confusing?
Qwen-Image-Edit was released in August 2025, and Qwen-Image-Edit-2509 in September 2025.
Why not just use the official names from Qwen rather than invent new ones?
Qwen-Edit
Qwen-Edit-2509
I appreciate what you are trying to say, but often times when I refer to Qwen Image Edit people automatically assume I am talking about the latest 2509 model instead of the "old" default model. This happens a lot, especially talking about it in Discord. Officially the Qwen team refers to the latest model as Qwen Image Edit Plus (see the diffusers pipeline on the main Qwen Image repo), so people have run with that name. So, imo I think it makes sense to use "non-plus" to refer to the older model.