Fuse this image into background
Description
FAQ
Comments (15)
Great lora! it would be amazing if you can train it to work with nunchaku.
you can use the nunchaku version of kontext: https://huggingface.co/mit-han-lab/nunchaku-flux.1-kontext-dev/tree/main
@Optimum the point is - it does not work with nunchaku lora loader and errors out cause it cant find all lora parameters
Yea, this doesn't work with nunchaku unfortunately. Throws an error with the nunchaku lora loader.
Optimum try using the nunchaku lora patcher - https://huggingface.co/lym00/comfyui_nunchaku_lora_patch
SpaceOperator88 Thanks! I changed it to recursivley check all my loras. Turned out only one needed repair! but it is good to have the tool!
Works amazingly well. Due to the lack of instructions - here my experience: input is only a single image with a poor montage of a person or object pasted on background (no separate images of object and background as I first thought). Use the default trigger, but an additional prompt is helpful, e.g., "fuse...background, the person gets light from the window" for indoors. Or "fuse...background, warm backlight from the sun, hard shadows" for outdoors. Thanks for your work.
Thank you!!! It was so helpful
I feel like this Lora training round is not enough, underfitting. Of course, maybe there is something wrong with my workflow
Just some clean-up: Would be good to have the before/after examples right next to each other in "Post Image Order" (you can re-arrange them), to make it easier to see before/after.
the trigger words is correct. but you have to add more detail about the lighting context after the trigger words.
kaytransg196 It wasn't when I wrote :)
hey what is the prompt?
merhabalar hocam
Prompt: "Fuse this image into background"
Amazing, thank! what is the Lora scale? 1 or 0.8?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.