fix: add scipy dependency for remove_black_bg

fix: pad-to-bucket instead of stretch, preserve exact aspect ratio, edge-fill padding

fix: use bucket system for inference resolution to prevent content cropping, then resize back to original

fix: resize output back to original dimensions to prevent 16-align cropping

add remove_black_bg: flood fill to remove only border-connected black, keep inner black elements

fix rmbg: add sigmoid activation to match ComfyUI exactly, remove letterbox complexity

revert rmbg post-processing: black bg needs no dilation/threshold

rmbg post-processing: dilate mask + soft threshold to preserve text edges

use diffusers QwenImageEditPlusPipeline instead of local qwenimage module to fix LoRA compatibility

switch to Qwen-Image-Lightning-4steps-V2.0 LoRA to match ComfyUI exactly

use official FireRed-1.1 model directly + Lightning LoRA, remove prithivMLmods transformer

add out_width/out_height params to infer: match ComfyUI EmptyLatentImage behavior

increase default steps to 6 to match ComfyUI quality

fix rmbg: aspect-ratio-preserving pad to square, avoid distortion on wide images

Remove 1MP bucket forcing, keep native resolution aligned to divisible_by=16

Add wide-format buckets and aspect-ratio-preserving resize for title graphics

Add Color Match feature (Reinhard RGB transfer)

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update requirements.txt

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

Update app.py

update app ✅

Update README.md

update app ✅

update app ✅

update app ✅

Update README.md

update app ✅

update app [notes]✅ (#2)

add examples ✅

update app [examples] ✅

add examples ✅

update app [examples] ✅

add examples ✅