eun2ce
fix: add factor=28 to smart_resize in prepare_image for Qwen VL models
f4ee71d verified