feel like its not worth trying for accurate flux gens on low vram, it just cant nail accuracy i want on loras compared to bigger models i cant run.