Updated: Nov 10, 2026
conceptSimply describe the areas you wish to modify, and the image will be generated based on your prompt. This approach performs better than the original LoRA node when used with a Turbo model (at 4–8 steps).
Supports both 3D and photorealistic reference styles.
支持4~8步加速,支持3D和写实参考。
服装细节一致性更高,工作流已给出。

