lovis93

/

next-scene-qwen-image-lora-2509

like 531

First, I love the concept. This seems super useful for people like me.

In your workflow, the reference image provided is the output of the last scene. I wonder, if we can use the image slots 2 and 3, too. What I want to do is providing a face close-up in addition for HQ character consistency, may be also the outfit. And let the lora do the environment consistency and the visual style.

If I do this, I get a super bad image quality output. Unless I take last frame to a slot other than image1, but in this case I'm not sure if the lora still does anything.

Could you share, if more than 1 reference image is even supposed to work and if yes, whether or not there is a logic behind it where order matters or the lora listens to a specific one?