Frustrated with manual prompting for large scenes, desires reference features

cocktailpeanut · x · 2026-08-29

A user吐槽 the frustration of manually prompting large AI scenes, noting that even using an LLM to help is frustrating. The wish is for a workflow where one can upload reference images, tick options like 'weak reference' or 'fully preserve', link audio to subjects, and provide a brief description to get results.

Original post →

More from Multimodal

Multimodal channel →