Help: How to maintain the first reference image in MiniMax rev2video
Fit_Satisfaction2953 · reddit · 2026-08-25
A user is encountering an issue with MiniMax's rev2video feature: when providing two reference images, the model sometimes generates a random starting image instead of beginning with the first reference image (e.g., a forest background) as intended. The user has tried explicitly describing the starting scene in the prompt but found the results inconsistent, and is seeking methods to ensure the first reference frame is maintained.
More from Multimodal
- Wan 3.0 Tested Against Seedance and MiniMax H3 — Aiden_Tech_Ai · 2026-08-26
- AI Series Mary Sue Episode 4 Made with CapCut and Seedance 2.5 — gen_ericai · 2026-08-26
- Microsoft MAI-Image-2.6 Tops Image Editing Leaderboard — ArtificialAnlys · 2026-08-26
- PatternEval and PatternRL: Aligning response patterns in hybrid-thinking MLLMs — YouJiacheng · 2026-08-26
- Higgsfield lands in Blender: prompt the scene, animate the camera, reblock in seconds — petewoodbridge · 2026-08-26
- Best current img2video solution for a 12GB VRAM rig — is WAN 2.2 still the answer? — FAUVEisEditing · 2026-08-25