About this prompt
Character-Locked Scene for a Multi-Clip Video
Four scenes, the same description, no reference image: a brunette with glasses in scene one, a blonde in scene two, a redhead in scene three. The audience notices it is a different person every cut. Continuity in AI video comes from three things used together — a reference image, a character block repeated verbatim in every prompt, and an explicit wardrobe lock — plus an end-frame hold so each scene hands off cleanly to the next. This prompt packages all four.
How to use. Generate or photograph a clean, front-facing portrait of your fictional character once. Write the character block and never edit it again. For each scene, change only the action, location, camera, lighting and end frame; keep the colour temperature consistent with the previous scene. Review every clip against the first one before stitching.
What to change. For product-led stories, the "character" can be the product: same bottle, same label orientation, same lighting temperature. For Veo, the same structure works; for Kling, the reference-image step is the one that does the heavy lifting.
Not a fit. Single clips, and real people — generate a composite character instead.


