AI scene images
Each scene can carry a generated still, written from the scene's own shot prompt and conditioned on your project's visual style — so a cartoon channel gets cartoons and a cinematic channel gets cinema, without per-image prompting.
AI images with locked identities and a learnable visual style.
Each scene can carry a generated still, written from the scene's own shot prompt and conditioned on your project's visual style — so a cartoon channel gets cartoons and a cinematic channel gets cinema, without per-image prompting.
Characters are library entries with locked identity — same face, costume, and age in every image that names them, across every reel. The planner even suggests characters your scripts mention that you haven't added yet.
Give a character a description and the studio generates their portrait — which then anchors their appearance in every scene image they appear in.
Point style training at a reference reel URL and it studies the look — color, composition, pacing of visuals — and distills style rules your project can apply to its own generations.
Upload reference frames — your own or from creators you admire — and the top references ride along with every image generation, keeping output visually consistent with the look you chose.
Character drift is the tell of AI video — the narrator whose face is subtly different in every scene. ReelMint locks identity at the library level: a character is an entry with a portrait, and that portrait anchors every image that names them, in every reel. A style description alone can't do this — text prompts describe a look, but only reference images hold one.
The same mechanism carries your channel's look. Style training studies a reference reel and distills rules; your uploaded reference frames ride along with every generation. The result is a feed that reads as one channel — not a sampler of whatever the image model felt like that day.
Characters, portraits, style training, and reference images are all in the free tier. No feature gates.
Start free