Direct long-form video,
with characters that never break.
dollyq turns a script into a five-to-ten minute film. Train a character once and reuse it across every scene and campaign — and approve each scene before a single frame renders.
SCRIPT → PLAN → APPROVE → RENDER → NORMALIZE → STITCH
- S1·01Establishing — city at dawn0:04.9
- S1·02Hero enters, mid push-in0:06.0
- S2·01Product reveal, locked face0:05.2
- S2·02Reaction, over-shoulder0:03.8
A character vault
Train a character once and lock it to a model family. It returns identical across every scene, every campaign — no drift, no re-prompting its face.
Approval at every gate
Nothing renders until you approve the plan. Nothing ships until you approve each scene. Three human gates stand between the script and the master.
Bulletproof stitching
Dozens of short clips, normalized to one color, frame rate and loudness, then assembled with clean cuts and scene crossfades into a single long cut.
Generation is fast. Judgment is yours.
We don't build the video model and we don't run GPUs. We own the part that decides a film is good: structure, character, and a human in the loop at every step.
- 1
The plan
The director agent decomposes your script into scenes and shots. You edit the draft — then approve it. Only then does anything render.
- 2
The scenes
Review each scene as it comes back. A scene can't advance until every shot in it is approved. Reject a shot and only that shot re-renders.
- 3
The cut
Sign off on the assembled film. Export a broadcast master, web 1080p, a vertical social cut and a preview — rendered in parallel.
Made for agencies running long-form, multi-part campaigns.
Branded and government content where a face has to stay the same across a dozen films. Not a one-click toy for creators — a director's suite for teams who ship.