v3.2.1
June 5, 2026
More creative control, smarter prompts, and Story Cutter cuts that land exactly where you planned.
✨ New
- AI Optimize — A new Optimize button in Cinematic Lab, AI Transitions, Motion Director, and Add Effects reads your current settings and attached images, then rewrites your prompt for that specific workflow — so you stop guessing what language gets the result you're after.
- Relight — multi-light control — Layer multiple light sources in a single relight session — each with its own type (Key, Fill, Accent), color, depth (in front of or behind the subject), and intensity. Go from a flat single-source look to a full studio setup without leaving the workflow.
🔧 Improvements
- Story Cutter — single-take accuracy — Single-take requests stay single takes. The pipeline now correctly reads one-take intent and stops second-guessing it — no more phantom violations on a cut you explicitly asked to leave in one piece.
- Story Cutter — batch plan fidelity — Batch cuts now pull the full transcript and your confirmed conversation plan for every clip. What you approved in the planning step is what gets assembled — not a best guess from a partial context window.
- Faster video attach — Attaching videos is significantly faster — whether you're dropping a clip into the chat area or loading footage into a Studio workflow. Large files that used to stall now load in a fraction of the time.
- Faster library & Recents loading — The Library and Recents panels load noticeably faster on launch and when switching between projects. Less waiting, more editing.
Relight — full multi-light studio control
Add as many light sources as you need — Key, Fill, Accent, or any combination. Each light has its own type, color picker, depth toggle (in front of or behind the subject), and intensity slider. The result is a lighting setup you can dial in rather than one you're stuck accepting.

AI Optimize — one click to a better prompt
Type a rough idea and hit Optimize. CVP reads your current workflow settings, any images you've attached, and the model you're using, then rewrites your prompt to match — turning a two-word starting point into a full cinematic brief without you having to learn prompt engineering.


