Video production — plan deep, generate elsewhere

The most important and most involved post type. GoUltra takes you as far as you want: most videos stop at the script; when you choose to go deeper it breaks the script into clips → shots, plans & generates the asset images, and writes the model-tuned prompt for each clip. GoUltra never generates the video itself — you take the prompts + assets to your chosen tool (Higgsfield / Google Flow) and bring the cut back. Demo brand: Fidelis Global · a 30s client-proof reel.

The model you'll generate on isn't where we make the video — it's the reference that sets each clip's max length, its approach, its shot style, and which assets it needs. Specs verified 2026.

1 Script default stop Shape the script in conversation (hook → beats → settings). Hand off a brief and you're done — or take the opt-in step into clips & assets. view → 2 Profile + clip breakdown opt-in Pick the generation profile (planner-recommended), then AI breaks the script into clips — each within the model's max duration, aiming high to minimize clip count — fully editable (merge / split / re-time). view → 3 Clip — shots + assets + prompt the heart A clip → its shots (single, or first→last-frame for Veo); the asset checklist each shot needs; and the model-tuned video prompt to paste into the external tool. Chat-refine on any prompt or asset. view → 4 Asset library — cast · locations · generate A reusable video-level library (characters, locations, products, logo) assigned across clips/shots. Per asset: already-have / upload / generate in-app / get-a-prompt — plus composite ("put this character in this scene → a frame"). view → 5 Hand off + bring the cut back Every clip with its prompt + assigned assets, ready to copy into Higgsfield / Google Flow (or a Claude project). Generate there, then drop the finished video back here to schedule. view →