Prompt-to-Render Intake Gate (Onboarding)

Prompt-to-Render Intake Gate (Onboarding)

1. 01

   Capture the client's existing brand assets before any prompt is written

   Collect logo files, hex codes, type specimens, and two to three reference clips the client already approves of. Renders made without this baseline get rejected at review and burn render credits twice.

2. 02

   Define the deliverable spec in seconds, aspect ratio, and resolution

   Most engines in this category output 4 to 15 second clips, so a 30 second explainer is a multi-render assembly, not one prompt. Write the shot count into the scope before quoting.

3. 03

   Separate hero assets from volume assets on the intake form

   Hero spots get human-directed finishing; social cutdowns and ad variants stay in the rapid prototyping lane. This split is what protects retainer margin on video work.

4. 04

   Confirm commercial-use and watermark terms for every model in the render stack

   Some workspaces bundle watermark-free downloads and commercial rights, others license per seat. Get the terms in writing before the first client-facing export.

5. 05

   Set a lip-sync and audio expectation note for the client

   Native audio and lip-sync exist in this category but do not hold up at scale for on-camera spokespeople. Flag dialogue-heavy concepts as human-shoot candidates at intake, not after the first review.

6. 06

   Log the model roster and a fallback for each render dependency

   Multi-model workspaces let you swap backends when one degrades, but single-model studio wrappers do not. Document the substitute path so a vendor outage does not stall a client deadline.

7. 07

   Agree the revision ceiling in writing

   Two render rounds per shot is a workable default. Unlimited AI revisions quietly convert a fixed-fee video project into an hourly loss.