Skip to main content

​
AI generation

Pindeck generates child images from library originals using fal.ai, with optional VLM analysis via OpenRouter. Runs are orchestrated through Trigger on app-vm; progress appears under Work in the app top bar.

​
In the app

1

Open an active library image

Gallery or Table → click to open the detail drawer.
2

Variations tab

Choose a mode (e.g. Action Shot), framing, and count—not a separate “Generate Variation” button on the tile.
3

Generate

Click Generate … to queue work. Watch Work for Trigger status.
4

Lineage

New images appear in the library with VAR styling; Lineage in the drawer shows parent and children.

​
Pipeline

​
Variation modes (UI)

Modes are labeled in Title Case in the drawer (e.g. Action Shot, Close-Up). Each run creates one or more child rows with sourceType: ai and a link to the parent.

​
Configuration

Set these in the Convex project settings: Orchestration toggle: PINDECK_TRIGGER_ORCHESTRATION_ENABLED — when true, heavy jobs use Trigger (see Architecture overview).

​
Smart analysis HTTP

The /smartAnalyzeImage endpoint uses OpenRouter VLM for analysis before or alongside generation. Requires INGEST_API_KEY Bearer auth and image ownership checks. See Smart Analyze endpoint.

​
Variation metadata

​
Rate limits and costs

  • fal.ai charges per generation — monitor usage in your fal.ai dashboard
  • OpenRouter charges per VLM analysis call
  • Generation time varies (often tens of seconds per image; Work shows live status)