CSV-driven image generation is useful when every output shares a layout but contains different text, photos, QR values or identifiers. The important work happens before rendering: clean data, explicit mapping and a useful preview.
A practical CSV shape
Column names do not have to match the template field names, because mapping connects them. Clear headings still make review and reuse easier.
| name | role | photo_url | reference_id |
|---|---|---|---|
| Asha Menon | Speaker | https://…/asha.webp | EVT-1042 |
| Ravi Das | Volunteer | https://…/ravi.webp | EVT-1043 |
Dynamic images need stricter validation
- Use HTTPS URLs that return an image without an interactive login.
- Keep source images reasonably sized; large originals increase transfer and processing time.
- Check orientation and crop behaviour with representative portraits and landscape images.
- For later regeneration, managed records preserve the source reference rather than retaining every uploaded source indefinitely.
Why preflight matters
Preflight provides a controlled checkpoint for mappings, output format, row count and invalid values. Fixing a bad URL before generation is cheaper and clearer than discovering it in hundreds of completed outputs.
Large jobs
The per-job row limit is a technical maximum, while available output credits determine how many rows can actually run. Large jobs are queued and processed in batches; workspace allowances are reserved atomically so two members cannot spend the same remaining credits.
Common questions
Can CSV values include image URLs?
Yes, when the template contains a dynamic image field and the URL is accessible to the renderer.
Can I regenerate one failed row?
Managed records support later correction and regeneration. Temporary output should be treated as a disposable batch.