That would be great, but I don’t see it as realistic.
Option 1: Camera Bits builds its own model to tag images, ideally running offline. Without sufficient training data, this is extremely difficult – and also expensive and complex to develop.
Option 2: Images are sent via API to OpenAI or Google, which then generate the keywords. Each request costs money. In addition, many creators do not want their images sent to such companies, since this data may be used to train their models.
The only practical solution I see is a small NodeJS script integrated into Photo Mechanic via a Droplet. That’s exactly how I do it — and it works perfectly.
