Description
Is there an existing issue for this?
- I have searched the existing issues
Contact Details
No response
What should this feature add?
I know this probably isn't the place to be asking, but I thought I would post.
With RAG becoming popular with standard chatbots for working with documents
Do you know if ImageRAG is coming down the pipeline?
https://rotem-shalev.github.io/ImageRAG/
Seems like a user-friendly feature. The ability to pull images from a folder or database rather than training a Lora.
Seems like this would be the way forward, specifically for image generation.
Say you had a client who had some photos and wanted them all amalgamated together, such as their office, company car and team into an AI-generated photo. It would take forever to do each team member individually, and then pictures of the office in a Lora and then the fleet vehicles. probably weeks of Lora training.
With ImageRAG you could bang them in a database or folder and instantly guide the AI to generate the photo with a prompt as if the AI had been trained on those images.
Especially for normal users who just want to make funny pictures of themselves. No more training Loras, put the pictures of yourself in a folder and then just write the prompt.
Will Invoke ever integrate this? I know it's probably years down the line. But just thought I would throw it out there..
Alternatives
No response
Additional Content
No response