2 ms·
It's cool, but I think if it accepts image prompts it would be 1000x more useful. In other words, I'd like to turn my existing images into 3D skyboxes.
by raincole 3y ago
It's cool, but I think if it accepts image prompts it would be 1000x more useful.
In other words, I'd like to turn my existing images into 3D skyboxes.
- brucethemoose2 3y agoYou can interrogate your image here and copy the text: https://huggingface.co/spaces/pharma/CLIP-Interrogator https://huggingface.co/spaces/pharma/CLIP-Interrogator But theoretically the devs could easily add a more "direct" pipeline with better results. There is a HF diffusers function for encoding input images like they are text prompts.
- raincole 3y agoI'm familiar with interrogation, and image -> interrogation -> t2i will usually generate a completely different image that the original (and it should!), unlike i2i.
- brucethemoose2 3y agoWell there is interrogate, there is i2i, and then there is something in between. You encode the image directly into latent space vectors, and it applies the general "style" of the input, much closer than a text prompt would. I think this is called pix2pix or style transfer.