4 ms·
Is there an AI that creates realistic text from photos?
by mjan22640 4y ago
Is there an AI that creates realistic text from photos?
- glenneroo 4y agoLook into image-to-text or CLIP captioning. There are several tools out there, although similar to running text-to-image tools locally, may require some time to setup via conda. The easiest way that I know of (currently) would be install Visions of Chaos. You will still need to follow the TensorFlow setup guide[0] which will take 1-2 hours, then, running Visions of Chaos > Mode > Machine Learning will download 3-400GB of models in order to be able to run all the available tools they have integrated, even if you aren't interested in most of them. The setup may require several retries and several hours due to some servers being extremely slow (or overloaded). EDIT: Just found this[1] as well, though setup might also be a pain. [0] https://softology.pro/tutorials/tensorflow/tensorflow.htm https://softology.pro/tutorials/tensorflow/tensorflow.htm [1] https://replicate.com/methexis-inc/img2prompt https://replicate.com/methexis-inc/img2prompt
- Jach 4y agoSort of. https://github.com/pharmapsychotic/clip-interrogator https://github.com/pharmapsychotic/clip-interrogator
- astrange 4y agoiOS is pretty good at it if you enable VoiceOver. Microsoft has an app called Seeing Eye that’s not as good at it. Both of those use older pre-CLIP technology.