Image above was created using Gallery Show in Studio Artist. I'm using generative ai paint synthesizer mutation. The source folder of images i'm working off of has nothing to do with seashore photos.
I then used above as the input to one-shot image resynthesis below. Increasing the disentanglement as you drop down for the first 3 below.
I then added additional text embedding (the text prompt was the title of this post) to help steer the resynthesis below.
The next one is a double resynthesis, where the one shot input is used to generate an embedding that is then augmented by the second text steering prompt embedding, and then text to image synthesis is used for the output. Note the difference in appearance vs the 2 above.
I then ran a set of text to image synthesis examples to show off how different the resynthesis approach looks as compared to straight text to image synthesis.
No comments:
Post a Comment