This is an AI image model, first of the Bobble series. Press space twice to start training. Hold space to see the data. Right now this is just a proof of concept, and at some point I will make it easier to use and with better UI. It fully reconstructed the example image I have in there after about 10 minutes in turbowarp with turbo mode, but now there are 2 images. Feel free to remix and add your own images! You can do so by taking a 10 x 10 image, converting it into color values separated by a space, (eg. 1464246 881863 864786) and the color values are (red x 65535) + (green x 256) + blue, then create 2 new lines in the data set list, the first line being the prompt for the image. Encode the prompt to numbers using the custom encode block. It outputs to encode.output. Copy that and put it in the first of the new lines in the data set list. Then the encoded image goes in the second. I'm working on a website to convert images into the string, but I'm at school rn so i cant see it
