The following images were made using my custom build of Disco Diffusion, photo-manipulation and digital drawing through Gimp and Inkscape, and post editing/composition in darktable.
I've been experimenting with training a new model using dreambooth and textual inversion, since it seems like thats where the fun is these days. Most of my friends/co-artists are all training very realistic robotic model these days, but I wanted to work more in a "comic"/"2d into 3d" approach. The kind of result you can get with MathRockDiffusion for example, but mixed with the madness of DiscoDiffusion, into Stable Diffusion.
So, I took a WHOLE BUNCH of my creations as the bases for this new model, as well as images and art pieces from artists going from Moebius to Enki Bilal, went on a creation bender, re-fed the result through textual inversion to the model, and repeat, again and again.
This is the first result coming out of it without an initial_image, purely prompted and composited in gimp and darktable! Do you think its okay? I feel like there's still room for improvement, and much left to train in that model, but so far it's looking quite good!
That's all for today, I hope you liked what you saw and what you read, if you did or not just let me know in a comment, you know you have the power to influence what I will make next! Let me know what you would like to see!
Also, if you want to check out more of my generative work, or 3D design, or even artworks, check me out online, where I publish most of what I do.
