

          --- Stepwise Delusions ---

        music & visuals by teadrinker
        Released at Jumalauta Party 23


Animation sequences was created with Stable Diffusion,
simply by changing the "steps" parameter. No inpainting, 
cropping or manual retouch was made. However, I did often
manually remove entire frames from the sequences to 
make them flow better. I also change framerates over time.

Tools: Stable Diffusion, "X/Y/Z plot" in A1111 web ui, ffmpeg, Reaper, Synth1 VST, Unity, KlakHAP, FFmpegOut, Custom Unity Scripts
Model: redshift-diffusion-v1.ckpt, 4xESRGAN

Typical Prompt: face portrait of attractive edgy male K-pop artist, (sophisticated style), fashion, album cover, love, (loving), outdoors autumn fog, creative, arty, epic composition, photography award, portra, leica, grainy film, analog film, redshift style
Negative prompt: (3d rendering), (3d), (rendered), bad anatomy, text, label, bad proportions, huge breasts, blurry, (photoshopped skin), ((overexposed)), (monochrome), lowres, low quality, deformed, disfigured, duplicate, extra arms, extra fingers, extra legs, extra limbs, fingers, morbid, mutated hands, mutilated, out of frame, tight crop, poorly drawn face, poorly drawn hands, too many fingers, ugly, eyes out of focus

I converted the sequences to HAP Q using ffmpeg,
A Unity project handles video playback, but main timeline
was composed in Reaper (clips are midi-controlled) 

Greetings!
Trepaan, Slipstream, Still, Omega, Poo-Brain, 
Logicoma, Mercury, Cocoon, Fairlight, Trauma, 
Conspiracy, rgba, Rage, Spaceballs, Future Crew!

Thanks to
 * Jumalauta Organizers!
 * Stability AI for releasing model/code!
 * nitrosocke for the redshift model!
 * the inventors of tea!

Peace