Bad Apple, but it’s rendered and colorized with ControlNet [Stable Diffusion Deforum test]

Obviously, based on the famous clip for the Touhou song ’Bad Apple!!’ Used Anything v4.5. The initial rendering was accomplished with canny edge detection and ’scribble’ ControlNet model, then the resulting grayscale video was put again through CN, but now with ’hed’ preprocessing and hed model to get the coloration. Deforum settings files: render — , coloration — The black and white canny edges scribble-mode mid-stage rendered video . You can check the prompts keyframing file to see if I got the characters names right. Despite my best efforts, some of the characters appearances were off, I think it was either the character names were incorrectly spelled (I was referring to the Touhou wiki) or they had been underrepresented in the dataset or there was overlap between their naming and some other franchises. There is also the issue of random unrelated characters appearing when the prompt contains no characters and instead is purely abstract, I guess it relates to this specific model retraining. The pure rendering process took around 12 hours, but it actually stretched to around two days because of my computer crashing multiple times amidst the process or the animation being completely off. To make it more smooth and accessible, I reduced the frame count threefold at first and then interpolated the video back using another free and opensource RIFE AI-model. Then upscaled the video x2 with realesr-animevideov3. This video was made for the purpose of exploration and showing off the new amazing ControlNet feature in Stable Diffusion AI-art system using the frontend of Deforum and AUTOMATIC1111’s webui. Any similarities with Apple logo are purely coincidental and related to the model’s bias. Feel free to ask me any questions on the process.