Exploring the universe through algorithmic whimsy. Advocate for naturism and trans rights. Mapping out my tiny region of configuration space. May contain nudity or unnecessary decimal precision.
This one was simulated annealing, trying to match a Gaussian blurred candidate image to the Gaussian blurred target image, with a tiny penalty for differing from the unblurred target pixels to keep it somewhat grounded in reality.
This one was simulated annealing, trying to match a Gaussian blurred candidate image to the Gaussian blurred target image, with a tiny penalty for differing from the unblurred target pixels to keep it somewhat grounded in reality.
- Find the pixel in the image that hasn't been quantized yet that would result in the lowest quantization error, and quantize it.
- Distribute the residual error to all remaining pixels that haven't been quantized using inverse square distance.
- Find the pixel in the image that hasn't been quantized yet that would result in the lowest quantization error, and quantize it.
- Distribute the residual error to all remaining pixels that haven't been quantized using inverse square distance.