Examples of Noise Painting Techniques
The noise painting
The final piece
The noise painting. This one collaged several generative images in addition to just using a sketch
The final piece
The noise painting
The final piece
The noise painting
The final piece (one of them, there were several) this. This one plays fast and loose with depth mapping in addition to the regular stuff we do with a noise paint
The noise paint. Here, we're experimenting with darkness and color shorthands
The final piece
The noise painting. With this one, we're using shorthands for depth and perspective
The final piece sums up my feelings on the importance of fighting art fascism
The noise paint
First Result: Insult
Second Result: Injury (but all is not as it seems)
Testing a new checkpoint, the noise paint, aimed mainly at creating a background and environment
And the final piece
The noise paint: second checkpoint test
The final piece: And I do believe we've made a breakthrough
Stable Diffusion as a medium: Noise painting is a technique that involves using a logographic shorthand to derail the denoising methods and training data used by diffusion models, in order to get either a more vibrant or more tightly controlled piece of art. It gets there by assembling more complicated noise patterns that force the computer to work harder. Using this technique, or techniques like it, you can make an art ai create more interesting textures, shading, or even assets that would not or simply could not appear from the reference data alone, in a standard use case.
The fun of noise painting is that you're not trying to create a final piece with the image itself. They're supposed to be raw, and only contain the markers the machine needs to know what you're talking about. Shorthands for backgrounds, textures, hair, and so on. All of these noise paints were made using Clip Studio Paint and older (long story on the reasoning) adobe products.
