I have a good feel for my limits when it comes to art, and where I fall short I try and augment my abilities with code. That’s what got me DEGAS (Digitally Extrapolated Graphics via Algorithmic Strokes), my own implementation of my own algorithm and supporting software in HTML5/Canvas. You can read more about it in my news feed.
I’ve submitted a paper to SIGGRAPH 2018 about it, so fingers are crossed! 🙂
It’s not a neural network-based system like Ostagram, or a filter like Waterlogue, but a painting program that uses a photograph as a reference image. The operator must chose brushes, strokes, and contours – which takes practice to get right – but the choice of orientation and color of each stroke is left up to DEGAS.
As a result, I tend to refer to DEGAS as an “impressionist gattling gun” since the program mimics the process of painting, not just the product, and at much higher speed.
DEGAS also keeps track of every individual stroke and generates a height map or normal map of the paint’s nooks and crannies, so when a painting is completed, it shades the texture of the bristles from the paint’s contour and can be rendered at high enough resolution to be printed at >200-300 DPI on very large canvases.
So far, my record is taking a 7 megapixel image, painting and upscaling it to 250 megapixels.
From the height map it can generate data to 3D print the final image so you can see the brush strokes as a physical texture. A UV ink printer, like what they used in The Next Rembrandt, would be perfect for this.
Virtual Painting Gallery
My Wife, Nayla
The Steampunk World’s Fair 2017