major breakthrough in rendering
recently, I saw a video that talked about a major breakthrough in the area of rendering, I would like to post the link in the forum for the development team and the general public to have a look: youtu.be/svLzmFuSBhk
7 Comments
Algorithms like ray marching known since 1989 are not exactly a breakthrough in 2019...
As explained in the video, you need a good estimator function. But they are only available for geometric primitives it seems. Thus, the usage is severely limited. It also doesn't seem compatible with Blender, more specifically polygon based models with arbitrarily shapes, which are used as input for raster and ray tracing based rendering engines.
* @codemanx * Don`t underestimate primitives. Triangles are primitives too. And metaballs are sort of primitives, but i made this {attachment 160246jpg}
* @ Vyacheslav * Yes, they are. But a whole lot of them complicate the estimator function's math and/or work to a point where you need to do a lot of things for good estimates or you need more steps. At this point, it may be cheaper to compute a result using rasterization or ray tracing techniques.
I don't know whether metaballs count as primitives. They are certainly powerful in modelling, but if they are expensive to render using ray marching then there is little sense in using parameterized objects instead of a triangulated representation of them.
* @codemanx * the good thing is raymarching works with parametric geometry, not with triangles. And primitives are not so primitive.
shadertoy.com/results
Yea, nice. But what do you want to see in Blender?
* @Eric * I see, that it can be useful like procedural volume fill.
* @ Vyacheslav * cool, I see now.