Does anyone know of voxel media without distinguishable individual voxels? In 2D there is both pixel art and regular art. I'm curious if 3D works with that, or only with vector graphics.
Volume rendering is an entire sub-field of computer graphics that deals with many representations of 3D data.
CG clouds are often volume rendered using continuous noise fields. CG haze and smoke are sometimes volume rendered using shape descriptions of the boundary, and 3D modulation functions.
Google "Volumetric siggraph" to find a ton of papers about other ways to do 3D volume rendering. Check the images tab for lots of examples of volume rendering.
*edit: btw I can tell you stories about volume rendering non-voxel data from CG production if you want. I used several non-voxel volumetric rendering techniques for fire, smoke and water on the Shrek and Madagascar movies.
A fun presentation I remember from siggraph was when Digital Domain presented their volume rendered they called "Voxel Bitch" and talked about rendering the avalanche sequence in the Vin Diesel "XXX" movie. IIRC, voxel bitch was a volume renderer, not limited to regular 3D voxel grids.
https://www.siggraph.org/conferences/reports/s2002/tech/film...
Yes!
So, volume rendering in production. I was in the effects department at PDI/DreamWorks, and I used several different volume rendering techniques over the years. We all did, part of why it was such a cool place to work was how free we were to use or invent whatever techniques we wanted to get our shots done - most CG production is a lot more fixed pipeline and regimented than that.
Particle rendering (aka splatting) is probably the main staple for volumetrics in the FX toolbox. Particles are best thought of as soft transparent spheres, but usually implemented as camera facing polygons. I used large volumetric particle sets for impact dust from feet & fight sequences in Shrek, and for smoke and fog and fire in later Shrek productions and in Madagascar. The visible texture actually comes from changes in particle density, and particle simulations were often in the millions of particles.
We used a ray-marching volume renderer implemented inside of a shader for some of the foggy sets in the Shrek productions. Pretty much the same tricks going around ShaderToy today. Dragon fire was another ray-marching shader that would ray-trace spinning spheres, and then ray-march through a noise field inside each sphere.
In one shot on Madagascar, I used a combination of particles and volume rendering and geometry when the boat blows it's horn and flattens the island. I happened to be working on the island's foliage as well. We didn't have low res standins for the plans for a shot this wide, and it was taking more than 12 hours to render IIRC. We wrote a script that culled any geometric or volumetric detail that projected down to smaller than a pixel, and it brought renders down to an hour or two, and turned out to cut something like 90% of the scene geometry. Doh.
examples: https://www.reddit.com/r/low_poly/ http://lopoco.tumblr.com
Just like pixel art is inspired by the aesthetics of low resolution graphics and/or limited color palettes, low poly is inspired by those of early 3d art and games, like N64, Playstation, etc.
I meant voxel art in the way of 2D art that is not inspired in the limitations of resolution or color palettes. Think of scanned drawings, "HD" video games or most DeviantArt stuff. These works are made of 2D pixels, but the 2D pixels are small enough to not be individually visible.
This is what I was talking about specifically: https://www.youtube.com/watch?v=5AvCxa9Y9NU
Raytracing also lets you produce a 2D raster image of a 3D scene with arbitary-resolution surfaces in it, although that still relies on geometric scene descriptions.
Otherwise, it's a bit of a problem to produce; too many voxels to hand-place.
I'm sorry, but this just cracks me up! LOL!
(To explain, "regular art" was and still is pixel art, and "pixel art" is pixel art at an artificially low resolution, emulating the lower hardware capabilities of old machines.)
And to answer your question: When computers get fast enough and have enough memory, you could have voxels at higher resolutions. When individual voxels do practically get too small to see, you have the analog of what happened with pixel art.
We were calling pixel artists "pixel artists" before "pixel art" came to be a distinctive low resolution style. That's just a fact.
And, no, there is no such thing as infinite resolution.
Now you're getting philosophical! :)
I thought they were fast enough and with enough memory for this, at least with static scenes in the style of this portfolio. Do you know of any art created this way?
I think the most practical way out would be procedural generation, but then it's arguably not manual art anymore.
I guess it could be practical with the right kind of environment. If you had all the tools of a good pixler like Deluxe Paint extended to a voxel-painting program.
As opposed to throwing rays and calculating the intersection with the fractal's equation.
I'm thinking that maybe rendering voxels is both slower and produces worse results in this case.
I can confidently say that the native description of the objects seen in that video is a direct mapping from 3d-coordinates to values. [Edit: So you can't just simply directly calculate an intersection point with a polygon. If you wanna cast a ray, you have to sample the mapping all along the way the ray is taking, in discrete steps, until you get a "hit". Optionally, you can then use e.g. bisection to get a more precise position of the intersection.]
For you edification, here's another video of a high-res voxel world. The voxel models used where converted from vectorized surface models. But the description of the models used by the renderer is a proper voxel description, maybe some octree-like thing, or whatever they came up with. https://www.youtube.com/watch?v=f3Ets6bWjEQ