It's more to do with the Wordpress theme that you're using. Hitchcock loads full resolution images for each portfolio item, rather than providing a smaller and more sensible thumbnail.
Consequently you're getting a home page that is 90mb in size.
If you have any web developer friends you can probably get this fixed for the cost of some beer.
When I tried visiting earlier, it was down because the database was over capacity, so a caching plugin will go a long way too. W3 Total Cache is a good option, but there are others as well.
Thanks!
For motivation and process, you can learn more on this interview: http://80.lv/articles/sir-carma-magica-voxel-was-like-having...
For the time it takes to do that, really depends on the size of the scene. For example, the recent temple I made took me around 5 or 6 hours.
Thanks!
I never start with a color palette (and I think I should correct this).
I start modeling with basic colors and then adjust the scene's color all along the modeling process.
And finally, in post-prod, I continue modifying lighting and colors in Lightroom.
Can you please briefly explain your working pipeline? i.e. What tool you use? Do you use 2d tools to sketch it out first? What are the materials on the voxels?
Hey! Thanks :)
I don't really have a complicated pipeline.
Since most of it is made for fun, as a hobby, I never really know in advance what I'm gonna do. So no first sketch on paper or 2D, I just go where my imagination takes me, directly in the voxel modeling tool.
I only use Magica Voxel, I love its intuitive controls and interface (but it's very limited in terms of max size - 126^3 voxels scenes- and no selection tools, so cube by cube).
There's no "real materials" in Magica voxel, it's only colors. But you can add some attributes to a chosen color (4 attributes available at the moment : diffuse, metal, glass, emission).
I render directly in Magica Voxel then, since the last 6 months, I started to do more and more post-production on my scenes with Adobe Lightroom (could have use Photoshop but it's easier for me on Lightroom), to work on lighting, colors, etc.
All of my work really have to be considered like concept arts (or maybe even sometimes target render) but not in-engine. There's so many cubes in my scenes that in Unity it would be very slow. But I believe we will be soon able to have that quality in-engine!
Hey Sir_carma! I was just curious if the resolutions posted to your blog were the highest-resolution/file type available? I love this kind of art and I'd love to be able to print some of it, but I'm afraid using the images you provide might result in some lower-than-desired quality prints.
Alternatively, if there's a place we can buy the art, that would work as well!
Two years ago, after seeing the incredible art style of Fez, I became a big fan of voxel-based sculpting. I implemented my own WebGL voxel editor and started a Tumblr with some example creations: http://eachothersbreath.tumblr.com.
Unfortunately it didn't take me long to feel the limitations of that art style. The lack of sloped surfaces is a huge problem, which can't be mitigated even with custom shaped voxels. Past a certain point, tasteful low-poly modeling (like http://www.chelseasaunders.com) is both easier and looks better, especially when animated.
I've not tried voxel art yet, but I've been playing with pixel art for the past 18 months or so and the limitations are actually fun for me. It means that you have to come up with some creative solutions.
Just like pixel art, voxel art looks great in a parallel projection where all voxels have the same visual size. But in a perspective projection, when some voxels are much closer to the camera than others, the limitations become insurmountable. It's similar to early FPS games like Doom, where enemy sprites would become very pixelated up close. The wakeup call for me was unlocking first person mode in Fez and slowly realizing that it doesn't look too good, even though the same scene would look amazing in side view.
I think the closest 3D analog to pixel art is not voxel art, but low poly art with low resolution textures. Something like this: http://orig03.deviantart.net/c602/f/2012/065/8/f/lowpoly_nak... is a restricted aesthetic in its own right, and looks great in a variety of views, but would be very hard or impossible to do with voxels. It's a shame that few people are pursuing that direction nowadays, everyone is jumping on the voxel bandwagon.
I agree re: low poly models (to achieve an aesthetic instead of for performance) being the next generation of pixel art. The Witness had a very distinct visual style where the models are almost entirely low poly but the textures are high res and lighting, post processing, etc. are all modern. There's a lot that can be done there.
Btw, does anyone have any examples of this they like? Games, old or new, that used low poly exceptionally well? Several PSX games come to mind for me...
The Witness has an amazing art style, but unfortunately it's completely out of reach for individual developers, so it's not quite analogous to pixel art. Check out Sky Rogue or Drift Stage for something that's more realistically achievable.
> But in a perspective projection, when some voxels are much closer to the camera than others
I think that's why in most of the examples in Sir_carma's portfolio are presented from a distance, where the perspective projection will have less effect.
> I think the closest 3D analog to pixel art is not voxel art, but low poly art with low resolution textures.
I disagree. Low-poly art is a vector format, so it makes more sense to me to compare it to 2D vector drawings with a small number of control points.
Voxel art has similar resolution and color limitations to pixel art, and they share a certain aesthetic that I don't see in low-poly designs.
I think that right about now is when we'll see a renaissance in low-poly artwork. I say this because I think the people who grew up playing PS1 era games are now going to do something similar to what the people who grew up playing 8-bit and 16-bit games did. See Bob Mackey's piece on this: http://www.usgamer.net/articles/back-in-1995-and-the-changin...
Orthographic projection can be used (you could bake a voxel model into a sprite sheet like Doom used). Still, even with perspective it looks "ok" IMO - but then again I am hugely biased! :)
Those animations for Babi Glo look fantastic! I love the whimsical style of the artwork. Are you still making progress on this? Any idea of a projected release date?
Sadly, I'm not Chelsea :-) As far as I know, she lost all her progress due to a hard drive failure and I'm not sure if Babi Glo will ever happen. But it's definitely a huge inspiration.
You mentioned having difficulty getting the shading right in one example (your chair) and not wanting to implement stencil shadows. Is there no lighting whatsoever being done in the renderer?
Check out the roofs I made with voxels: http://www.voxelquest.com/uploads/9/5/4/0/9540564/3786947_or... its possible, just depends on your resolution and style. I use voxels with normals (direction of light reflection) which is unusual in most voxel art.
Out of curiosity, do you still feel that voxels are worthwhile? Compared to the amazing stuff that you can achieve with polygons, e.g. http://www.orsispanyol.com/art/.
In your typical commercial project it would be pretty foolish to use anything other than polygons, just because that is where the tools, engines, and APIs are at.
Still, from the perspective of making something unique looking, many people seem to like voxels. It is almost comparable to the preference of pixel art over vector art. As a warning, if you do anything with voxels, even just something Minecraft-ish, you will be opening the floodgate for many technical battles. :) From a conservative perspective, it is hard to justify the use of them.
One of the take-home lessons from the last Global Game Jam was that there is https://ephtracy.github.io/ which can produce stunning voxel art (I saw quite a few participants using it).
When I was young and first dialed up to the budding Internet I somehow found myself making modding tools for games, including voxel editors for Tiberian Sun. A few years back I found some exes and they still ran in Wine: http://williamedwardscoder.tumblr.com/post/69583927490/i-was...
Global illumination adds lots of charm to most of the works there. I personally like the minimalist 3d style and global illumination of some sort makes even single cube alive.
I think that simple 3d models with GI effectively make programmer's art that look minimal, but not necessarily cheap.
Very nice work! I like the combination of the discrete/precise geometry with the haze/fog atmospheric effects. I also make architectural art in 3D but prefer the geometric flexibility of NURBS rather than voxels. (shameless self promotion: https://www.instagram.com/jeffreyemanuel/ )
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.
Yes, there's quite a lot of it in fact. It's usually called "volumetric", as "voxel" tends to imply volumetric data that is represented by a regularly spaced 3D grid.
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.
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.
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.
That looks pretty good, but it's not what I was asking for.
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.
I think they mean more like volume rendering, like might be generated by an MRI or CT scan. In those cases, you end up with a voxel-like 3D data set of discrete points. You can either render the data as voxels, or connect the points into a 3D mesh (like in a normal 3D model).
Not sure, that looks to me like it might be a primitive version of same, although as it's mono, it's probably generated from some sort of laser/ultrasound rangefinder.
I think some CT systems produce this kind of image. There are also laser point cloud scanner systems that have been used for archaeology, and various attempts at reconstructing 3D from two or more cameras. And the Microsoft Hololens demo of free sculpting in AR.
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.
(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.
Your own definitions do not even add up. Regular images (not regular art) are technically infinite resolution images with the only limitation being the 32bpp colour scheme (which can go up infinitely too, but displays are a limitation) Pixel art is a style and movement replicating the low resolution and limited colour (and sometimes other limitations such as sprite limits per scanline of NES) sprites, especially of older games.
Even an image created in a physical medium (e.g. oil paints) has a limited resolution. Vector/polygon art is "infinite resolution" (but not infinite detail/complication), and so are some other forms of mathematically-defined images (example: fractals like the Mandelbrot set, which are infinitely detailed as well).
> When computers get fast enough and have enough memory, you could have voxels at higher resolutions
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?
So that's what you were getting at. I've never seen anything like what you describe. I suppose it would be a very tall order to pay the same attention to every detail in a high resolution voxel-image as in a high resolution pixel-image.
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.
I think the question of how you are rendering your voxels is a technicality. Redering voxels does not to my mind imply a specific rendering technique any more than rendering vector graphics does.
When I render a low resolution voxel image using vector descriptions of the cubes containing those voxels and then passing them through a rasterizer, is it still a voxel image? Certainly, because the image is natively described as a mapping from coordinates to present/not-present flags (and optionally color).
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
Have faith in yourself! You can go a long way with generative art, which boils down to applying algorithms to build interesting pictures. Building an really good generative artwork requires a lot of work, but the techniques are those of the developer.
- Google "generative art". You'll find lots of images and you can usually work out roughly how they were made. Try to reverse engineer them and create your own facsimile.
Start with something that looks a bit like something you might want to make into art. One of the standard fractals, Perlin noise with a fixed seed, anything that's got a bit of randomness to it. Look for a section that's inspiring. Then look to see if you can tweak the logic to make that closer to what you want it to be.
if you see beautiful art and feel a yearning to explore art it probably means you might have some talent for it.
the best place to start in my opinion is by learning about colour theory which is one of the fudamentals of visual art, like with programming learning some fudamentals at the beginning saves alot of time in the long run. Learning how to use colours will instantly stop the stuff you make from looking like it is made by a 5 year old, it will make it look at minimum presentable and something you will want to show to other people and that they will react positively to.
It is a time taking investment but a very emotionally rewarding one.
Clicking any of the pieces gives me the WordPress "Error establishing a database connection" error. Not sure if it's just the hacker news hug of death, but it would be worth taking a look and seeing if your hosting is having trouble with the number of visitors you're getting.
Most of the shots are rendered using Magicavoxel, which make things looks fantastic with its realistic shadows and reflections. Anyone know how that's done? I've used the program and it renders those scenes very quickly.
Sorry for my website having problem handling so many visits coming from HN :x Really need to change provider.
If it still doesn't work for you, you can check my art on my twitter page : https://twitter.com/Sir_carma
And don't hesitate to ask me questions over there, I'll gladly answer! Cheers Sir_carma