Slashdot Mirror
POV-Ray 3.6 Released
ehmdjii writes "After a long betatesting-phase the POV-Ray team just released version 3.6 of the popular opensource raytracer. It's been two years since the last version and many bugs have been fixed as well as some changes in the render core. This release concentrates on stability and providing a framework for future re-implementations."
Not nearly as nice as some of the pro stuff out there, but definitely usable for the rank amateurs among us.
I wonder if that might be more accurately stated in the reverse: Definitely better than some of the pro stuff out there, but not nearly as usable for the rank amateurs among us.
It's actually far harder to use than simple point-click-and-drag solutions like 3dsmax or Maya, but the results can be just as good. Two of my favorite POV-Ray images:
'The Wet Bird'
'Chado'
I can't even imagine putting those images together using POV-Ray. Using 3dsmax, sure. But POV-Ray? Wow.
"Linux was 3 times faster on the same machine."
Yeah if that result was typical, most of us 3D artists wouldn't be using Windows 2000.
"Derp de derp."
Hmm, then why do the best povray images look so... fake... and plastic?
"Sure the pro ones can claim all kinds of features that are barely more than excuses to use patents,"
Huh? What features are you talking about? I ask because if you go back and forth between MAX, Maya, and Lightwave, there's very little you don't get. Usually the big difference is in implementation, can't say I know anything about the 'excuses to use patents' bit.
As for 'claiming all kinds of features'... what exactly is your motivating thought behind that comment? The last 3 releases of Maya, 3D Studio MAX, and Lightwave (probably true for XSI as well, but I have not followed it) have all had impressive features that made the creation of quality 3D art easier to do. You claim that povray's renderer is 'second to none'. Well I honestly can't tell you where it stands as I haven't used it. I can tell you, though, that these days the big bottleneck is the artist, not the engine moving the pixels around. 'Features' are a big part of making sure an artist can express themselves. If the interface is getting in your way, then the renderer isn't doing you a lick of good.
I have NFI why anybody'd dismiss 'software for the pros' so easily.
"Derp de derp."
I hope you were just being sarcastic. If it got any more realistic no one would believe it was computer generated. In the low-res version, it looks like a photo from a digital camera. Only when you look at it in it's hi-res image can you see those "too perfect to be real" nuances. Personally, I rather like the effect it creates.
IF (and this is a big "if") Moore's law holds (i.e. processor speed doubles every 18 months), then it should happen easily. Some quick calculations come up with:
40 years * 12 months / 18 months = 26 doublings of processor speed .0000008940 fps on a current processor.
2^26 = processor is 67108864 times faster than before
Something that runs at 60 fps in the future / 67108864 =
Taking the reciprocal, we get 1118481 seconds/frame or roughly 12 days per frame on a current processor. So anything we can raytrace now in 12 days, we would be able to raytrace at 60 fps 40 years down the round. Again, assuming that Moore's law holds.
However, I don't think it's going to be so cut and dried. It seems like we're always on the cusp of the failure of Moore's law by reaching some sort of limit on how small we can make things. On the plus side, proper raytracing isn't exactly a prerequisite to good looking, photo-realistic images -- we're getting better and better at "cheating" with more efficient rendering algorithms that look good enough. So we might not exactly have what we're looking for, but I don't doubt for a second that rendered graphics 40 years from now will look amazing.
Don't you mean "the source is open, but it's not Open Source" ?
It's not *closed* source if you can download the source; if you mean "it's not GPL", why not say that?
POV ray definately has its place. However, it really is not a replacement for 3dsmax and a real artist.
Someone who can't draw, use colors and compose will not be a good 3D artist regardless of the tools used. In the same way that a novice photographer can't (normally) compose artistically interesting photographs, a programmer won't be able to make an interesting raytrace without artistic skills even when the tools are perfectly photorealistic.
As far as commerical packages go, they have really nice tools for "massaging" forms like nurbs which are like 3d splines. Artists need to be able to make innumerable iterative adjustments to forms, colors and positions. You can't do that efficiently by only typing equations or adjusting numbers in property boxes.
Both Softimage (XSI) and Alias (Maya) report that rendering is faster on Linux (not 3X, but marginally at least).
The problem is that the interactive portions of the programs are often better in Windows, and sometimes (especially with XSI) include things that aren't possible with Linux (integration with IE, for one, and also often the use of VBS).
So you might think that we should just use Windows for interactive and Linux for rendering, but there were subtle differences in the rendered images - not better or worse, just subtle differences, that made compositing with images generated by both impossible (this is several years ago with Maya). Also, at the time, Linux multiprocessing was a bit more difficult. If we'd have had something like Irix's "runon" command, the animators would have been a lot happier. Moreover, many of our distributed rendering tools are happier running on a single platform.
Add to that the fact that a lot of other tools used - photoshop and other Adobe packages, as well as some other image manipulation programs (like ACDC), and we ultimately require Windows for interactive. Since we are a small shop, we also use those boxes for evening and weekend renders.
Still, rendering IS reported to be faster under Linux, and if you can live with the other tools available under Linux, it's probably a better, more cost-effective option than Windows.
Keep in mind that much of what you are seeing in film these days IS rendered with Linux, not just because it's cheaper (hey, multiply a 1000 boxes times the cost of Windows) but because it IS better.
Stupid sexy Flanders.
Well, with the 486DX it would have been insanely faster even then. You were having to emulate FP math with the SX.
it's interesting to see comments that pit povray against these other applications which are really intended to put a great amount of creative control at the hands of the user...
there are some fundamental differences between povray vs maya/max/etc which aren't so apparent but important to distinguish the applications.
pov models and renders objects based on the mathematical description of the surface. maya/max on the other hand models objects based on triangulating the surface. while it seems esoteric, it is a fundamental difference which really puts these applications into two different classes. try modeling a quaternion fractalor other mathematical shape/function in maya... additionally the complex and random model generation that pov has(onyale's pipe macro, chris colfax macros) is not something easily created with these other apps.
maya and max on the other hand have a strong UI to help put a layer between the user and the code itself. there's no way you could do the type of modeling in pov that you could in maya. pov is very methodical; plan out the image, heavy previsualization, utilization of macros and includes to manage the items you previously created. maya provides immediate feedback as items are created, providing a more fine art approach of creation, kinda like working with clay as opposed to architecting a building.
comparing these applications (pov/max/maya) is like comparing a car to a plane... they are both used for transportation, but the mechanisms to implement the main function is vastly different(it's not the best analogy, take it with a grain of salt)...
perhaps a more productive discussion would be to have some kind of shootout between maya, max, and blender.
three can keep a secret, if two are dead - benjamin franklin
Problem is when you are a professional, its not just about output quality, its about production quality.
If it takes longer to produce acceptable output then its NOT as 'good' in the general sence.
Raw productivity goes down in that situation, which costs money.
Amateurs dont count in the professional world, its just a fact of life. ( being an amateur 3D person myself, i can say that honestly )
---- Booth was a patriot ----
Conversely, POV-Ray is more mature and more portable. For example, Yafray required the very most recent point release of GCC when 0.0.6 came out...sigh). Also, POV-Ray comes with documentation (fancy that!).
Vote in November. You won't regret it.
You mean, if they'd just released version 20.5, you'd have been more impressed ? :-)
The developers concentrate on stability. Lots of other software comes and goes but POV is a stable application that you can start a render on, come back three days later, and not be looking at an 'access violation' messagebox. In fact version 3.5 was so heavily beta-tested that it was able to go two years without a single point release due to the absence of any significant 'crash the app' bugs.
Stability is essential when you want to run an app that may need to make several trillion calculations in a large render. Many POV-Ray users prefer stability over features, and those that don't use one of the unofficial versions :-)