archkre Posted April 28, 2006 Share Posted April 28, 2006 Dear friends: Do you have any news if the new Maxwell 1.0 is faster and less grainy? I never understood Next Limits' concept that "the result of a rendering depends on how long you want the processing of the light solution! I understand it is the othe way about: I want quality, how long is it going to take? -------------------------------------------------------------------------------- Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 It's as fast as RC5 but the jury is still out on whether it's faster, bottom line is you still have to wait a very long time for the noise to clear. The only positive thing in the speed department is cooperative rendering which I'm still testing but it looks like with enough PC's you will be able to produce work in a reasonable amount of time. Once I get network rendering working I'll be able to tell you exactly how long it takes to produce a high rez image. Link to comment Share on other sites More sharing options...
archkre Posted April 28, 2006 Author Share Posted April 28, 2006 So it takes a render farm! Get out of here Goombahs! Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 Its scene dependent and it also matters whether you’re doing interiors or exteriors and whether or not you want to use any noise reduction software. Exterior scenes render reasonably fast, you could probably get one done overnight if the resolution wasn't too high. Interior scenes are completely different, I rendered one last night at 2500x1875 this was the result. I'm also using a dual core dual processor system with 4 gigs of ram, this rendered for 15 hours. Link to comment Share on other sites More sharing options...
archkre Posted April 28, 2006 Author Share Posted April 28, 2006 I am sticking to Vray then! Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 I think that is a smart idea, at least until we get to V2 and see some speed improvements. Link to comment Share on other sites More sharing options...
leoA4D Posted April 28, 2006 Share Posted April 28, 2006 ... at least until we get to V2 and see some speed improvements. In the "V2" spot above, insert beta, RC1-5 or V1. The same or a similar phrase has been used within hours of every new M~R release. Depressing. OT: "goombahs!" That made me chuckle; it has been a while since I heard that one. Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 I know it seems like were never going to get there. Link to comment Share on other sites More sharing options...
AJLynn Posted April 28, 2006 Share Posted April 28, 2006 Guys, I hate to be the one to say this, but: Maxwell will not become significantly faster than it is now. There, I said it. Maxwell is a render engine based on well established algorithms, that must peform a certain amount of calculation in order to render an image. The only way to reduce this amount of calculation is through adaptive sampling algoritms, which would be "biased" and kill the "beta love" image quality, so NL will not do this. They will continue to find minor shortcuts, but most of the improvement we've seen and will see is through code optimization - finding places where computation is happening than is required, e.g. inefficiencies, and making them more efficient. I can't believe that after all this time Maxwell is still only half as efficient as it could be - my guess would have to be in the 80-99% range - so the 100% speed increase we've been hoping for will not happen, and the 10x that was mentioned fleetingly is really very unrealistic unless NL plans to abandon unbiased calculation entirely. I say this with no particular knowledge of Maxwell's code or MLT, but enough programming background to be confident. Link to comment Share on other sites More sharing options...
edub Posted April 28, 2006 Share Posted April 28, 2006 Guys, I hate to be the one to say this, but: Maxwell will not become significantly faster than it is now. at least cpu speeds will continue to increase - so, following Moore's Law, in 18 months you'll have your render times cut in half Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 Actually CPU speeds are relatively stuck where they are and the CPU makers are instead going to continue to add additional cores to their processors in order to continue the performance increases. This is going to cause Maxwell some problems since you are only allowed to use 4 CPU's per each license. If computer makers continue with this trend and we have 10 core processors in a few years, Maxwell in it's current state will only be able to use 40% of the available power of a single chip. So it really doesn’t matter at this point how much faster chips get, unless NL changes there core = CPU policy no one will be rendering any faster unless they purchase more licenses which may have been their plan all along. Link to comment Share on other sites More sharing options...
_PopArt Posted April 28, 2006 Share Posted April 28, 2006 heheh but moore's law doesnt work anymore...hence the dual cores. Yes i think it is true, maxwell 1.0 represents the best mlt implementation possibile at this time. It is good, it is slow, and that's that. Network rendering is the best solution right now, but NL has implemented a complicated system, it will take a bit of time to verify just how efficient it is. We all know this should have been the RC6 with all the bugs and still alot of optimization to be done, like for ram usgae and memory leaks, but we all know how this works by now. Take it or leave it... Link to comment Share on other sites More sharing options...
AJLynn Posted April 28, 2006 Share Posted April 28, 2006 Victor has said that the licensing policy will be updated as the "standard" chips are updated. No idea exactly what this means. For this whole thing, the old saying applies: "Fast, good or cheap. Pick two." Link to comment Share on other sites More sharing options...
AJLynn Posted April 28, 2006 Share Posted April 28, 2006 We all know this should have been the RC6 with all the bugs and still alot of optimization to be done, like for ram usgae and memory leaks, but we all know how this works by now. Take it or leave it... Technically, all the previous versions should have been labelled Alphas, RC5 should have been labelled Beta and this one should be RC. You just don't get to publicly release five release candidates unless you're on a nightly build system and still have your release scheme taken seriously. Link to comment Share on other sites More sharing options...
archkre Posted April 28, 2006 Author Share Posted April 28, 2006 Wasn't ol' Lighjtscape of similar quality results but much more faster and grainless about ten years ago? Despite the modeling nuissances I felt is has no posible competitors in Archviz field! Link to comment Share on other sites More sharing options...
Devin Johnston Posted April 28, 2006 Share Posted April 28, 2006 Lightscape was a raytrace renderer, not the same technology but it is considered by many as the Holy Grail in terms of its ease of use and image quality. It was later reincarnated in Max and Viz as a radiosity engine but for some reason I don't think it ever was able to recreate Lightscapes images. Link to comment Share on other sites More sharing options...
AJLynn Posted April 28, 2006 Share Posted April 28, 2006 Lightscape was a raytrace renderer, not the same technology but it is considered by many as the Holy Grail in terms of its ease of use and image quality. It was later reincarnated in Max and Viz as a radiosity engine but for some reason I don't think it ever was able to recreate Lightscapes images. They didn't fully implement the radiosity engine (though now that Max has adaptive subdivision it's close) and they didn't use the Lightscape raytracer, which was a strong point, or the shader system, which was easy to use and had a lot of presets. As I see it, the main advantage that the newer GI programs have is the ability to calculate more types of light transmission - radiosity is for diffuse light, but Vray, Maxwell, mental ray et al can handle specular and glossy as well. Link to comment Share on other sites More sharing options...
_PopArt Posted April 28, 2006 Share Posted April 28, 2006 Technically, all the previous versions should have been labelled Alphas, RC5 should have been labelled Beta and this one should be RC. You just don't get to publicly release five release candidates unless you're on a nightly build system and still have your release scheme taken seriously. Yes I agree, this 1.0 is an RC for sure, maybe an RC 4 or 5 while RC5 was more like a RC1 or late beta ,whatever, its confusing:) . Its important to emphasize, even if it seems silly, that NL is willing to release at this point and RC with alot of bugs as a complete 1.0 version, because most, almost all other professional companies would not do that. Look at vray who prefer to push things back indefinietly and piss alot of people off, rather than release a 1.5 that isnt rock solid. But this is NL, and we've seen it all before with realflow etc, short term sales seem to be the most important thing. The material system, even if not as reliable as beta, creates really cool stuff, i give them credit here, bsdf bi directional scattering distribtion function material are light years ahead of lightscape or the max 'implementation' of lightscape:) Link to comment Share on other sites More sharing options...
AdamT Posted April 29, 2006 Share Posted April 29, 2006 From all I've been reading 1.0 certainly isn't worthy of RC status. It's clearly still in beta form. Link to comment Share on other sites More sharing options...
Giovanni Scocchera Posted April 30, 2006 Share Posted April 30, 2006 From all I've been reading 1.0 certainly isn't worthy of RC status. It's clearly still in beta form. I think exactly the same, but probably I'm wrong... probably it must be considered only an alpha SW. Ciao, Jo Link to comment Share on other sites More sharing options...
runrun Posted April 30, 2006 Share Posted April 30, 2006 I think NL can improve the cooperative network rendering. But so far, I think it's pretty cool. It was actually easy for me to setup, and the merge makes sense to me. My only problem is that there is no clear S.L. number that I know a final image will be after being merged? As it stands, I use 3 nodes, different machines and I don't finish the S.L. on all of them, so they are merged at different levels. Link to comment Share on other sites More sharing options...
yog Posted April 30, 2006 Share Posted April 30, 2006 From all I've been reading 1.0 certainly isn't worthy of RC status. It's clearly still in beta form.If I'm being as un-biased as I can, I would say that Maxwell Studio is a Ver-1, but the MAX and Lightwave plug-ins are still back at the beta stage (possible Rc-1). Still no faster though Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now