Anyone notice?

[_PopArt]

No its been developed by a single programmer, Csaba Kelemen. His delta sampling technique is pretty interesting, I'm sure NL's spent some time on his webpage, the preview engine in the RC's seems as though they've taken note.

 

 

http://www.hungrycat.hu/index1.html

 

Check these threads:

 

http://forums.cgsociety.org/archive/index.php/t-190261.html

 

"Maxwell is great on making pretty pictures, I really like it and Im glad that there is so many unbiased rendering fun but it seems to me, Maxwell adapts the school style knowledge only.

My task are researching and developing in one. This can explain why the unusual, underground style program name and uncertainty on deadlines."

 

It still looks pretty basic, i hope it comes to light. Cool guy...

[Chris_TC]

Keep in mind that MLT/path tracing rendering has been around for many, many years. Its not a new discovery, and there are quite a few render engines which implement MLT engines.

 

That's true. I found out the other week that even Mental Ray has a path material you can use for path tracing. The results are very similar to Maxwell, but the rendering process is even slower. Which is a shame, because it would make the lighting always turn out "right".

[_PopArt]

That's true. I found out the other week that even Mental Ray has a path material you can use for path tracing. The results are very similar to Maxwell, but the rendering process is even slower. Which is a shame, because it would make the lighting always turn out "right".

 

Exactly!

 

And look at the limitations of Mental Ray's monte carlo brute force engine: SLOW, NOISY, AND PROBLEMS WITH TRANSPARENCY & GLASS!!!

 

"Why isn't everyone using it then? Because it's VERY slow if you want good results. In order to not have any graininess, you need to use very high sampling settings in the Render Globals like Min 4 Max 4. It uses what's known as brute force Monte Carlo sampling to calculate the GI. You can control the number of lighting bounces this shader calculates by setting the raytracing Reflection/Refraction and Max Trace Depth attributes in the Render Globals. For example, setting the Reflections attribute to 5 in the Render Globals and the Max Trace Depth to at least 5, the Path shader will calculate 5 bounces of light. One strange thing is that the shader doesn't seem to work with transparency/refraction."

[Chris_TC]

One strange thing is that the shader doesn't seem to work with transparency/refraction."

 

There's nothing strange about this. Mental Images never implemented it or put a lot of work into the path material because they didn't see anyone want to use it because it's such a slow and old technique.

 

The last time I read about the path material on XSIBase was in a one-year old thread.

In the Maya Rendering forum on CGSociety however they've used it a couple days back to do a comparison between Mental Ray's path material and Maxwell's lighting solution. It's in the "Vray like interiors..." thread. The two images look practically identical. But looking at the rendering times non-Maxwell users might faint.

[_PopArt]

There's nothing strange about this. Mental Images never implemented it or put a lot of work into the path material because they didn't see anyone want to use it because it's such a slow and old technique.

 

The last time I read about the path material on XSIBase was in a one-year old thread.

In the Maya Rendering forum on CGSociety however they've used it a couple days back to do a comparison between Mental Ray's path material and Maxwell's lighting solution. It's in the "Vray like interiors..." thread. The two images look practically identical. But looking at the rendering times non-Maxwell users might faint.

 

I'll look for it...can you post a link? Thanks...

[rivoli]

here's the link, howardb posts (this one and the following page):

 

http://forums.cgsociety.org/showthread.php?t=190232&page=67&pp=15&highlight=maxwell

[_PopArt]

Thanks for the link!! Amazing similarity between the images.

[Ernest Burden III]

Is Brute Force Monte Carlo similar or the same as what Maxwell does? I don't kniow the 'under the hood' stuff. Cinema4D had a version of that (slow, noisy though glass works fine, I did exterior animation with it recently) and so does FinalRender2. I haven't tried the FR2 version yet, but now I will to see how it does with a physical sky/sun and a glass material.

[_PopArt]

Well its all supposition, but one could make an accurate guess.

 

Maxwell probably is a monte carlo derivative based on the metropolis light transport alorithms (mlt), meaning its a path tracer, which could be considered 'inverse global illumination', in the sense that it traces paths from the light sources to the camera lens, which is an inversion, if i understand correctly, from traditional montecarlo brute force, and is capable of much more powerful things:

 

http://graphics.stanford.edu/papers/metro/metro.pdf

 

http://www.3dvirtualight.com/mlt/

 

From this (mlt) it evolved (or more likely was from the beginning) a bi-directional path tracer - with bi directional path tracing rays are shot & traced both from light sources and the camera, and is capable of handling even more complex situations - scattering, anisotropy, etc, etc...

 

Then maxwell has also integrated spectral values into some of the core algorithms, but the values generally has to be interpreted back to rgb for image viewing. Once again, NL isn't the first to used the spectral system and surely wont be the last.

 

Vrays ppt is pretty similar to bi directional path tracing, but it currently has some heavy calculational limits as compared to maxwell...

 

Path tracing, as slow as it is, tends to excell in certain scenes, and not so well in others, as compared to monte carlo brute force. In reasonably well lit scenes such as the mentalray path tracing/maxwell comparison (kindly linked above), it would seem that the difference is not so great, although i'm not too familiar with mentalrays path tracing technique - need to check it out:)

 

So to sum it up its generally a pretty sophisticated evolution of montecarlo brute force, but it can often be difficult to tell the difference with monte carlo brute force, except in certain situations (caustics, spectral refractions, diamonds, glass, etc - in these situations its true power starts to show)

[olbo]

Well a lil bit off topic but ... just take a look at this thread at cgtalk. This

guy is developing a renderer for 3dsmax. He is doing it for his studies and

I'm very impressed what one person can do in five month. And now ... just

think a lil bit further ... :D

 

 

take care

Oleg

[Karlfucious]

Why is everyone in such a hurry to get ahold of RC5? Is everyone in a hurry to just get dissapointed again by another crappy release? Another thing I keep hearing is the problem with the physical sun not going through glass. You Max users are acually blessed that the physical sun acually does anything. In Maya the physical Sky and sun doesnt even work at all. The only thing that you can use in Maya is the emiters. If you want to physical sky and such (Which is really the reason to use maxwell in the first place) you now have to use Maxwell Studio. Pain in the Arse. Also they havent addressed ANY of the bugs that I could think of from the alpha version. Acually they seem to have added even more bugs as they went along. Aside from the terrible speed and extra noise the alpha version seems to be the most stable for Maya which is pretty crappy if you ask me. Acually I take that back they did fix one bug. The Maxwell Cameras don't hijack your perspective camera anymore in the beta version lol. Also still no nurbs. For some reason Maxwell can render nurbs in Rhino, Solidworks, and probably a few other apps but not in Maya? Thats rediculous.

 

Personally I am just going to give up on Maxwell because it turned ot to be a total waste of my time and money. I dont think that RC5 will fix any problems with Mayall. Even if they do fix the Sun/dielectric problem that doesnt do too much for me because it dont work in maya anyway.

[Maximus3D]

If you assume RC5 will be a failure then you're wrong, maybe you set out to be dissapointed even if it renders the whole world down to every little neutron in realtime on a pocketcalculator. I know there's alot of people who think like that. Be realistic and not stupid please.

 

/ Max

[frankmeyers]

Maximus the Faithfull, what do you consider stupid actually?

[Ernest Burden III]

If you assume RC5 will be a failure then you're wrong, maybe you set out to be dissapointed

 

Its a fair question to ask, then, what would make RC5 a failure? Will its very existence (not that any of us have it yet) make it a success to you, or is there some level of function and progress that must be met short of which it would be a failure in your eyes? I guess different people have a different bar. Where's yours?

 

 

To be fair, I don't know how I would answer that.

[bricklyne]

Its a fair question to ask, then, what would make RC5 a failure? Will its very existence (not that any of us have it yet) make it a success to you, or is there some level of function and progress that must be met short of which it would be a failure in your eyes? I guess different people have a different bar. Where's yours?

 

 

To be fair, I don't know how I would answer that.

 

An RC5 failure would include;

 

1. Inability to perform some of the basic functions that were easily achieved by the Beta, and this includes both render speed and image (lighting) quality.

 

2. Inability to maintain a fair degree or minimum level of stability when running ( i.e not a crash/bug-fiesta).

 

3. Not addressing some of the glaring shortcomings of the Beta version given that it's coming almost 7 full months after the Beta.

 

-At this juncture I should point out that given the above criteria, RCs 1-4 were all esentially failures, and I would dare challenge anyone, (including you, oh faithful and Loyal Maximus) to prove this not to be so; the Studio notwithstanding.

 

4. Lack of more than half the features originally announced for version 1.0, given that the RC is supposed to be closer to version 1.0 than the Beta is.

 

5. Lack of significant or observable improvement in speed - code optimizations or lack thereof, notwithstanding. 7 months is 7 months!!!

 

6. and of course, Lack of an appearance before the Groundhog Phil sees it's shadow (or not), or while snow still sits on the ground. Given that the longer an acceptable and working RC is delayed, the final version is also likewise delayed by a commensurate factor of about 1.5 times the same time it took us to get from RC 1 to RC5; the continued delay of RC5 essentially implies that a final and acceptable Version 1.0 (or 1.xxx) with all announced features may not make it into the market until late 2006, which may not be a problem for some people, but for others whose operating costs are tied into their equipment, this translates into loss of projected and expected revenue and productivity.

 

7. Don't even get the Mac/OSX crowd started.

 

 

 

Maximus, we have every single reason to believe that RC5 WILL be a failure given NL's most recent release history ( not just RC's 1-4 but also the Beta) and given the fact that they themselves have also attempted to temper expectations by pre-emptively announcing that not all expected features ( or even funtionality ;- Sun/Glass) should be expected anytime before version 1.0. Don't even get me started on their so-called expected "optimizations".

 

Normally I would like to believe that given that they are taking so long to release RC5, we should expect a stable working version, but I know better. So should you. Blind loyalty and indefensible support by fanboys is just as harmful, if not more so, as reputedly unwarranted pressure to release the product, by the clientbase.

[jackb602]

If you assume RC5 will be a failure then you're wrong, maybe you set out to be dissapointed even if it renders the whole world down to every little neutron in realtime on a pocketcalculator. I know there's alot of people who think like that. Be realistic and not stupid please.

/ Max

 

It is perfectly realistic to expect progress (or lack thereof) on Maxwell, as well as any communication, to proceed at exactly the same pace it has so far. If you let go of a ball 100 times, and it drops to the ground each time, you can safely assume it will do so the next time you drop it.

 

Given that Next Limit hasn't changed its outward behavior at all, the only reasonable conclusion I can draw is that they will continue to release incomplete/buggy software, miss their own deadlines, fail to communicate, and to treat their paying customers poorly. I would love to see some evidence to make me think differently.

 

Jack

[AJLynn]

6. and of course, Lack of an appearance before the Groundhog Phil sees it's shadow (or not)

 

His name is Punxsutawney Phil, and he's a he, not an it

[AdamT]

I would consider RC5 a success if it more or less matches the beta, but with integrated new material system. Not a big success, mind you, but not a failure either.

 

To be *really* successful it would have to accomplish the above *and* do one or more of the following: 1) show a good speedup (2x or more); 2) render sunlight through glass.

[bricklyne]

His name is Punxsutawney Phil, and he's a he, not an it

 

 

Sorry Andrew, but I didn't know how to spell "Punxsutawney" , and didn't want to embarass myself (so much for that)/was too lazy to look it up. In any case you know what I mean......

 

It's funny that that's the idiomatic reference that first came to mind, as an example, seeing as this whole Maxwell episode is like one bad and ugly software version of the Groundhog Day movie. All we need to do is have NextLimit play Sonny & Cher's 'I Got You Babe' (as would play on the Phil's Alarm clock in the movie every morning) on their website, everytime they announce a release or everytime they miss a deadline. Would be fittingly appropriate.....

[Ernest Burden III]

All we need to do is have NextLimit play Sonny & Cher's 'I Got You Babe' (as would play on the Phil's Alarm clock in the movie every morning) on their website.....

 

Followed by a cold shower (There isn't any hot water? Not on a Tuesday) Oh, and Bill Murray'c character had an eternal number of chances. But he got it right in the end!

[AJLynn]

It's funny that that's the idiomatic reference that first came to mind, as an example, seeing as this whole Maxwell episode is like one bad and ugly software version of the Groundog Day movie.

 

LOL! Damn that's a scary thought. What would be worse - every day it's the same old, no new release, no news from Victor - or every day a new busted-ass RC version?

[Fran]

FWIW, I would consider RC5 a success if the features that have been steadily removed (clip maps, fog, coop. network rendering, etc.), were reincorporated and the engine got a tune-up.

 

After asking very politely for a refund and being patently ignored, I have no sympathy for them. They get exactly what they deserve.

[Devin Johnston]

Have they removed cooperative rendering as a feature?

[Maximus3D]

Don't get me wrong, i'm not blind to what happend in the past with RC's and everything else but i can say that i do talk alot with several people from the team, several times a week and sometimes also daily via email and i never had a problem with the communication with NL. I know stuff i cannot say as i sworn to keep silent but RC5 will be good and worth waiting for, that's as much as i can say even tho noone here will believe me on that. And i know that Mike V agrees with me on that too even tho he's bound by his NDA to not say anything revealing.

 

I understand that frustration and anger took a big deep bite outta most of you and that it's hard to let go of the past problems the delays and lack of communication caused us all. NL knows about them very well, and why they don't do more than they do about it i can only speculate has something to do with them working so hard and having so few people and time to sort out the problems on the forum with the communication with us users. That's my guess based on my long conversations with NL. Don't quote me on this please.

 

And to those who prefer to call me names like "fanboy" or whatever i don't care, go ahead if it makes you happier. But i know some about developement and people and i judge this situation based on what knowledge i have and not on what other people think about it and me.

 

/ Max

[AJLynn]

I know stuff i cannot say as i sworn to keep silent but RC5 will be good and worth waiting for, that's as much as i can say even tho noone here will believe me on that.

 

I really hope you're right. It's just that after th last 2 months I have a hard time believing anything that involves M~R development being reality-based. But I'll wait and see.

 

What I really have a hard time with is, if they had the thing working well enough to do those renders they showed before the RC release, what the heck happened?