New Maxwell version

[dave davidson]

well i cant use old v1 scenes in c4d its always complaining about the uvs which i have no idea wat so ever how to solve and even on a new scene with custom maxwell tags it still gives the same error and i still have no idea what to do to solve it.

im still trying it to see what the crack is so we/ me will see

oh and the camera selction thing is doing my head in. i have a camera with a maxwell tag and it automatically goes to the viewport that is active, this has got me loads of times cause im not use to it doing it

 

oh and for the atackers of complainers /whinners ( that hide in the shadows) i am mearly stating what i have on thier forum and stating the obvious workflow killers, etc. i use i say i use some more its that simple and they learn from the user base saying stuff when they use it OK

[pixelperfectg]

Well, it looks like I was partly to blame for the frustration in my previous post. Apparently I overlooked one or two of the .jpg's projector numbers while I was trying to change the arroway texture collection materials.

After reading the feedback on the Maxwell forum, I went back and made sure I had all the projector maps set to 1.0 and it renders fine now. But it's still a shame I have to go through the material sets and modify all the projector numbers when bitmaps are involved.

Hopefully they will quickly release a patch to fix this...and not mess up all the materials I've already changed & saved...

[danatgia]

BTW, here is a link to that piccie for those who can't access the forum:

http://surrealstructures.com/maxwell/porch04.jpg

Very nice image but does that Say Render Time 21hours 20 Minutes????? I thought it was supposed to be faster?

Dan

[Fran]

Very nice image but does that Say Render Time 21hours 20 Minutes????? I thought it was supposed to be faster?

Dan

 

What was supposed to be faster?

[Devin Johnston]

Faster in Maxwell means a 5% or 10% improvement in the clearing of noise, if your expecting a 2 or 3 fold increase your dreaming. NL has even said that they have two solutions for dealing with the speed issue, one is cooperative rendering and the other is waiting for hardware to catch-up with the software.

[Fran]

Yes, but did he mean V1.0 versus V1.1? My image was rendered with V1.0, which is stated on the image. Here is a link that shows comparisons. Anyone can look at it, whether they are registered or not.

 

http://www.maxwellrender.com/forum/viewtopic.php?t=17120

 

Oops

 

People can draw their own conclusions. Whether or not this amounts to significant improvement for complex architectural scenes remains to be seen.

[Devin Johnston]

Just so you know if your planning on using V1.1 with any model you made in an older version your going to have to recreate all your materials. Also you are still unable to view materials applied to geometry in Max.

[Ricardo Eloy]

Hehe...so they're saying something like "we released a PS3 game, even though you're all Atari 2600 owners, so wait another 10 years"? Clever.

[Fran]

Just so you know if your planning on using V1.1 with any model you made in an older version your going to have to recreate all your materials. Also you are still unable to view materials applied to geometry in Max.

 

It is true about legacy materials. V1 materials have the problem of the default projector = 0 thing, which the max plug-in has a problem with. I expect that they will be updating the max plug-in soon (?). There are several threads on the forum dealing with this issue, and it is very easy to get the information about it confused.

 

You can view your mxm in the viewport if it has a bitmap. You open the material and click on the eye at the top of the mxed window and select the bitmap you want visible for that material. Then you refresh the material in the max mat editor.

[Ernest Burden III]

Hehe...so they're saying something like "we released a PS3 game, even though you're all Atari 2600 owners, so wait another 10 years"? Clever.

 

When you get tired of blaming your customers, blame their computers.

 

I hope to get some time to try the C4D version, though AdamT has already found some new problems with the plugin.

[Devin Johnston]

Thanks Fran, you saved me a bunch of time with that one!

[Bongo51]

Faster in Maxwell means a 5% or 10% improvement in the clearing of noise, if your expecting a 2 or 3 fold increase your dreaming. NL has even said that they have two solutions for dealing with the speed issue, one is cooperative rendering and the other is waiting for hardware to catch-up with the software.

 

LOL...

 

Our 'Life simulator' needs to run on a planetsized computer, where the entire program IS the planet... what? You don't have a planetcomputer? Well then, I guess your really not cut out for ownership ARE YOU! So quit your whining!

 

PS... isn't that the 'greatest american hero' logo?

[Devin Johnston]

;-)

 

Yep, the Greatest American Hero rules!!

[Guest imanobody]

In all fairness to Maxwell - MTL is just a slow process on to itself and nobody has been able to make it faster. Even when the hardware reaches a point that you can do a nice interior rendering in 1 hour, that only means that all the other renderer's can do it in 1 minute, so it will still be slow no matter what. Anyway, I don't think that hardware is increasing at the pace it did in the 90's; I've had my computer for nearly 2 years and it's still pretty much top of the line, but it would be obsolete by now if it was 1995.

[Bongo51]

imanobody,

 

I think most of us understand that true light simulation isn't a walk in the park mathmatically... the problem lies in the fact that there is noise that NEVER goes away. a 1000 hour render would still have noise, and THAT is unacceptable for any renderer... IMO.

 

I can understand some noise, but to have really bad noise, and only in certain spots it's makes it obvious and unusable.

 

So, when I complain about maxwell being slow, it's only slow to remove enough noise to make the product actually usable. Most 'clean' renders of architecture we have seen have been rendered larger, and cleaned up in post. I've had many renders which just wouldn't clear up. So, if NL needs to introduce SOME biased filtering... at least something we can turn on for shorter renders, this would eliminate a large percentage of the casual complaining. If after 10 hours my render is clean.. even if it comprimises total accuracy a little bit... I don't care. A hybrid is really the ticket here... it's obvious that NL is having trouble making a true unbiased renderer that is not glaringly noisy. Many serious issues have already been fixed by cheating, window glass being one.

 

why not have built in noise filtration? That more then anything would make Maxwell appear faster. I've had many renders that looked great, but I kept waiting because a section of glass, or a light was 1000% noisier then the rest. If some noise control and filtration were active I'd guess my renders would need to cook for 50% less time! Isn't that all that vRay and the others (simplified) are doing... they aren't faster, the just cheat better.

 

Ian

[Adehus]

why not have built in noise filtration? That more then anything would make Maxwell appear faster. I've had many renders that looked great, but I kept waiting because a section of glass, or a light was 1000% noisier then the rest. If some noise control and filtration were active I'd guess my renders would need to cook for 50% less time! Isn't that all that vRay and the others (simplified) are doing... they aren't faster, the just cheat better.

Ian

 

Noise filtration, interpolation, whatever the proper technique may be, it just seems to make good sense. Since Maxwell has an awareness of polygons/edges, that would be the place to do it (vs photoshop, etc.)

 

Related to that... a question for those in the know about rendering technology. Am I right in thinking that most radiosity/GI apps sample in a manner similar to Maxwell to some degree, and then just make educated guesses about tone/color for adjacent pixels from that starting point? (I know, I think I'm asking the same question that Ian asked)

[Guest imanobody]

But I think the noise is a common thing in MTL. Indigo has the same noise as the Beta and so does Mental Ray's path tracer, so I think it's a safe bet that the fault is in the process. After seeing the renderings done with V1 I tend to think that they have already added a biased solution into the renderer due to the "cg" look of the images. Some people have made good images, but that's after they spent hours screwing around with the materials, but wasn't Maxwell supposed to get them away from that. If they start adding more biased solutions then it will turn into nothing more then another raytracer with MTL. With biasness (is that even a word) comes all the problems that everyone who bought Maxwell was trying to get away from.

[Bongo51]

here's a thought, perhaps reaching the true solution, math that truly simulates our natural world utilized to generate photographs is either a much larger project then Next Limit imagined, or it's something beyond our current math/computer skills?

I mean, think about what is being simulated, it's complex beyond what we fully and currently understand correct? There are aspects of light's behaviour which we really are only guessing about, so who's to say until these issues are understood, that computer math will ever get it right.

imanobody,

I understand what you are saying, however, the way maxwell handles light in all it's activities is the ideal, I still think by moving the target just a weeee bit toward bias... just so there are not GLARING noise patches in the render... at least this choice should be available as a render option... just like in fPrime rendering in only ray tracing can come in handy. As for what maxwell is trying to get away from? I don't think even Next Limit knows what direction they are going. Maxwell started as simple and the most elegant plugin, now it's a massive monster, which yes, is far away from what it was originally 'supposed' to be. But they were 'supposed' to have several other plugins but they have mysteriously vanished without comment.

[Devin Johnston]

The way I see it is that we've already moved so far from where we were supposed to ever be that adding a biased option wouldn’t be that much of a stretch. The idea of unbiased rendering is a whole lot more attractive than the actual process it's self. If NL knew at the beginning what they know now they never would have taken that approach. The concept of never having a finished rendering is ridiculous and it screams out for another solution. Noise filtering is something that is seriously needed, but that should come after they fix all of the current problems. Getting V1.1 to look like the beta did is the highest priority for me, if they can't do that then all the other stuff is a waste of time.

[Ernest Burden III]

If NL knew at the beginning what they know now they never would have taken that approach.

 

I'm not so sure. Unbiased was not so much a practical consideration as a mantra. And the best part of arrogance is never having to say you're sorry.

 

 

The concept of never having a finished rendering is ridiculous and it screams out for another solution.

 

Nature provides perfect renders in realtime. Yet even a stripped-down reductive simulation can take forever and a day. That suggests that embracing reduction is a good idea, which means either leaving things out or accepting a biased solution that looks right but maybe isn't.

 

I had hoped NextLimit would address their position on 'unbiased at any cost' regarding MWR v1, but I don't think they have. Does it actually matter to anyone beyond them?

 

Something I've learned early and often in art is that it can be quite difficult to match a great sketch. Beta was a great sketch.

[Devin Johnston]

That's why it makes since to have the option available for those that don't mind the biased way of rendering. I'd bet that if it was available 95% of the people would use it and if it were done right you probably would have a hard time telling the difference.

[Ernest Burden III]

That's why it makes since to have the option available for those that don't mind the biased way of rendering. I'd bet that if it was available 95% of the people would use it and if it were done right you probably would have a hard time telling the difference.

 

I agree completely.

 

Now go post that on the MWR forum. Maybe post a poll, could be illuminating.

[Guest imanobody]

Despite being a big critic of NL, I've kind of understood why they took the 100% unbiased approach. I definitely believe that the future of rendering will be 100% unbiased, but when will that be... 5... 10 years? Who knows. But after seeing the Beta images, this is what I believe.... but I also believe that eventually we'll only have one material (SSS) to work with, because, in theory, that's basically what every material is. I wouldn't put much weight in my predictions though.

I think that maybe NL thought the same thing and wanted to usher in the unbiased age, but I highly doubt that's going to happen now; I'd be surprised if they lasted another two years. I do think that that another company (please let it be Mental Ray) will pick up the idea and run with it later when it's more viable option.

[AdamT]

I think NL is going to have to introduce some kind of sample interpolation to make Maxwell practical for interior renderings. Noise reduction is okay, but kills texture detail.

 

Anyway, they really can't claim to be unbiased, what with AGS glass, hidden objects, the ability to turn caustics/dispersion on and off, etc. I like that you can do almost-unbiased renders with M~R, but the user should have the ability to decide how much of that he or she can take.

[Devin Johnston]

I agree completely.

Now go post that on the MWR forum. Maybe post a poll, could be illuminating.

I'd do it but we both know that half the people would think it was a good idea, the other half would think it's blasphemous and Mike and Tom would tell me that I need to become a better artist and learn the material editor better.