maxwell Render - first facts and screenshots

[Mudvin]

Vlado, there's obvious light calculation errors in the corners of ceiling cubes in your image.

[Mudvin]

the lighting quality is great and the noise problem will sure be solved.
this is still a alpha version.

I don't think that the noise problem can be solved. It's obviously the signs of undersampling, which itself in turn is consequence of that feature with fixed maximum render time. The only solution is to use some interpolation which also already used almost in every modern GI renderer, and can't be used in maxwell, because they said they will not.

[finish]

olivierfR, you look at the problem from another point of view.

i said maxwell can't be compared to vray, that's all. if you absolutely want to compare, you can of course. what i said about vlado's post still applies. to compare he should have used direct computation secondary bounces because it's the nearest method to maxwell's one.

i know vray can do much faster job and achieve, at least, same quality. that's not the prob here. and i don't want to remember speed and quality there is so much more about a renderer than its speed and quality, but as a professionnal you konw that very well, isn't it olivierfR ?

you and others claim you try to be professionnal in your critics, then accept that a slower renderer can fit needs that a faster one can't in some kind of jobs.

and i don't think vlado made a good point here, he just demonstrated that with lightmap with QMC vray can render a noisy image much faster than maxwell

[RErender]

Isn't the point that a brute force aproach to GI rendering is arcane and generally a big waste of time? And as far as Maxwells "inherent accuracy" In my tests with this scene and just skylight the results are nearly indistinguishable from the maxwell render nomatter what methods I use in V-Ray, the highlights, shadows and ambient light are all in the right places...

For light studies I would suggest Lichtect's solution with graphing and false color it is a powerful tool.

[olivierfR]

Well, That's endless discussion ,until we can't test it on full version...at least the one just after the alpha...

Also, anyway waiting for a DEMO !!! ... things could be a bit more clear for everybody...

btw, finish did you try the available version of Maxwell ? if yes, could you share your tests ?

For light studies I would suggest Lichtect's solution with graphing and false color it is a powerful tool.

Also, Radiance....well the "desktop Radiance" for Autocad is also super nice...I think the same method as Vray, mmm, Irradiance map...
and it's free !

http://radsite.lbl.gov/deskrad/


also undless discussion...

[finish]

brute-force is the mother of all optimizations. respect you parents ! :P
steve blackmon from splutterfish said : "premature optimiszation is the root of all evils". that's something to consider too.

+ as i said, some people (not me) still needs absolutely unbiased results. this kind of people often need to write their own programs because there is nothing on the market (or because they want to). again having the possibility to buy one is great, just because it's now possible.

and finally i prefer to see loads of super-expensive computers *wasting* their time on creating astonishing images, rather than computing the next ideal millitary target (which is porbably done in a brute-force manner too ) just look at debevec work, the programs he used in the past (radiance) and the one he use now (arnold) are much more similar to maxwel than vray or brazil or any others. just because it is what he actually need/want (i assume). simple isn't it ?

[olivierfR]

The next century will be direct compution or not...

I wish i could use this direct computation everyday day

[olivierfR]

btw, is there any kind of relation between Arnold and Maxwell ( i don't talk about their direct computation)...like the Beta tester are all spanish...both with Arnold and Maxwell... Beta tester Friends ??? Let's move to spain and find out !!

[mildse7en]

Hi vlado, regarding to your 24min render pic post above, what is that render machine's configure? I mean, what cpu, how much ram?

Thanks!

[flino2004]

I am architect and the most part of the time I need renderings for clients and marketing stuff but I'd like to use engines like Maxwell as a design tool (Physically accurate illumination)....Vray,fR,Brazil and MR are very similar at the end. Maxwell is a different animal....maybe it is a hybrid between (radiosity and GI).

[cpnichols]

There are a few point to make.... one for Vary, and one for Maxwell

For Vray:

When my supervisor tells me. That corner over there looks dark, or that reflection should not look like that. I want to see more "bling" in the fresnel (whatever the neck that means), or here is a good one. I want to make it "feel" blue, but not "look" blue... I will have better chance to that with Vray, because I can do it, and I can do it faster.

For Maxwell:

When I'm doing a PHD in Physics focusing on optics and professor tells me, I don't care if it look bad, is it correct? Well, I don't know enough about Maxwell, but can I really say that about Vray right now?

[RErender]

When I'm doing a PHD in Physics focusing on optics and professor tells me, I don't care if it look bad, is it correct? Well, I don't know enough about Maxwell, but can I really say that about Vray right now?

Looks pretty good http://www.chaoticdimension.com/foru...ghlight=tilted

[vlado]

Well, I don't know enough about Maxwell, but can I really say that about Vray right now?

Yes, to a certain extent. Lighting and GI in VRay are physically correct, at least as far as diffuse GI is concerned. Hilights and glossy reflections are also correct with the default settings. Of course, if you use an approximate GI solution, there will be errors because of the approximation. By default, VRay also cuts some corners to make things faster (GI caustics), but one can always turn this off.

Best regards,
Vlado

[cpnichols]

Well... "touche" vlado and rerender... But I was referring to the wavelength based GI (whatever that is). Meaning that RGB considerations of light (HDR or not) are not an accurate measure of light energy, according to Maxwell. I have no idea it will it have a true effect or not.

On the other hand, I remember back around 1991/1992, We used a program called POVray (yeap it was around back then) to help us figure out our optics in the physics lab.

Vray would do a great job of that. A friend of mine that build telescopes wants to try to simulate his telescope in Vray. The only think it will not be able to do is chromatic aberrations.

[plastic_]

the only thing that seems special to me with maxwell is the sky setup thing. it looks great. i'm not very concerned about real locations and daytimes, but some procedural auto-sky generator would be handy for vray.