TPA: First V1.0 WIP and tests

[Adrian Brown]

Hey Guys, well now that I'm finally posting on these forums (it only took me 3+ year to finally do so) I thought I'd let you know that I've posted a couple trial run pics with 1.0 over on the Maxwell forum.

 

I gotta say things are looking good with it so far. Anyhow heres the link (if their server is up that is). Going to let in render the rest of the night and maybe I can really start tweaking it tomorrow.

 

http://www.maxwellrender.com/forum/viewtopic.php?p=144316#144316

[Adrian Brown]

Since it's kind of hit or miss over there on Maxwell forums (with the current server loads), maybe I should just post the image here too.

and here a Vray one for reference:

http://www.thepixelartist.com/c2003002/p2006011_files/20060421_finals/p2006001-final1-view1-2000wm.jpg

[CHE]

Hi Adrian,

 

Nice test, thanks for sharing. Did you use the new AGS material for the glass? What was the render time?

 

Thanks,

Ernesto

[TRANSPARENT]

I also am wondering about the render time. Also I will wait for the final rendr, but as of now the Vray render looks better. Lets see some more.

[Devin Johnston]

Yes, tell us how long it rendered for, what the resolution is, and if you used AGS. I don't see any reflections in the glass.

[CHE]

I am runing some tests right now and so far I have nothing good to say about V1. I have not been able to render something good enough to post it here. Render times are long, maybe longer than Beta. The quick preview in 3dsmax takes twice as longer than with the Beta plugin.

[Adrian Brown]

Good morning gang,

Well I let another one (with a few tweaks) render last night at a little higher res (1600x1000) for 6 hours @~12 samples (that first render took 30mins with a little PS de-noising). The first image is the raw output and then the other is with about 15mins of Pshopping to tweak the colors, slap on a quick background (haven't figured how to do it yet in Maxwell, definitely not intuitive), and a little perspective correcting (again, another big missing feature in maxwell as far as us arch renderers are concerned). But I wanted to get a feel for how finishing might be with a Maxwell render (compared to Vray). This geometry is straight out of Sketchup BTW (and there are some issues there too).

 

OK issues so far. Obviously there is supposed to be glass in those big windows. And there is, but it's not showing up (it is physically there in studio). There is actually two single poly thin planes of glass there, both with the same AGS mat that is on the artwork frames (which look great). Also its missing on the picture on the right, and its exactly like to frame on the left except for a different picture in the frame(???). I am a little disappointed that after 6hours at this res (it would take 45mins in Vray) there still a lot of noise on the interior (as least to me), especially since there is no glass in the outside frames, I'd imagine that there would be more if the glass was showing up. There a few other issues but I'll post more about them on the Maxwell Forum.

 

Original Maxwell Output:

http://www.thepixelartist.com/images/maxwell_tests/dodsen/20060426-maxwell_test3-dodson.jpg

 

Photoshop Edited:

 

 

Hi Ernesto, yes I did use the AGS mat first one as well.

 

Hey TRANSPARENT, thanks. Yeah, heres the thing, I don't need Maxwell for it's realism, as I can get Vray to produce as much realism as I need. I was originally hoping Maxwell would make my life easier. But thats a long ways from happening yet.

 

Here a quick one the Vray a did last week, again I don't need any more realism than this.

 

[Adehus]

Thanks for the tests, Adrian... Everything's looking good except for (surprise, surprise) the glass.

[Adrian Brown]

Well.... OK, here a strange one. I got glass issue fixed, but not sure why. I reimported just the window glass, reapplied the AGS mat and it looks to work now. Oh, and the picture frame glass one was my fault though, it really was just missing it's glass (was on a wrong hidden layer for some reason).

 

Heres a quick Studio preview image. So later when I get some time I'll do some more tweaking of the mats, do some more finishing to this, and let it rerender.

 

[Devin Johnston]

I've been messing around with it this morning trying to get Network rendering set up. I'm very disappointed that there isn't a simpler setup, there is to much manual inputting and starting up of things. This is something that has to be simplified and automated like 3D Studio Max backburner has been.

[Adrian Brown]

Maxer: Yeah, as I've said before, right now Maxwell has a lot of potential (always has), but it kind of take two steps forword in render quality and some parts of the workflow but then take about 6 step backward in really production workflow and other standard features. It's too bad too. But I guess it's getting closer.

 

Ok, one more quick update with the glass working now (I've really gotta to get some work done today). I let in render for 30min at the same res as the test 2 one I did last night. I wanted to see how the noise resolved behind the glass compared to without it. I'm pretty happy to see that the noise is pretty much the same and best of all there is sunlight and shadows inside. I also think it's obvious that the image realism has really gone up now that the glass is showing.

 

 

Anyhow, there's obviously a lot of tweaking still to do on the materials and setups here in this Maxwell scene (and still a lot of learning its quirks). So I think will be much much improved with some work later.

 

[Devin Johnston]

I'm still not seeing any reflections in the glass.

[Adrian Brown]

Well it's there, but it is a little hard to see yet. If you look towards the bottom of the glass you'll see the ground reflecting. But again, in fairness thats not really Maxwells fault. The problem is that there isn't anything really to be reflected in the windows yet. Only a bit of ground and then to far away even skydome. So I think it would just take a while longer to really start seeing the sky color reflecting on the glass (like it does on the painting glass in the pervious render.

 

 

The Vray one has a reflection screen behind the camera with a photo of the other side of the street scene, which obviously really gives a nice reflection to the glass. I'm going to do the same thing later for the Maxwell one, which should really help.

[yog]

I'm still not seeing any reflections in the glass.

Thats because the "new" and "revolutionary" AGS-preset is nothing more than two ordinary BSDF layers. One allows light to pass through it but has no reflections at all, and the other has reflection enabled, with the already understood problems with caustics passing through them. The weighting between the two layers is 85/15 in favour of the non-reflective layer by default.

 

This means that although you will have a lot of light pass through the glass, you will generally only get around 15% of the reflection value you would normally expect.

Of course you will also lose light refraction as well, something I have found is quite obvious (not in a good way) when it comes to curved glass.

 

Not so much a new material, rather a half assed preset.

[Msamir]

The problem is that there isn't anything really to be reflected in the windows yet.

 

I was just going to say so

[Devin Johnston]

NL wonders never cease

[CHE]

Here is my first interior test straight out of Maxwell. Render time 5hrs at 1500 pixels on a Dual Xeon 3GHz 2Gigs RAM. I applied three standard materials (One diffuse material, on emitter material and one AGS for the glass). All the windows have glass including the one on the main entrance. I couldn't get the multi-light utility to work. I guess there is a limit on the number of emitters you can load to it. Also, I noticed that Maxwell V1 requires more PC resources than Beta. I couldn't get anything to render in my laptop. (P4 3.6 1 Gig RAM).

Sorry for posting my test here but I thought that we could all post our findings and wips here. Let me know otherwise.

 

Ernesto

[Devin Johnston]

Yes someone on the Maxwell forum said the same thing about windows resources, I think it was Max3D.

[Maximus3D]

Sigh yes, sadly enough it requires more horsepower and ram now to work well. I still haven't been able to get anything outta it with one gig ram it just closes down the renderwindow without a warning once i try to render something. It ain't making me too happy, that's for sure. But i seen what other people been able to squeeze outta it and so far it looks really good, but i wish i could make use of it too non of my highpoly detailed models will render at all.

 

/ Max

[3dworks]

Sigh yes, sadly enough it requires more horsepower and ram now to work well. I still haven't been able to get anything outta it with one gig ram it just closes down the renderwindow without a warning once i try to render something. It ain't making me too happy, that's for sure. But i seen what other people been able to squeeze outta it and so far it looks really good, but i wish i could make use of it too non of my highpoly detailed models will render at all.

/ Max

 

one good thing is always to save an MXS file from studio (or with help of the plugin) and run MXCL from commandline in RAM critical situations. don't use the -d option if RAM is really tight and you don't need the other functions. also, be careful with the glare processing and the multilight function, both take away lot's of resources. as it is for release 1.0 glare should be used only AFTER the rendering process stopped, if not MXCL may crash.

 

hope this helps

 

markus

[3dworks]

Thats because the "new" and "revolutionary" AGS-preset is nothing more than two ordinary BSDF layers. One allows light to pass through it but has no reflections at all, and the other has reflection enabled, with the already understood problems with caustics passing through them. The weighting between the two layers is 85/15 in favour of the non-reflective layer by default.

 

This means that although you will have a lot of light pass through the glass, you will generally only get around 15% of the reflection value you would normally expect.

Of course you will also lose light refraction as well, something I have found is quite obvious (not in a good way) when it comes to curved glass.

 

Not so much a new material, rather a half assed preset.

 

simply add a third layer as a 'reflection enhancer' if you need more reflections!

 

markus

[yog]

simply add a third layer as a 'reflection enhancer' if you need more reflections!

markus

Apart from the fact that it is the IOR that is stopping caustics appearing through glass (i.e. the original problem).

[3dworks]

Apart from the fact that it is the IOR that is stopping caustics appearing through glass (i.e. the original problem).

 

you will have to set the Nd value much higher for this extra layer, for compensation. if you use a value like 50 you will get pretty strong reflections even with a 5% layer weighting. it's well worth to experiment a bit with this...

[CHE]

I thought I post this here.

 

mverta wrote: [Ok, first of all. I'd begin by using .mxm's for EVERYTHING. I mean, EVERYTHING in the scene.

 

Second, those emitter .mxms should begin as just pure emitters. You create a fresh material in MXED and add an emitter layer.

 

Third, to test the emitters are working, set the Watts to like, 100,000 with an efficiency of 500. This is like putting the freaking sun in your room, so you should definitely see some action there.

 

Assuming that's working, dial the emitters back down. Maybe start with 1000W and 100 efficiency, just to make sure there aren't some enormous other errors with your scene. Remember, you can always dial the settings into "real world" later, but there's no point in doing that if you're trying to hunt down potentially giant bugs.

 

USE MXM's FOR ALL YOUR MATERIALS. And make sure ALL of your solid objects have at LEAST ND 3. Begin without bump maps.. disable them.. just dial the scene down to its basic illumination components.

 

This stuff works -so have confidence, but begin the debugging process by simplifying.

 

_Mike]

[_PopArt]

AGS is a cheap work around, but i guess we should be happy to have the solution they offer. To the Pixel Artist, thanks for the posts, very nice work. But i think you posted the same on maxwell forum with a Vray render, which looked much better. This is the way maxwell is IMO - in some situations unbeatable, in others just a big PITA