[Petero]

Hello,

I am using an AGS glass for rendering which include Glas cases. this test render is rough(noise) probably because of too low SL (13.66), but my real problem is that

1.)the glass seems to be casting very dark shadows, and

2.) the light doesn't seem to be falling on the tops of the display shelves

This could be an important consideration:the displays are of THREE materials- the GLASS, and a MATTE and SEMI GLOSS Gray matallic material. The Gloss if for most of the Matal you see, EXCEPT the shelf piece which is matte.

here is my Glass composition in 2 BSDFs:





the test render:



thanks for any comments, ideas.

Peter

[Half Life]

Why not just use the new "AGS Clear" from the MXED wizard -- that looks like an old Maxwell 1.X formula...

Best,
Jason.

[Petero]

Thanks Jason,

I am testing that now. In the context of my current renderings, does this post supply options in your opinion? It seems like Tim Ellis has looke dclosely at glass options, but i am not sure how to interpret his conclusions?

http://www.maxwellrender.com/forum/view ... 598#243598

Does he imply that actually putting a piece of of Glass mxm is a good idea in your opinion?

He implies that 'mixing' AGS and 'Glass' Glass is a good idea?

Of course anyone's ideas on this are welcome!

thx-

Peter

[brodie_geers]

Those tests were from 1.x so I wouldn't trust them for 2.x. At any rate there are 2 advantages to real glass, caustics and refraction. You probably don't need caustics for what you're doing and you'll only get accurate refraction with appropriate glass thickness. I'd suggest starting with either real glass or AGS from the wizard and if you can't get the result you want maybe doing some tests to see if any of that old mixing business still applies.

1/2 life understands materials better than myself but I'm not even sure what you have there is 1/2 AGS. In 2.x AGS basically just involves lowering the layer opacity which you don't have there. If you were to mix them you'd probably need to use multiple layers rather than BSDF's. Could be mistaken on that however, and I've put a great deal of work into forgetting most of my 1.x knowledge.

-Brodie

[Half Life]

Tim is a very smart guy but that is an older version (with older material parameters) as Brodie says, which does invalidate the data for the most part since modern AGS is nothing like classic AGS -- especially the clear formula.

Real Glass always looks better IMO which is why I hope Maxwell gets to the place where AGS is no longer necessary ... that said I did some side-by-side tests of my own (with the newest version of Maxwell) for the Sketchup to Maxwell series and came to the conclusion that the "AGS clear" formula is closest to real glass visually (in both transmission and reflection).

Glass will look better but you will need to render to SL 35 (or so) for it to render clear with full detail (see this thread http://www.maxwellrender.com/forum/view ... 25#p340525 ) -- since most are not as crazy about these things as me, I always suggest AGS clear

I've heard many tricks to try to use glass and still get accurate lighting (like hide to GI), but my thoughts are the second you modify how the material renders you lessen the realism that Maxwell can/should produce... at that point you might as well use AGS clear and be done with it.

Best,
Jason.

[fv]

I find the results rather artificial with AGS. You might try this.

Edit your glass geometry in to seperate surfaces for the edges and the rest. Hide everything but the edges of the glass from GI. Then assign all with high or low grade glass.
The rendered results will just be perfect in most cases, without much extra rendertime. I do all my architectural glass like that. If you also hide the edges of the glass from GI its ok as well but you will loose the typical glass look at the edges. And the edges really do make a difference in the overall effect.

It may seem a lot of work but I have made this part of my workflow and it hardly adds to the total time spent on a job.
Another tip, use the special preview for glass while editing the attenuation and transmittance. That saved me lots of time as well.
Here's the link. Tried to find the glass or sss preview scene with a stairstepped element from 0 to 10 cm. Can't find the link anyomre. Maybe someone knows a link to download it again.

Francois

[brodie_geers]

Francois, do you have an example of how your method renders out?

-Brodie

[Petero]

Hi François,

thanks for that technique. a few questions before i try it:

What do you consider 'high' and low' quality glass? Are they MXMs you have made? or did you download them?

So if i understand you are suggesting to work with surface objects only?

Doesn't glass need to have a thickness( be a solid) to work with the attenuation setting?

Thanks-

Peter

[brodie_geers]

I believe fv was referring to the high quality and low quality glass settings that you get when going through the material wizard.

I don't think glass has to be a solid like SSS but I could be wrong.

-Brodie

[Half Life]

SSS is just glass( or any dielectric) with scattering -- they both need to be solids to render correctly.

I'm not opposed to any workarounds using various strategies, but I do feel compelled to point out that none of them will be technically "correct"... the refraction cannot by definition be correct if all parts are not glass and assembled in the same way as they would be in life -- and that being the defining characteristic of glass vs AGS, at that point why not use AGS.

Simply put, edge refraction will not be correct if surface refraction is not also taking place -- now you may say "having something going on there is better than nothing"... but my reply is since at that point the render data is already false why not simply add fake refraction in post and opt for the faster/cleaner render.

And this assumes you even need it -- often times the edges are anisotropic and show minimal refraction anyway... and you can certainly do roughness (and anisotropy) with AGS Clear, see this tut:

http://www.maxwellrender.com/forum/view ... 84#p338384

Best,
Jason.

[Petero]

Her's a test following Jason's advice to try the Wizard for Clear glass at 12.5. SL of 18.25 .

I probaly should tweak the color , and maybe make it more reflective.








Here is the MXM- i noticed the Wizard created an attenuation of 1 nm on the top bsdf...does this mean it is an SSS type material?




OK I went back into the wizard and tested the use of the"AGS" choice..

- when i make clear glass in the wizard, it gives me a "Ghost" and "Specular" layer, with the ghost layer having an attenuation of 1 nm. no mention of "AGS"

-when I use the 'Normal' setting in the wizard I get only one layer called "AGS", with 'Attenuation' off.

1.) Are both considered "AGS"?
2.) Are both faster to render than "real glass"?
3.) What is the difference between "Normal " and "Clear" ?
4.) there are very few 2.0 versions of MXMs of glass in the forum that i can find. and when i open some from 1.x version, i am pretty sure they need rebuilding. If I have not yet the proficiency to tweak these complex settings, where can I go to get 2.x version 'Real Glass?'

5.) Does something like a Real Glass Template exist?

Thank you!!

Peter

[Bubbaloo]

SSS is just glass( or any dielectric) with scattering -- they both need to be solids to render correctly.

True for SSS, but glass only needs an entry and exit normal. Although visible holes in the glass (where refraction would be calculated as infinite) might render strange refractions, it would not be detrimental to the render engine like a non water-tight mesh would be for SSS.

Here is a render I did quite a while ago in 1.x using AGS on the top and bottom surfaces and low wizard glass on the edges (multi-material):

[Half Life]

The ghost layer is an additive mode layer that only exists to pass its transparency to the layer below -- increasing it's opacity will increase the "invisibility" of the material, whereas decreasing the opacity of the ghost layer will make the AGS more "solid".

The attenuation does not matter because any non-zero attenuation value will be full transparency due to pure white transmittance... to see what I mean try this:

1) Set the ghost BSDFs Transmittance value to 1 and test render it on the leaf preview scene.
2) Now set it for 254 and re-preview.

The results are identical -- the only value that gives the full transparency is 255. No refraction or reflection occurs because the Nd is set to 1.00 (vacuum)...

So you can see the ghost is simply an infinitely transparent (or invisible if you wish) layer that is passing it's "invisibility" to the layer below through the layer opacity percentage.

AGS is faster to render because it is basically a metallic object with the opacity turned down -- glass (or any dielectric/SSS material) takes longer because the refraction and particularly the refracted caustics take a lot of rendering power and time to chew through and will generally cause noise.

For what it's worth Brian I have rendered many hundreds of tests with SSS with open volumes with no issues worth mentioning, but I also know in those cases the results are not fully realistic (exactly the same issue with glass)... thinSSS (particularly ghost thinSSS) is often a more useful option (but there are instances where it totally breaks plausibility, much like AGS).

Generally my reason for discouraging that "cheat" is because I think it is unnecessary -- it adds render time and noise without adding true accuracy... and since there is little accuracy those edges could be painted in post in a few minutes instead of waiting extra hours for the (already inaccurate) render. Don't get me wrong, I think something needs to be there for plausibility -- I'm just not convinced the inaccurate render looks any better than a quick 10 minute paint job in Photoshop would... and the render cost is much higher than AGS.

To be clear I hate the idea of AGS -- I would love nothing better than to see it banished forever from Maxwell... Glass is far superior in every way visually, but when efficiency is the issue then AGS is the better option (right now).

Best,
Jason.

[brodie_geers]

I'd never really realized there were two options with the AGS wizard (normal and clear). I see the difference in how they're created but I'm unclear as to what the advantage/difference is between the two.

Brodie

[Petero]

Brian,

I will try to make a test which reproduces what your render shows.

thanks!

Peter