CGTalk > Main > General Discussion
Login register
reply share thread « Previous Thread | Next Thread »
 
Thread Tools Search this Thread Display Modes
Old 07-19-2010, 12:27 AM   #1
Bubbaloo
Rendered Speechless
 
Bubbaloo's Avatar
portfolio
Brian Looney
Owner
Lunatic Studio
Sand Springs, USA
 
Join Date: Dec 2006
Posts: 958
Frosted Glass Challenge (Any Render Engine)

Over on the Maxwell forum, there has been a challenge issued to all other render engines, biased and unbiased, to create a realistic frosted glass material and render it on the supplied scene. Please have a read below, and if you care to, submit an example from your render engine of choice so we can fill the empty slots:

Tom wrote: A few days ago, I've revisited H.Jensen's frosted glass dragon scene with Maxwell Render V2 and
it turned out pretty satisfying in terms of quality I've expected. As you see, the scene is very simple
and it's only demonstrating rough dielectrics. The subject is not very challenging for most of the biased engines
but, it could be a real pain when it comes to unbiased cores unless they are not properly optimized/implemented,
especially for interactive and realtime rendering as their BSDF and microsurface approach is playing a huge role.

Here's the original work done by Henrik Wann Jensen (2003):
http://graphics.ucsd.edu/~henrik/images/raytrace.html




Below is the frosted glass dragon rendered with Maxwell Render V2:




So, I thought I should share the scene and you may give it a try with any other engine you have access.
Of course, this is a quality/ability test rather than being a speed-challenge.

The scene has a blue studio ground, a dragon, a small emitter plane and a dim skydome as ambient to match the original reference.
But, you can use either of them or both or more when the features of the engine don't match (only emitter, only skydome or only IBL or all).

Here's the scene OBJ files (all pivots at zero, no further relocation needed):
http://www.fileden.com/files/2009/5...n_scene_OBJ.rar

Resolution: 800 x 640
Lens: 85 mm

You may also try FBX file instead:
http://www.fileden.com/files/2009/5...193//dragon.rar


There are 3 goals in doing this:
1) Highly rough transmissive surface achieving blurry refractions.
2) Full caustics and TIR including the caustics on its own shadow.
3) Clear glass material core with no SSS/translucency or other effects is essential.

I want to leave this test open to any kind of render engine for comparing the quality of outcomes to Maxwell Render.
This is not a duel, so my respect goes to the ones which can at least succeed and demonstrate its quality in comparison.
And of course it's very sad for the ones which may fail awfully. Let's see... Awaiting for your renders...


COPYRIGHTS:
Dragon model by Stanford University Computer Graphics Laboratory.
Reference image rendering by Henrik Wann Jensen.



-Edit: Building the table of entries... Below is the current state:


Full Size:
http://www.atillaakin.com/maxwell/dragonmatrix.png
__________________
Brian Looney

lunatic-studio.com

Last edited by Bubbaloo : 07-22-2010 at 12:25 AM.
 
Old 07-19-2010, 01:07 AM   #2
CHRiTTeR
On the run!
 
CHRiTTeR's Avatar
Chris
Graphic designer extraordinaire
Belgium
 
Join Date: Feb 2002
Posts: 4,372
Just a few questions who might come handy for those participating...

What is the exact rgb tint of the blue ground?
What ior is used for the glass?
What color and power does the ambient light have?
What color and power does the emitter plane have?


Any chance of sharing the maxwell scene?

Last edited by CHRiTTeR : 07-19-2010 at 06:08 AM.
 
Old 07-19-2010, 12:37 PM   #3
Bubbaloo
Rendered Speechless
 
Bubbaloo's Avatar
portfolio
Brian Looney
Owner
Lunatic Studio
Sand Springs, USA
 
Join Date: Dec 2006
Posts: 958
Hello,

Here is the Maxwell file:
http://www.fileden.com/files/2009/5...sted_dragon.rar

Background:
R=20
G=20
B=120

Glass IOR = 1.51

Ambient Light:
Sky Dome
Color 255
10000 cd/m^2

Emitter Plane:
Color=255
Power=300 watts
Efficacy=125 lm/w
Output=37500 lm

Camera Settings:
Focal Length=85mm
Shutter=1/90
F-Stop=8
ISO=300
EV Number=12.492

These are the initial settings. MultiLight may have been used to fine tune the final output. The scene was set up as simply as possible so as to easily compare results of rough glass between the various render engines. It should be sufficient to start with these initial settings and make minor adjustments to lighting levels to approximate overall scene lighting.

Good luck! Hope to see all engines' results!!
__________________
Brian Looney

lunatic-studio.com
 
Old 07-19-2010, 01:17 PM   #4
mister3d
Expert
 
mister3d's Avatar
portfolio
asdasd adsasd
Kiev, Ukraine
 
Join Date: Nov 2004
Posts: 6,031
Yeah, vray looks more realistic. Thank you, now I will use vray.
 
Old 07-19-2010, 02:00 PM   #5
Bubbaloo
Rendered Speechless
 
Bubbaloo's Avatar
portfolio
Brian Looney
Owner
Lunatic Studio
Sand Springs, USA
 
Join Date: Dec 2006
Posts: 958
Quote:
Originally Posted by mister3d
Yeah, vray looks more realistic. Thank you, now I will use vray.


I'm pleased that you found this render engine comparison helpful in your decision making process, although so far only 3 engines have been represented.
__________________
Brian Looney

lunatic-studio.com
 
Old 07-20-2010, 08:29 PM   #6
Grolicus
Veteran
Luc GORIS
Belgium
 
Join Date: Oct 2004
Posts: 92
Interesting challenge, I was looking forward to this thread .
Too bad the entries seem to stall.
Anyone care to step in ? (Luxrender anyone?)
 
Old 07-20-2010, 11:41 PM   #7
Bubbaloo
Rendered Speechless
 
Bubbaloo's Avatar
portfolio
Brian Looney
Owner
Lunatic Studio
Sand Springs, USA
 
Join Date: Dec 2006
Posts: 958
Yes, I wish more engines to be represented.

P.S. Updated the Thea render with a new/better result in the first post.
__________________
Brian Looney

lunatic-studio.com
 
Old 07-21-2010, 12:26 AM   #8
CHRiTTeR
On the run!
 
CHRiTTeR's Avatar
Chris
Graphic designer extraordinaire
Belgium
 
Join Date: Feb 2002
Posts: 4,372
Well i've had a quick look at the maxwell scene and it seems that the glass material is a blend between 2 BSDF's with the same settings (glass with ior of 1.51) except that they have different roughness and reflectioncolor settings... which explains why i had such a hard time trying to recreate the same material.

I'm rendering 2 seperate images for each BSDF's and will post those here. This will make it easier for others to recreate the same glass as close as possible.

To be honnest i dont know why a blend of 2 different glasses was used, as i find it doesnt give such a good result compared to having a single BSDF (no blending) for glass.

Also, Bubbaloo, you might want to adjust the enviroment/scene settings you posted earlier. There is no enviroment lighting used in the maxwell file you posted.
If i use the enviroment settings you posted i get a totally different image.
Without enviroment lighting i get a correct result.

And last but not least, i dont think that vray render is using simular shader/scene/lighting settings it looks much too different.
I'll try to make a better vray render. But pls have patience for this because im doing this all on my laptop and i have to work on it also, so i dont have much time to do this.


BTW, is the challenge about getting as close as possible to the original work done by Henrik Wann Jensen or is it the challenge to try and get simular results compared with the maxwell render? Which one supposed to be more correct/accurate.
Im a bit confused about this

Last edited by CHRiTTeR : 07-21-2010 at 01:01 AM.
 
Old 07-21-2010, 02:40 PM   #9
tomNL
Next Limit Technologies
 
tomNL's Avatar
Atilla Akın
Maxwell Render Team
Turkey
 
Join Date: Dec 2006
Posts: 65
Special thanks to all who are showing interest and especially contributing to this challenge. Let me try to answer some of your critical questions:

Q -- Why do I use 2 BSDF? Is this a limitation or challenge for Maxwell?
A -- No. It's just the fact there is no surface in real life which has a perfectly uniform roughness. Every surface is a set of different roughness levels blended (usually in a chaotic/textured way). It does not mean a single substance should be made of single BSDF. Contrary, you can easily say less BSDF means less realism due to the above fact I've outlined. So, of course I could make a frosted glass using 1 BSDF (means a uniform isotropic roughness) but it'd not reflect the reality as needed. This is a common fact for any engine, nothing to do with Maxwell.

Q -- Why do you use ambient light? Is that essential for this look?
A -- The first post says: "The scene has a blue studio ground, a dragon, a small emitter plane and a dim skydome as ambient to match the original reference. But, you can use either of them or both or more when the features of the engine don't match (only emitter, only skydome or only IBL or all)." So, it's not crucial. I was trying to match the mood in Jensen's and that's all.

Q -- Is the challenge about getting as close as possible to the original work done by Henrik Wann Jensen or is it the challenge to try and get simular results compared with the maxwell render?
A -- The first post says: "There are 3 goals in doing this:
1) Highly rough transmissive surface achieving blurry refractions.
2) Full caustics and TIR including the caustics on its own shadow.
3) Clear glass material core with no SSS/translucency or other effects is essential.
I want to leave this test open to any kind of render engine for comparing the quality of outcomes to Maxwell Render. This is not a duel, so my respect goes to the ones which can at least succeed and demonstrate its quality in comparison."

Q -- What's the purpose in doing this?
A -- Rough dielectrics may easily become a pain when a render engine has several limitations/issues in several areas. Because it majorly covers the adaptation of BSDF, microfacet theory (roughness), caustics and more... I thought it's good having such a table of results to see how each engine competes in this area and to see which ones are out, yet.
__________________
Next Limit

Last edited by tomNL : 07-21-2010 at 02:42 PM.
 
Old 07-21-2010, 06:48 PM   #10
th3ta
Veteran
 
th3ta's Avatar
portfolio
3d Artist
In a dark office, USA
 
Join Date: Dec 2003
Posts: 739
I think this is kind of a cool project, to see how all the renderers compate to each other. But I can't help thinking that at the end of the day, it doesn't really matter, because usually you can just combine multiple passes and effects in comp to get a much different/realistic render. I think it's very rare (at least here at our studio) to have a 3d render go out the door untouched in comp.
 
Old 07-21-2010, 07:15 PM   #11
mister3d
Expert
 
mister3d's Avatar
portfolio
asdasd adsasd
Kiev, Ukraine
 
Join Date: Nov 2004
Posts: 6,031
Quote:
Originally Posted by th3ta
I think this is kind of a cool project, to see how all the renderers compate to each other. But I can't help thinking that at the end of the day, it doesn't really matter, because usually you can just combine multiple passes and effects in comp to get a much different/realistic render. I think it's very rare (at least here at our studio) to have a 3d render go out the door untouched in comp.


I hope one day comes we don't rely on compositing to look realistic. If you shoot it with a camera, it looks realistic, and it should be the same with rendering.
I looked at renderings done with arion and fryrender, and they look just more realistic as they are.

Last edited by mister3d : 07-21-2010 at 07:17 PM.
 
Old 07-21-2010, 07:44 PM   #12
CHRiTTeR
On the run!
 
CHRiTTeR's Avatar
Chris
Graphic designer extraordinaire
Belgium
 
Join Date: Feb 2002
Posts: 4,372
thanks for clearing that up tom... i fully agree on everything you said on mixing different BSDF's.
I guess i wasnt thinking straight when i posted that


Quote:
I want to leave this test open to any kind of render engine for comparing the quality of outcomes to Maxwell Render. This is not a duel, so my respect goes to the ones which can at least succeed and demonstrate its quality in comparison."


Yes, but in order to compare quality you first have to know what the correct results should look like and use simular settings, right?

Not trying to sounds annoying, thats just how i personally like to do my quality tests...

Anyway, i'll just use the same settings as the maxwell scene and post that (in a few days probably, sorry).

Contacted a collegue who owns an indigo licence, he'll have a look at it also.


ps: anyone knows how to convert 'cd/m' to 'lm' or 'lm/m?/sr' or 'W/m?/sr' or 'watt' (radiant power, not consumption)? if possible
*edit: according to my test (using cd in a photometric light and then changing it to lm) 2000cd/m is about 25132,7422lm, is that correct?

Last edited by CHRiTTeR : 07-21-2010 at 09:56 PM.
 
Old 07-21-2010, 08:09 PM   #13
Samo
Lord of the posts
 
Samo's Avatar
portfolio
Alvaro Luna Bautista
Roses, Spain
 
Join Date: Jul 2004
Posts: 981
Quote:
Originally Posted by mister3d
Yeah, vray looks more realistic. Thank you, now I will use vray.


My sarcasm detector is detecting something here.
__________________
www.yafaray.org
Free rays for the masses
 
Old 07-21-2010, 08:59 PM   #14
tomNL
Next Limit Technologies
 
tomNL's Avatar
Atilla Akın
Maxwell Render Team
Turkey
 
Join Date: Dec 2006
Posts: 65
Quote:
Originally Posted by CHRiTTeR
Yes, but in order to compare quality you first have to know what the correct results should look like and use simular settings, right?

Not trying to sounds annoying, thats just how i personally like to do my quality tests...
Now, here comes the most tricky part. Because, it's partially relative. It means, even if the engine doesn't have a basic problem with illumination and BSDF and even if the compared engines use same algorithms, it can still look different due to some other reasons. These reasons could be, for example an engine may easily miss the required caustics due to optimization or another one might only have problems with roughness or TIR and again it will fail. On the other hand, different engines may prefer different microfacet models and again it's understandable. All these factors will make the result look vastly or slightly different. Besides, there are several ways of making frosted glass in real life and it's also affecting the look. Sandblasting, acid etching, oil and dirt added later etc. Therefore, you shouldn't take this test more precise than needed. Current two contributions are more or less valid as they don't look completely like something else or insufficient about caustic noise etc. Once we receive more contributions, the differences may grow.
__________________
Next Limit
 
Old 07-21-2010, 10:05 PM   #15
SL
Hobbyist
 
SL's Avatar
portfolio
Sam Lavoie
USA
 
Join Date: Feb 2010
Posts: 60
Here's an Indigo render.

 
reply share thread


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 12:05 PM.


Powered by vBulletin
Copyright ©2000 - 2016, Jelsoft Enterprises Ltd.