Render Comparisons
Render Comparisons
Hi,
for people interested in different rendering engines and global illumination. Here is a (still kind of hidden) link:
http://www.janwalter.com/RadianceVsYouN ... ameit.html
I just started using Indigo RT and the scenes are more about bouncing lights and IES usage than artistic scenes. There are hardly any textures being used to make those scenes "beautiful" but at least you can see different renderers rendering the same scene, although it might be difficult to match materials, tonemappers etc. ...
Anyway, let me know if you have questions, or if you find it useful (or not) ... it's work in progress ...
Cheers,
Jan
for people interested in different rendering engines and global illumination. Here is a (still kind of hidden) link:
http://www.janwalter.com/RadianceVsYouN ... ameit.html
I just started using Indigo RT and the scenes are more about bouncing lights and IES usage than artistic scenes. There are hardly any textures being used to make those scenes "beautiful" but at least you can see different renderers rendering the same scene, although it might be difficult to match materials, tonemappers etc. ...
Anyway, let me know if you have questions, or if you find it useful (or not) ... it's work in progress ...
Cheers,
Jan
Re: Render Comparisons
Is it just me or does the indigo render lack contrast? You would expect the area directly behind the light to be strongly illuminated too (3.6) . The lighting looks oddly.. uniform?
Re: Render Comparisons
Hi wahn,
it's one of the most thoroughfull comparison I have ever seen, keep it up !
What I see is that Indigo is the only engine casting the mirror's caustics on the ceiling
it's one of the most thoroughfull comparison I have ever seen, keep it up !
What I see is that Indigo is the only engine casting the mirror's caustics on the ceiling
obsolete asset
Re: Render Comparisons
I assume you are talking about the right image (with three lights on the ceiling and two sconce lights mounted at the wall). Here is a screenshot how the lights are modelled:neo0. wrote:Is it just me or does the indigo render lack contrast? You would expect the area directly behind the light to be strongly illuminated too (3.6) . The lighting looks oddly.. uniform?
The original Radiance scene uses a sphere as light source (like a lightbulb), and the rectangle you see from the camera is also a dim light source. To make sure that the light bounces off the rectangle I duplicated the geometry (as you can see in the screenshot) and offset it a bit. The front rectangle does emit (dim) light, the back rectangle has a simple diffuse material and I wanted the light emitted by the (brighter) sphere to bounce off this rectangle and against the wall.
Do you think it might be the multiplier(s) for the "Material Emission" part? The "Power" values are as in the original Radiance scene, but I used a multiplier of 10^6 for all light emitting geometry.
Anyway, I will run another test with different mutlipliers and see if things change ...
Re: Render Comparisons
Thanks. At some point I want to give the developers of each renderer the scenes and ask for help with tweaking the parameters and material settings to get the best and fastest results. So far I just want to get "some" images for all of them before I start asking for help.CTZn wrote:it's one of the most thoroughfull comparison I have ever seen, keep it up !
You are right, but to be honest I removed for some renderers the "reflectivity" of that "antique copper"-like material and you can see some caustics at least on the wall. If I let more light bounce off the material it should reach the ceiling too, but so far I still have trouble finding the right intensities etc.CTZn wrote:What I see is that Indigo is the only engine casting the mirror's caustics on the ceiling
Re: Render Comparisons
Excellent, that's exactly what most comparisons are lacking: a good setup for each renderer.wahn wrote:At some point I want to give the developers of each renderer the scenes and ask for help with tweaking the parameters and material settings to get the best and fastest results.
I am right but impetuous, that was not necessary from me. We don't know rendering times for other renderers for instance.wahn wrote:You are right, but to be honest I removed for some renderers the "reflectivity" of that "antique copper"-like material and you can see some caustics at least on the wall. If I let more light bounce off the material it should reach the ceiling
Can we help ? I think you should use the linear tonemapping method to clarify intensities. Indigo can emit lux, nit, w/m² and cd. See the "model" section of the technical reference.wahn wrote:but so far I still have trouble finding the right intensities etc.
obsolete asset
Re: Render Comparisons
I added a new image (of a new scene):
A larger version can be found here:
http://www.janwalter.com/RadianceVsYouN ... igo_01.png
It originally comes from Luxrender and can be found here:
http://www.luxrender.net/forum/gallery2 ... temId=3535
The guy who modelled it is called Martin Lubich and provided the model for download:
http://www.loramel.net/
I will ask him if I can use the model for my render comparisons but I assume I can, because he published it under the Creative Commons Attribution license.
A larger version can be found here:
http://www.janwalter.com/RadianceVsYouN ... igo_01.png
It originally comes from Luxrender and can be found here:
http://www.luxrender.net/forum/gallery2 ... temId=3535
The guy who modelled it is called Martin Lubich and provided the model for download:
http://www.loramel.net/
I will ask him if I can use the model for my render comparisons but I assume I can, because he published it under the Creative Commons Attribution license.
Re: Render Comparisons
BTW an image of the real crown can be found on Wikipedia:wahn wrote:I added a new image (of a new scene) ...
http://en.wikipedia.org/wiki/File:Weltl ... _white.jpg
Re: Render Comparisons
This is quite some nice bling-bling
Only thing I would change is your tonemapping setting from Reinhard to linear.
Linear looks better (not so washed out colors) and also should be included in the most other renders.
what materials did ya use for the gemstones and pearls? Indigo has some nice physical based gemstones from tabulated spectrum scans in the database, but this maybe hard for other engines to compare....
Only thing I would change is your tonemapping setting from Reinhard to linear.
Linear looks better (not so washed out colors) and also should be included in the most other renders.
what materials did ya use for the gemstones and pearls? Indigo has some nice physical based gemstones from tabulated spectrum scans in the database, but this maybe hard for other engines to compare....
polygonmanufaktur.de
- pixie
- Posts: 2332
- Joined: Sat Dec 29, 2007 4:54 am
- Location: Away from paradise
- 3D Software: Cinema 4D
- Contact:
Re: Render Comparisons
The reinhard doesn't look as washed if (a) you tweak the values a (b) bit and have a good light setup.Zom-B wrote:This is quite some nice bling-bling
Only thing I would change is your tonemapping setting from Reinhard to linear.
Linear looks better (not so washed out colors) and also should be included in the most other renders.
what materials did ya use for the gemstones and pearls? Indigo has some nice physical based gemstones from tabulated spectrum scans in the database, but this maybe hard for other engines to compare....
Re: Render Comparisons
Yes, 2/2/10 settings can do some magic for example, but I was actually more pointing in the comparison direction then blaming Reinhardpixie wrote:The reinhard doesn't look as washed if (a) you tweak the values a (b) bit and have a good light setup.
Anyway, I look forward to see some more engines rendering that crown!
Tell us when you are done wahn.
polygonmanufaktur.de
Re: Render Comparisons
Thanks for the tip. I changed that and you are right about the colors. Looks much better now.Zom-B wrote:This is quite some nice bling-bling
Only thing I would change is your tonemapping setting from Reinhard to linear.
Linear looks better (not so washed out colors) and also should be included in the most other renders.
For the gemstones I looked at the sources of the materials in the library and used a "specular" material with "transparent" on. The medium was "basic" with "IOR" and "Cauchy b Coeff." taken from the examples of the library . The absorption color I took from the materials I saw being used in the Luxrender counterpart (RGB values within [0,1] range) and the "Brightness" I attached by eye during rendering to make the gemstones more or less transparent. I assume it's like a absorption based on the thickness of the stones.Zom-B wrote:what materials did ya use for the gemstones and pearls? Indigo has some nice physical based gemstones from tabulated spectrum scans in the database, but this maybe hard for other engines to compare....
For the pearls I used a simple trick and blended 3 materials, two phong materials with a hard and a soft specular highlight, and one semi-transparent glass-like material which gave it a subtile subsurface scattering look. With Indigo RT I can't use real subsurface scattering and I was just messing around with the idea, which I also got from looking at the original Luxrender counterpart.
In general it looks like sometimes the exported values (e.g. the bump mapping multipliers) get messed up during the Blender export. For those things I adjust the material settings while rendering with Indigo RT. I have to figure out at some point why this is happening. I'm quiet a beginner with Indigo and I did not try to leave e.g. the subdivision surfaces up to the renderer (is that possible to send the control mesh instead?). I tessellated everything in Blender, exporter OBJ files (also for the other renderers), brought them back into Blender, and tried to work on some bigger junks of geometry by joining meshes, using multi-materials (several materials on the same mesh), and bringing in geometry step-by-step before exporting from Blender. Sometimes things go wrong during the export and the error messages are not really helpful. This way I have more control and see when things go wrong, which makes it easier to investigate. For example I realized that if you use the same material twice in a material list, I rather combine those into a single material by selecting the one (or the other) and assigning the already existing material to both parts. Then I get rid of the double listed material, so it exists only once in the list. This was caused mainly by the OBJ re-import. I also make sure that materials are shared between different meshes, if they are reused on several parts.
Who is online
Users browsing this forum: No registered users and 32 guests