Download Trial Buy now
Log in or Sign up for the chat.

Search  | Login  | Register

Post new topic Reply to topic  [ 46 posts ]  Go to page Previous  1, 2, 3, 4  Next
Offline
User avatar
Hi there,
very nice to see my tests being discussed here.

In terms of "Eneen" that wanted to know the AA settings that were used:

Attachment:
Screenshot.PNG
Screenshot.PNG [ 50.6 KiB | Viewed 336 times ]


I guess thats the default settings.

In my personal experience I rather have a clean looking image. I do not care if its a tad softer then what other renderers produce. I professionally render either for animation in FullHD or for Print in really high res.
Personal stills I normally render in 4K. Normally there is enough room to use a sharpen filter if you ever have the feeling that the render looks a bit soft. If you compare it with photographs you normally do not get a pixel accurate result because of the bayer interpolation that needs to happen.

So for me Indigo does a great job here. But thats only my two cents!

The only reasons I prefer Octane over Indigo are:

Feature completeness with SSS, Dispersion, Volume rendering.

Nodal based material system.
Especially how it makes it possible to alter textures. Mix them with noises / other textures, multiply them remap the lightness values (ramp) etc.

Shader based (per object) Dirt (AO) / Inverted Dirt shader. Vertex Map Support.
There are so many things you can to using those effects working around UV-Unwrapping and object and painting textures for it.
You can produce used edges with inverted Dirt (AO) or get interesting effects by using it as a mix texture.

Falloff / Fresnel Shader that is uable inside any channel.
Material often display other angle based properties them. Therefore its important to have a tweakable shader that reads out the facing angle depending on the camera / viewer.

Cheers,
Raphael

_________________
You don´t dream in cryo.


 Profile  
 
Offline
Thank you Silverwing.
In the meantime I wanted to be picky and it seems all are almost same except thea which is sharp. So it will most probably give less noise when set same as others but I really doubt if it will be faster then indigo.

_________________
http://www.mastersfinger.com


 Profile  
 
Offline
User avatar
2nd Place 100
Thank you very much for your clear but kind speech Silverwing!!


Gesendet von iPhone mit Tapatalk

_________________
Cheers, David



DAVIDGUDELIUS // 3D.PORTFOLIO
·
Indigo 4.0.47 | Indigo for C4D 4.0.48 | C4D R14.042 | Mac OS X 10.12 | Windows 7 Professional x64


 Profile  
 
Offline
Comparing the quality of the Maxwell and Indigo renders based on the transparent dragon image (the left-most one at http://www.silverwing-vfx.de/Download/Convergence_Test/Dragon_02.jpg), it appears that Indigo is blowing out the highlights.

I don't know if this is because Indigo has a more limited dynamic range than Maxwell or if Maxwell is doing some dynamic range compression. To my eyes, the Maxwell image appears more natural and faithful to how those highlights would look in real life.

What is your opinion on this? Which of the two (Maxwell vs. Indigo) is producing the more realistic render?


 Profile  
 
Offline
Silverwing wrote:
Hi there,
very nice to see my tests being discussed here.


Hi Silverwing, I would like to further explore the differences between the Maxwell and Indigo rendered versions of the transparent version of the dragon from http://www.silverwing-vfx.de/Download/Convergence_Test/Dragon_02.jpg. Would it be possible to post the scene file here, so that I can correct it as much as possible to match exposure.

It appears to me that highlights are blown out more on the Indigo version as compared to what Maxwell is rendering, thus losing detail in those highlights. On the other hand, the Maxwell render is happening more slowly, so given more time, Maxwell may also blow out those same highlights. Or, it may be possible that Maxwell has a higher dynamic range than Indigo (and/or is doing some sort of highlight compression) and this is being illustrated by this image. I would like to explore this further.

SharpEars


 Profile  
 
Offline
User avatar
SharpEars wrote:
Comparing the quality of the Maxwell and Indigo renders based on the transparent dragon image (the left-most one at http://www.silverwing-vfx.de/Download/Convergence_Test/Dragon_02.jpg), it appears that Indigo is blowing out the highlights.

I don't know if this is because Indigo has a more limited dynamic range than Maxwell or if Maxwell is doing some dynamic range compression. To my eyes, the Maxwell image appears more natural and faithful to how those highlights would look in real life.

What is your opinion on this? Which of the two (Maxwell vs. Indigo) is producing the more realistic render?


Indigo blowing out the highlights is probably because the rays are being magnified through the glass and form a very bright highlight. It's quite easy to fix this by choose a tonemapping method with more dynamic range or use Indigo's colour curves to reduce contrast. Judging from the weak caustics in the Maxwell render, I'd say it might do some clamping or something.

_________________
http://fb.me/OscarJDesign
https://www.behance.net/OscarJohansson


 Profile  
 
Offline
SharpEars wrote:
Comparing the quality of the Maxwell and Indigo renders based on the transparent dragon image (the left-most one at http://www.silverwing-vfx.de/Download/Convergence_Test/Dragon_02.jpg), it appears that Indigo is blowing out the highlights.

I don't know if this is because Indigo has a more limited dynamic range than Maxwell or if Maxwell is doing some dynamic range compression. To my eyes, the Maxwell image appears more natural and faithful to how those highlights would look in real life.

What is your opinion on this? Which of the two (Maxwell vs. Indigo) is producing the more realistic render?


Maxwell has default highlight compression using burn parameter set to 0.8. It works similar to reinhard and gives about 1EV in highlights with 0.1 and about 3-4EV with 0.0. To get linear response you have to set burn to 1.0.

_________________
http://www.mastersfinger.com


 Profile  
 
Offline
User avatar
Hi there everyone:
Yes I can make the scene available. Give me a bit of time to prepare it and upload it,

P.s. I am aware that Maxwell has normally a Reinhardt mapping of 0.8.
Thats why I set it to 1 for the test. Because I want to compare all render engines as linear as possible.

Remember that this is Maxwell CPU I have been comparing it against. Mainly because Maxwell GPU only has a ray depth of 8. Indigo GPU has a ray depth of 16 per default (at least that´s what I could figure out by testing the refraction through multiple layers of glass)
Maxwell can be a bit slow in resolving rays that have multiple reflections / refractions. Thus It could be when let it render longer it would become brighter.

Maybe I should do another test. Not about speed but about the final noise free look ?!

Also the algorithm that is used to calculate blurry rays is highly responsible for a different outcome of the look.
Maxwell is known for factoring in the fresnel strength per sample rather then per surface normal (or at least has clever algorithms), thus the influence of the fresnel becomes smaller the higher the roughness of the material is. This is because you more and more dispersing the rays more and more averaging the fresnel as you reflection more and more converges towards a diffuse distribution.

Since Next-Limit is really closed up about what algorithms they are using inside maxwell (Ray Depth etc) its really hard to tell whats different from Indigo here. I guess its rather the material refraction glossiness algorithm then any global sampling setting or tone mapping.

Again only my two cents :-)


Attachment:
Maxwell_Ray_Depth_GPU.jpg
Maxwell_Ray_Depth_GPU.jpg [ 342.31 KiB | Viewed 186 times ]


Attachment:
Ray_Depth_Indigo_GPU.jpg
Ray_Depth_Indigo_GPU.jpg [ 202.52 KiB | Viewed 186 times ]

_________________
You don´t dream in cryo.


 Profile  
 
Offline
User avatar
Interesting. Are those rendered for the same/similar amount of time?

_________________
http://fb.me/OscarJDesign
https://www.behance.net/OscarJohansson


 Profile  
 
Offline
Silverwing: thank you, wasn't aware of that ray depth limit with maxwell gpu, glad haven't upgraded. I wonder why Indigo gives brightest rough glass of all tested engines. Maybe devs will comment it when they come from vacation. Can it be related to really fine resolved caustics it has? And why cycles rough glass is so dark...
BTW: it probably won't be related but can affect look of plastics and ceramics, problem with additive layers:
http://www.maxwellrender.com/forum/view ... 90#p391997

_________________
http://www.mastersfinger.com


 Profile  
 
Offline
User avatar
Oscar J wrote:
Interesting. Are those rendered for the same/similar amount of time?

In this test it was not about achieving the same lighting situation, nor the noise. It barely was about the ray depth. So no I do not think that those renders were being cooked the same amount of time.

Eneen wrote:
Silverwing: thank you, wasn't aware of that ray depth limit with maxwell gpu, glad haven't upgraded. I wonder why Indigo gives brightest rough glass of all tested engines. Maybe devs will comment it when they come from vacation. Can it be related to really fine resolved caustics it has? And why cycles rough glass is so dark...
BTW: it probably won't be related but can affect look of plastics and ceramics, problem with additive layers:
http://www.maxwellrender.com/forum/view ... 90#p391997

Yes unfortunately maxwell utterly fell flat of my expectations. Also you can only use one GPU as of now. Its about to change no doubt, but I felt really let down by the initial release.

Where the darkness comes from I really have no idea to be honest. Might be the caustics but I rather think its the overall BRDF function that has been used.

About that additive layers. Actually that's a good test. Normally you should not get brighter results then the environment on the Object.
Also a 100% matte reflection within a completely uniform environment should look completely flat and not exhibit any fresnel characteristics (where the rim becomes brighter)

P.s. sorry for taking so long. Here´s the Indigo as well as the Maxwell file. I do not know if Maxwell has the correct texture link. Probably not. But you can re-export it using the C4D file that is provided also:
DOWNLOAD


EDIT:
I rendered the Maxwell scene for 1h on 5 machines (CPU network rendering) and this is the result.
It takes damn long but looks really convincing... don´t you think.
I could achieve something similar in Indigo using some absorption.
Generally I would say that there is something fishy going on in Indigo making the glass too bright.


Tha´s the Maxwell result:
Attachment:
Dragon_Maxwell_01.jpg
Dragon_Maxwell_01.jpg [ 304.08 KiB | Viewed 116 times ]



That´s the Indigo Render with absorption for comparison:
Attachment:
Dragon_Indigo_01.jpg
Dragon_Indigo_01.jpg [ 276.66 KiB | Viewed 116 times ]

_________________
You don´t dream in cryo.


 Profile  
 
Offline
Silverwing wrote:
Tha´s the Maxwell result:
Attachment:
Dragon_Maxwell_01.jpg


Wow, the Maxwell dragon looks much more realistic at this longer render time. I wonder how the clear glass dragon would look given more time. I have a feeling that the Maxwell dragon would look best in this case as well.


 Profile  
 
Offline
User avatar
2nd Place 100
Why does it look more realistic? It looks more homogenous, but is that realistic?


Gesendet von iPhone mit Tapatalk

_________________
Cheers, David



DAVIDGUDELIUS // 3D.PORTFOLIO
·
Indigo 4.0.47 | Indigo for C4D 4.0.48 | C4D R14.042 | Mac OS X 10.12 | Windows 7 Professional x64


 Profile  
 
Offline
User avatar
Interesting comparisons.

Rendering a glossy transparent object in a uniform environment is a good way to make sure they follow the laws of energy conservation. As you can see Indigo (CPU) renders the glossy transparent monkey pretty much exactly homogeneously with the uniform background. As does Maxwell, after a very long rendering time, at least fairly close.

I agree with zeit that I don't think the Maxwell one looks more realistic. To compare apples to apples you should probably compare Indigo CPU with Maxwell CPU though, as Indigo CPU has virtually infinite path depth, while the limited path depth of Indigo GPU probably affects its appearance (see the darker edges in your render). Would also be interesting to see how Maxwell GPU compares.


Attachments:
glossy transp maxwell test.jpg
glossy transp maxwell test.jpg [ 162.33 KiB | Viewed 102 times ]
glossy transp indigo test.jpg
glossy transp indigo test.jpg [ 205.08 KiB | Viewed 102 times ]

_________________
http://fb.me/OscarJDesign
https://www.behance.net/OscarJohansson
 Profile  
 
Offline
User avatar
2nd Place Winner
zeitmeister wrote:
Why does it look more realistic? It looks more homogenous, but is that realistic?
Exactamundo!
In the provided scene we have a lightsource from the left, the frosty dragon could act like a magnifying glass and bundle the light from that side.
Here a long CPU rendering of that dragon (20kspp), where you see that its even brighter then on GPU!
Attachment:
quickrender.jpg
quickrender.jpg [ 345.46 KiB | Viewed 98 times ]

_________________
polygonmanufaktur.de


 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 46 posts ]  Go to page Previous  1, 2, 3, 4  Next



You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  

Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group