Why no objective comparisons between available Rendering solutions?

[art3mis]

First let me say I am a novice user and also really addicted with Houdini as a creative tool.

At the moment trying to get a basic mastery of the rendering pipeline, shaders and the various rendering solutions available for Houdini.

A no brainer to me would be to have a site dedicated to comparing the various render engines out there. No single render solution is THE best. They all have their strengths and weaknesses, and,

as someone who doesn't have 100's of additional spare hours to master every rendering solution out there ( ie being told to learn them all!), it is those strengths and weaknesses I need to know.

 

[Skybar]

Well if you don't have the time, who does?

Although I reckon there's tons of info on any particular solution if you just look for it. Maybe not straight up comparisons with graphs between renderers and things like that, because that would be stupid, but yeah. Evaluate what works best for you and go with that. No one can do that for you.

[michael]

the only useful way to do this would be to build a scene (or small set of scenes) that contained a full range of geometry types (polys, nurbs, curves, volumes etc), animated and static and then build a set of shaders with all the characteristics possible in the various renderers (BRDF, sss, refraction etc etc).

then render the scene(s) with all the different renderers to try to get the exact same resulting image - then compare the render times, how long/difficult was it to get the settings right, what each renderer could and couldn't do etc

it's not impossible but would take a lot of time and effort, something that not many people have to spare.

[mestela]

Agilles at 3dlight has started a super basic raytracing throwdown between various renderers:

https://3delight.atlassian.net/wiki/display/3DSP/Ray+Tracing+Stress+Test

 

 

[haggi]

The problem with setting up scenes for different renderers is that you have to know how the renderer quite good to get the desired results. e.g. you have differnt GI engines in Redshift or VRay. So to set up a scene in an optimal way you need a deeper knowledge of the renderer. So to do a good comparision, you have to digg deep into the depths of the renderer. 

Before switching to other renderers, I'd try to use the buit in one unless you already know that you do not want to use it for some reason. But then you know what to search for. As soon as you hit the limits of the default renderer, you know what to search for in other engines.

A intersting comparision can be seen here: http://furryball.aaa-studio.eu/aboutFurryBall/compare.html

[edward]

15 hours ago, haggi said:

A intersting comparision can be seen here: http://furryball.aaa-studio.eu/aboutFurryBall/compare.html

Ugh ... That page has only 2 scenes and they don't even compare against the same renderers!

[Guest tar]

On 8/7/2016 at 6:04 AM, eco_bach said:

 

as someone who doesn't have 100's of additional spare hours to master every rendering solution out there ( ie being told to learn them all!), it is those strengths and weaknesses I need to know.

 

Easy one; start with the so-called fastest renderers out there, Octane and Redshift. Then when or if they start not doing it for you move onto the others.

Rendering can be a very deep subject, but if you're willing to invest the energy and effort then you wouldn't be able to gain any benefits from using more involved renderers.

[pezetko]

On 8. 8. 2016 at 2:08 PM, haggi said:

A intersting comparision can be seen here: http://furryball.aaa-studio.eu/aboutFurryBall/compare.html

This comparison was heavy biased: http://forums.cgsociety.org/showthread.php?f=87&t=1121380&page=2&pp=15 You can see more realistic render times and better looking results from correctly set up render engines in that thread. I wouldn't believe Furryball's guys a single number imho.

Speed isn't the most important stuff for renderer. Price, stability, scalability, support and setup time plays important roles too.

[lukeiamyourfather]

24 minutes ago, pezetko said:

Speed isn't the most important stuff for renderer. Price, stability, scalability, support and setup time plays important roles too.

This! Not everything that counts can be counted. I tried Octane for Houdini and it was underwhelming. Instances didn't work, it ran out of GPU memory, the install was not well documented, and the materials were cumbersome.

I get that the original poster wants to know what renderer is best but it's not that simple. There are tons of factors to consider and those factors change depending on what you're working on so there's no silver bullet renderer. I suggest starting with the renderer that comes with an application, Mantra in this case, and then branch out from there if there's a need to do so.

[art3mis]

Thanks for the replies. Reading thru the excellent (but already dated) overview of the state of rendering on fxphd my main takeaway is that choosing a renderer is a lot like choosing a religion:).

 

[Guest tar]

2 hours ago, eco_bach said:

 is that choosing a renderer is a lot like choosing a religion:).

 

Yes - but don't let that line be the end of your research - it's used by a lot of people not to investigate more about different renderers.

It's best to always test things yourself, as almost everyone is talking from a point of view. It's about working around other's views to see what is suitable for you, as new rendering tech, and options, are always coming out.

Edited August 11, 2016 by tar

[Federico]

On vimeo you can find some comparison videos, especially for volumes. In this thread there is a comparison between differen render engines (focused on the noise reduction for indoor scenes)

 

[Mandrake0]

i think when you really want to compare rendering engines you need some knowledge and i advice you to read the pbrt book. http://pbrt.org 

just kidding 

make a list of requirements what you want to achieve and with that you can find your rendering solution. if you want to be a rendering master then you need the pbrt book. 

[art3mis]

Op here. Came across this thread over on the redshift forums. Although probably biased(pun intended) much food for thought

https://www.redshift3d.com/forums/viewthread/8011/

[Guest tar]

In the end, it comes down to the artist. I've seen great and shit renders over the last 20 years, regardless of biased, unbiased, gpu, cpu new fandangle renderers.  Just get stuck into it