R20 Physical Render Engine Speed Increases???

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

 
Thread Tools Display Modes
  1 Week Ago
R20 Physical Render Engine Speed Increases???

Was just watching one of 3DFluff's R20 intro videos, and he spent only a quick moment showing that the (progressive) physical render engine has seen a seemingly substantial speed increase. Depending on just how that manifests (he mentioned with "certain materials & certain lighting") - that alone would be worth the upgrade for me. Has anyone else who has had early access to R20 experienced these gains? It's been a LONG time since Maxon did anything to speed up either the Physical or the Standard renderers.
 
  1 Week Ago
Originally Posted by sneather: Was just watching one of 3DFluff's R20 intro videos, and he spent only a quick moment showing that the (progressive) physical render engine has seen a seemingly substantial speed increase. Depending on just how that manifests (he mentioned with "certain materials & certain lighting") - that alone would be worth the upgrade for me. Has anyone else who has had early access to R20 experienced these gains? It's been a LONG time since Maxon did anything to speed up either the Physical or the Standard renderers.

You probably did miss the significant R19 speedup for the standard renderer. See:https://www.youtube.com/watch?time_...2&v=tQvKpxqA1Jo

Best regards,

Wilfried
 
  1 Week Ago
Originally Posted by wbj: You probably did miss the significant R19 speedup for the standard renderer. See:https://www.youtube.com/watch?time_...2&v=tQvKpxqA1Jo

Best regards,

Wilfried
Well, I'm not sure I would call it "significant" on my end. I've battle-tested R17 versus R19, and I've only seen some marginal (well under 10%) gains with R19.
I have no idea how they got that ~50% boost in that quick demo.

P.S. They did indicate "lots of transparencies and reflections" - so perhaps when the materials don't contain too much of either, the gains drop rapidly.
 
  1 Week Ago
It isn't a general speedup, it affects PBR rendering mostly, and only with progressive mode. If you light with the diffuse reflectance channels or the pbr material, then youll probably notice it the most. Especially in scenes which seem to clean up quickly for a few moments but then just never seem to clean up no matter how long you leave them.

It is also tricky to benchmark, it renders as fast as before, but each iteration will look cleaner than before. Overall don't expect much difference in final renders, but your tests with bad grain in progressive mode will look better quicker. I'll see if I can make a scene which is possible to properly test
__________________
Matthew O'Neill
www.3dfluff.com
 
  1 Week Ago
Originally Posted by imashination: It isn't a general speedup, it affects PBR rendering mostly, and only with progressive mode. If you light with the diffuse reflectance channels or the pbr material, then youll probably notice it the most. Especially in scenes which seem to clean up quickly for a few moments but then just never seem to clean up no matter how long you leave them.

It is also tricky to benchmark, it renders as fast as before, but each iteration will look cleaner than before. Overall don't expect much difference in final renders, but your tests with bad grain in progressive mode will look better quicker. I'll see if I can make a scene which is possible to properly test
Thanks! And also thanks, if you can explore and document that further. By chance, did you try HDRI lighting, to see if that affected the quality of the passes? Or was it only the area lighting you experimented with in the time you had?
 
  1 Week Ago
Originally Posted by sneather: Well, I'm not sure I would call it "significant" on my end. I've battle-tested R17 versus R19, and I've only seen some marginal (well under 10%) gains with R19.
I have no idea how they got that ~50% boost in that quick demo.

P.S. They did indicate "lots of transparencies and reflections" - so perhaps when the materials don't contain too much of either, the gains drop rapidly.

It clearly depends on the scene, but on average you should see something in the 2 - 2.5x range (the CineBench scene for example is in that range, but there are extreme cases with 10x speedup). Basically anything with transparency, a lot of poly intersections or AO should see a good speedup.

Best regards,

Wilfried
 
  1 Week Ago
Originally Posted by wbj: It clearly depends on the scene, but on average you should see something in the 2 - 2.5x range (the CineBench scene for example is in that range, but there are extreme cases with 10x speedup). Basically anything with transparency, a lot of poly intersections or AO should see a good speedup.

Best regards,

Wilfried
Based on this "revelation" to me, I just ran a few tests. I created a scene, with two objects (one transparent and reflective, and one only reflective). I used a random HDRI on a sky object to provide a reflection to the scene, and I set the "ray depth" and "reflection depth" each to "12". In a variety of various camera angles, I was able to consistently see around 30-35% boosts in R19. But that's a FAR cry from the 2-2.5x or beyond. Now, I didn't try pushing this to the extreme, with AO or complex surface intersections - so maybe that is ALL needed, to actually see the extreme gains? I'll try a few more things to see.
 
  1 Week Ago
I just took my same test scene, and enabled AO, and merged a couple of the same (CAD) models into the similar space, so that there were complex geometric intersections.
Results, were essentially the same. In fact, I'd say R19 dropped a little under 30% faster. So, I'm still missing the "secret sauce" settings which would demonstrate such massive differences in render times.
 
  1 Week Ago
Originally Posted by sneather: I just took my same test scene, and enabled AO, and merged a couple of the same (CAD) models into the similar space, so that there were complex geometric intersections.
Results, were essentially the same. In fact, I'd say R19 dropped a little under 30% faster. So, I'm still missing the "secret sauce" settings which would demonstrate such massive differences in render times.
Well there 's no secret sauce. I just took the ContentBrowser ao_bike.c4d scene. Left part of the picture (dark color scheme) is R18, right part is R19:



Best regards,

Wilfried
 
  1 Week Ago
Wow. Interesting. Good suggestion, for that objective test. I just ran the same thing: ~15sec R18 and only ~6sec R19.
So, clearly, there are a number of other variables at play, because my "normal" setups for client work in the Standard Render engine simply don't get anywhere near this kind of performance difference in R19.
 
  1 Week Ago
Originally Posted by sneather: Wow. Interesting. Good suggestion, for that objective test. I just ran the same thing: ~15sec R18 and only ~6sec R19.
So, clearly, there are a number of other variables at play, because my "normal" setups for client work in the Standard Render engine simply don't get anywhere near this kind of performance difference in R19.
So maybe copy a scene over into the AO Bike Scene and compare it further? Or save out a render settings preset from the bike scene?
 
  1 Week Ago
Originally Posted by zeden: So maybe copy a scene over into the AO Bike Scene and compare it further? Or save out a render settings preset from the bike scene?
Certainly, a reasonable idea. However, I don't think it's so much the overall render output settings. I think it has to do with the subtlety of a myriad of material settings.
 
reply share thread



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 06:48 AM.


Powered by vBulletin
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.