Chaos Corona for Cinema 4D > [C4D] General Discussion
Speed of different Versions
(1/1)
wsiew:
Hi,
actually i am working with Corona 10 on a MacBook pro Intel Core i9 (2019, 64GB Ram, 8GB GPU) and Cinema 4D R21 on Monterey. Now i want to upgrade to Sonoma, but also with C4D R21. Is there a different in speed with Corona Version 10, 11 or 12? In case i should use Corona 12: Can i use all Renderfarms or do you recommend any special farm?
Wolfgang
BigAl3D:
I too have had this question about the actual render engine being improved, speed-wise. Previously, when 3D apps were updated, they'd always show you have the engine has been optimized to give a 20% speed boost or something like that. This meant on your existing system, you should see a speed improvement, but I have never seen Corona announce that the engine's speed has been improved, just new and improved features. I'm not complaining about Corona as I do love it, but it would be nice to hear the v12 engine is 15% faster than v11. Having said that, there is never a good reason not to be running the newest version of Corona. You are paying for it, why not use it?
As far as render farms go, you need to check with each one to see who can render Corona and up to which version. You also need to think about any plugins you need for output. I haven't used a farm with Corona, but there are many out there. Here is a link from their site.
https://www.chaos.com/render-farms?product=chaos-corona&country=
TomG:
We do post measures of speed ups when they are introduced, such as we did with the faster caustics in 12 Update 1 (though this was better expressed in terms of quality, given that any speed up would depend on how dominant the caustics were in the scene, so this was "caustics are about visually twice as good in the same amount of render time")
And there have been "x% speed ups" in releases in the past. Thing is, the algorithms and calculations are now very well optimized, and as such there is less opportunity for something that is just a raw speed up to rendering time, and as noted most improvements that Corona can give are about speeding up YOU by making the workflow easier, with rendering time speed ups most often coming from hardware improvements.
For instance, https://www.chaos.com/blog/corona-renderer-7-for-3ds-max-released had some noticeable and measurable speed ups.
Ultimately though, light transport calculations are what they are and as time goes on there are less and less ways in which they can be improved upon (if you've done your job well in the first place and things are handled efficiently) so this is why over time improvements are less about faster renders and more about faster you :)
As a note, for speed ups we don't want "cheats" or "approximations" or "short cuts" because Corona is focused on accurate, realistic results, so any speed up has to come without any sort of compromise in result to achieve it. This is one of the reasons you can see claims of speedups for GPU renders, e.g. "now 3 out of 4 frames are AI approximations, rather than 1 out of 4 frames!" is the kind of thing that wouldn't fly for Corona, were there a CPU equivalent, at least not for final renders (you can see a similar 4 times speed up by using the new Upscaling, which is fine for test renders and that can be very valuable, but it's not really applicable to final deliverable work).
Hope that helps!
BigAl3D:
Maybe we're hitting the coding version of Moore's Law. As fast as it can go without melting. Or running Microsoft Word on a 64-core Threadripper vs. a 96-core Threadripper. Still fast, but as Scotty used to say in Star Trek, "Aye Captain, without new dylithium crystals, she's going as fast as she can."
TomG:
Yep - to grossly over-simplify, if you have to do "a = b + c" there is only so much you can do in code to speed that up; the calculation is what it is, and needs to be done, and what determines whether that gets faster is the hardware it is run on.
Of course this is not to say we don't still look for optimizations, mostly through new algorithmic approaches than simply optimizing our code, but they do turn up less and less often the more mature a product gets (if it's done right!) as you are already exploiting the best code and best approaches. But we'll never stop looking, looking at research and seeing if it can be applied, and so on!
Navigation
[0] Message Index
Go to full version