Corona Interactive

[cecofuli]

I'm not 100% sure that, with heavy scene and DR, you'll get slower feedback...
In V-Ray the interactivity performance is near the same in DR or in Local.
Ok, you have to wait more for the first step, because you have to share all the stuff inside network. But this isn't th main topic of this discussion.
I don't understand why you continue to you continue to argue the opposite. For example, you can change the V-Ray bundle size to get a better interactivity performance.
And, I prefer to wait 1-2 seconds and to get in 10 seconds (with 5 nodes for example) a clean preview, instead to wait 50 sec for the same result with one machine.

[Ludvik Koutny]

Yes, that's Vray. But i am sure you very well know Corona DR is nowhere near Vray level, so I can't imagine current system working realiably with interactive rendering. As i said in my previous post, i do not think it's impossible, but there needs to be a lot of steps taken before this could turn into usable solution. There are also lots of things of higher priority, be it hair rendering, skin shader, adaptivity, proper refraction handling, improving reliability of regular distributed rendering, etc...

It's really not as easy as simply switching on current DR for IR mode.

[cecofuli]

Yes, I know a little =)
I can understand that it isn't easy to implement it. But, I think, it's very useful to have IR in DR mode.
And, we will have no IR slower feedback (except when you add a new, heavy geometry), because we know Corona's guys ;-) 

[Ondra]

about rawalanche: as a rule of thumb, he has just strong opinions - not necessarily always in favor of Corona ;)

about personal jabs of any kind - please stop them

about DR vs. interactive - the problem is the way Corona interactive is implemented - everything supported, everything works as final rendering - it would be probably possible to do a different model "send scene to all machines, wait a while, and then have fast camera viewport updates" and record nice demos with it. But transmitting all scene changes would be much more complicated/slower.

In the end, we are currently happy with existing IR and we will focus on more new features, stability and performance fixes, before investigating options of DR+IR

[MarkusB.]

I know it has been a while since something was posted here. But what is the status on IR with DR?
Is this still something you are investigating or did you dropped the idea completely?

[Ondra]

we still have no plans to do it

[Benny]

I don't know if this is a totally different discussion, but I understand Corona is only using CPU for IR as this allows for identical functionality and code between final render and IR. I also remember another argument was that CPUs are getting more and more cores, and that you don't have worry about fitting your scene in VRAM, and that this trend is likely to continue for the foreseeable future.

It's been a while since Corona alpha now and I wonder if the same still holds true? With some nvidia cards having 12GB and rumored to go to 16GB for the next Titan it seems CPUs will have a tough time keeping up. A 10 core 6950 is $1,600 and Xeons even more, whereas a Titan X is about $1,000.

Is it time for Corona to consider a GPU strategy for the future, or is the CPU track still the way to go?

[danio1011]

Hi all, sorry to resurrect an old thread, but I'm assuming still no plans for IR + DR?  Twould be sweet :)

[Benny]

I don't know if this is a totally different discussion, but I understand Corona is only using CPU for IR as this allows for identical functionality and code between final render and IR. I also remember another argument was that CPUs are getting more and more cores, and that you don't have worry about fitting your scene in VRAM, and that this trend is likely to continue for the foreseeable future.

It's been a while since Corona alpha now and I wonder if the same still holds true? With some nvidia cards having 12GB and rumored to go to 16GB for the next Titan it seems CPUs will have a tough time keeping up. A 10 core 6950 is $1,600 and Xeons even more, whereas a Titan X is about $1,000.

Is it time for Corona to consider a GPU strategy for the future, or is the CPU track still the way to go?

Almost six months later I also wonder if anything has changed on the GPU front?

[maru]

No plans for IR+DR - the reasons are the same as previously - we want to keep IR responsive.

No plans for GPU rendering in any foreseeable future. We might however offload some other tasks to GPU, such as VFB post processing. There is a huge list of reasons why we don't want to switch to GPU. We are planning to release an official blog post, or other kind of article, explaining why exactly we believe in the CPU.