Why is there no "Real Time Rendering" in the plugins

[Amaterasu]

Recently I decided to test the octane standalone and compare the speed of the plugin and the standalone version, as a result of my tests I found the "Real Time Rendering" function in the octane standalone. It increases vram consumption, but it also increases rendering speed, it is independent of parallel samples parameter which works great, can you tell me about how this function works and if it is possible to enable it in octane for cinema 4d (maybe there is some analogue of this function in the plugin that I don't know about).

The results of my tests for those who are interested ↓

[bepeg4d]

Hi,
Standalone Real Time mode is not completed, and it will be finished in 2023.1 SDK.
For now, it only moves the tone mapping to GPU, and a lot of features are not supported:

• Noise pass
• Denoised passes
• Cryptomatte passes
• Upsampler
• Adaptive sampling
• Export deep image
• Display color management
• Multi GPUs rendering
• Network Rendering

So, better to wait for next SDK.

ciao,
Beppe

[SSmolak]

Hi,
Standalone Real Time mode is not completed, and it will be finished in 2023.1 SDK.

At the moment Standalone is 25% faster than C4D plugin so when SDK will be finished it will be 50-75% faster than plugin right ? I think that Amaterasu asked about speed difference not current SDK stage...

[frankmci]

At the moment Standalone is 25% faster than C4D plugin so when SDK will be finished it will be 50-75% faster than plugin right ? I think that Amaterasu asked about speed difference not current SDK stage...

Amaterasu asked how it works and if it is possible to enable it in C4D.

Since Beppe says Standalone is currently only moving tone mapping to the GPU for real time, who knows what the different efficiency gains might or might not be made from the various other necessary processes? For all we know, they did tone mapping first because it had the greatest potential for speed gains, and all the others are significantly less promising.

Personally, I find my quality of life is much better when I don't try to extrapolate from sparse information. Of course when OTOY makes the announcement, they will use standard promotional language like, "Up to XXX times faster!". And that will be true, "up to," in certain specific circumstances, which most of us won't experience in our day to day work. For me, what's working now is pretty darned good. When some significant improvements are available, I'll be delighted, but overly anticipating the future diminishes the quality of the present. Already, images look so good and render times have already become such a small consideration compared to the overall project that I barely worry about it. There was a time when everything, especially image quality, was constrained by how long it would take to render finals by the delivery date. At this point, almost every improvement is icing on the cake. It won't be long before nearly everything is real-time. That will be nice, but still a relatively small, incremental advance compared to how far we have come since the early 90s.

(This is, of course, in my particular corner of the industry. We're not doing Hollywood blockbuster effects or Superbowl commercials. But the customers are happy and most of us manage to work close to normal business hours these days, instead of the insanity of a few decades ago.)

[SSmolak]

Right, 25% speed increase in Standalone compared to plugin version is not important if you work with small images. But if you going to render four hi-res images for print and one of them is rendering one hour there is big difference - 4h vs 3h right ? So the question is - why plugin version is that much slower than standalone ?

[bepeg4d]

Sorry, could you share an example scene that shows 25% difference in speed between c4doctane and Standalone?

ciao,
Beppe

[SSmolak]

Amaterasu, can you send this scene to Otoy for compare ?

166 seconds vs 194 seconds is exactly 25% if nothing has changed in recent years.

[aoktar]

Real time rendering is not accelerating the render speed. It's more like having faster viewport refreshes as I can know. As providing direct memory transfer between GPU and CPU. Less time to display images, because transferring the render buffering is taking more time without it. Adding this support will be examined because I don't know that C4D will allow it or not.

[funk]

Amaterasu, can you send this scene to Otoy for compare ?

166 seconds vs 194 seconds is exactly 25% if nothing has changed in recent years.

I'm confused how you are getting 25% from these 2 render times. Can you explain your formula?

I searched google for some different formulas, and percentage increase/decrease is calculated like this:

If Standalone is the original render (166s), I'd calculate C4Ds render time percentage increase with:

((194-166) / 166) * 100 = 16.87%

or if C4D is the original render (194s), then Standalone's render time percentage decrease is:

((166-194) / 194) * 100 = -14.43%

[SSmolak]

Amaterasu, can you send this scene to Otoy for compare ?

166 seconds vs 194 seconds is exactly 25% if nothing has changed in recent years.

I'm confused how you are getting 25% from these 2 render times. Can you explain your formula?