A machine with 4x GTX 980 should crush anything Keyshot or Octane could produce. What render device mode is selected in Tools > Options > 3D Viewport? Try unchecking "Automatic" and select "GPU". This will force Visualize to just use your 4x GPUs, as a good test starting point.
Even with 1x GTX 980 your renders should be light years faster than Keyshot.
Do you have a reseller you could work with as a starting point to benchmark your rendering performance with your CAD files?
It is set to GPU only.
I do have a re-seller to work with and they have been very cooperative but for Visualize the forum has been a great source of info!
Are there specific GeForce drivers that work best for Visualize?
The Visualize forum is a great wealth of knowledge! That's because it's so active with all the global users
We recommend at least NVIDIA driver 354.56. A google search for that driver will bring up the download link. Of course any newer driver should work just fine as well, but this is the minimum driver recommended with Visualize.
It wasn't clear from your original post; are you rendering Visualize in "Accurate" mode? Have you tried a Fast mode render for comparison? Also, are you benchmarking final render quality and time? Or interactive viewport performance?
I've used Octane a bit and can attest to their speed. You CANNOT quote me on this as ti is only a guess at this point, but I believe Octane is strictly Optix based. Which means they'd be more freely able to make optimizations their engine, that specifically suit their users pipelines and their product. The downside is that the engine doesn't automatically support fallbacks (ie rendering in CPU only mode, which iRay does) or anything else that comes along (now and in the future) as part of the broader iRay package.
Anyhow, just sharing my $0.02.
Wow... 8 Westmere cores faster than 11264 Maxwell cores!!! Very curious to see in which conditions!!!
Is it into the same scene with same settings.. (models, materials, lighting etc)?
That said, Keyshot is very fast for a CPU rendered and Ron is right, Octane is faster than Iray/Visualize because fully GPU integrated (even if it will be in crossing platform), while Iray/Visualize is CPU+GPU neutral. Taht the price of the homogeneity.
If you want to correctly compare (as said by Ron) Visualize to Keyshot and Octane, you have to set all your renderers in BIASED KERNEL MODE it in FAST MODE and Octane in Direct Lighting. For Keyshot, I don't know how many ray bounces are set as default but more or less it would be the same quality.
And as a prouf of it... Here you will see 2 images of renders I have done in the past during "Bunkspeed area", so same as Visualize:
My computer had an Intel I7 860 (old CPU!) and my graphic card was Nvidia GTX 480 (old too, it was my radiator during winter!)
First one done with Keyshot 5 (default settings):
Second one with Bunkspeed Pro 2014 in fast mode (default settings too)
That's 3 times faster...
I think it's not fair to compare product shots with hdri only lighting. Clearly iray has a better advantage.
I would be more interested to see a comparison in interior renders.
I think this article clears some things up:
(read the "Common Misunderstandings" section)
So I guess in the end GPU renderers and SW Visualize offer a better work flow and I would agree.
I would encourage you to do comparison shots with the scene attached.
test scene.skp.zip 1.4 MB
Well, Ok when I will have time I will do the test. Thanks for sharing.
But I'm not sure too fully understand what Ronen Bekerman is meaning. It looks like he is mixing different stuff (even if I'm very respectful of his work!). Because he compare Vray, which is a very high-end render but biased renderer to Octane in Pathtracing mode (so unbiased...).
For me (maybe I'm wrong):
- GPU is faster than CPU in biased like unbiased modes, outdoor or indoor scenes.
- Biased is faster than unbiased but not as qualitative, especialy in indoor scenes.
- Indoor scenes are more complex, so take more time to compute, than outdoor scenes
But in some case, like yours I suppose, a CPU in biased mode can be faster than GPU in unbiased mode, like in interiors.
About this, because of its 100 pass only, Fast mode is not adapted to the indoor scenes. Fast mode is more set for cars and objects, so in outdoor.
Hi Nico, Thanks as always for your voice in our Visualize Community! It's great to hear direct feedback from power users such as yourself.
Hi Marios, as Ronen said in his article, it's not only about the final render time but the total time from fresh import to final render. I personally find the Visualize interface, with its integrated Viewport and simple settings for materials, the fastest to produce photo-quality content for my workflow.
Even though the NVIDIA Iray render engine is truly identical to a photo, they are always looking at ways to speed up the render times to get to that photo-quality convergence. While exterior shots, studio shots and others like that are completely solved, the real speed improvements are coming with interior scenes. Each year, NVIDIA almost doubles the render speeds from one chip set architecture to the next, and I'm really looking forward to getting my hands on their newest board, the Pascal series.
Was your original issue on your first post solved when using the latest version of Visualize 2017 Beta, that was available on Aug 16?