Page 2 of 3

Re: render issues for printing large

Posted: Sun May 22, 2016 7:46 pm
by calus
Nice board, thanks Ahmet.

One precision, it seems to me V3 is still limited to 76 millions triangles so not filling yet the 12 GB VRAM of a TitanX.
see here:
viewtopic.php?f=33&t=53732

Re: render issues for printing large

Posted: Sun May 22, 2016 8:14 pm
by aoktar
calus wrote:Nice board, thanks Ahmet.

One precision, it seems to me V3 is still limited to 76 millions triangles so not filling yet the 12 GB VRAM of a TitanX.
see here:
viewtopic.php?f=33&t=53732
Thanks! Do you think CPU renderers are not limited by system RAM? Both are limited by their memory amounts. I specified that in table.
Hmm, i think i missed that point.

Re: render issues for printing large

Posted: Sun May 22, 2016 8:30 pm
by calus
aoktar wrote:Do you think CPU renderers are not limited by system RAM? Both are limited by their memory amounts. I specified that in table.
Hmm, i think i missed that point.
my point:
V3 will be theoretically limited only by the amount of VRAM yes,
but not yet! there's still a limit to 8.5 GB VRAM,/76millions of polys due to range overflow.

:arrow: "Increased the maximum number of triangles to 76 million (see viewtopic.php?p=270444#p270444 ). This equates to ~8.5GB of memory. Higher numbers will result in a range overflow. We have a few ideas how we can solve this problem, but this will require some more work, which we plan for the time after 3.00."
viewtopic.php?f=33&t=53732

Re: render issues for printing large

Posted: Sun May 22, 2016 8:55 pm
by aoktar
Yep apparently. I feel bad myself due this mistake. Sorry!

But i don't think that Octane is ready hundreds of millions triangle to work. Because compile performance is not good yet as other engines. It will be perfect if guys could improve it on speed.

Re: render issues for printing large

Posted: Mon May 23, 2016 11:14 pm
by jakeamin
I've tried up to 5% on the polygon reduction but it's still having issues, stopping too soon when it's stil black? Do i need to upgrade to 3V :S

Re: render issues for printing large

Posted: Mon May 23, 2016 11:20 pm
by aoktar
jakeamin wrote:I've tried up to 5% on the polygon reduction but it's still having issues, stopping too soon when it's stil black? Do i need to upgrade to 3V :S
I think that i give the all clues to decide for that. I can't answer that without knowing any good info about your specs, free vram on rendering, etc..

Re: render issues for printing large

Posted: Mon May 23, 2016 11:31 pm
by calus
Yes it may be just that you stop the render to soon, the process to export to Octane the 15 million polygons object is very very long,
but I think you may have a better way to reduce your object, go in the sculpt menu and choose "unsubdivide", then go in the sculpt tag open the "layer editor" and revert to a lower subdivision level before to converting it back to polys (this only works if you object is subdivided"

About last c4d plugin beta, you have nothing to loose to try it (don't forget to put "parallele sample" to 1 ). You can revert to V2 if this doesn't work.

Edit : or just gives to Ahmet the infos he asked you to help you... ;)

Re: render issues for printing large

Posted: Mon May 23, 2016 11:40 pm
by jakeamin
Sorry I'm new to octane aoktar so still getting confused by alot of this
running an i7 4790k at 4ghz
32gb ram
gtx 980
If anyone's able to fix this scene as I am needing to print for wednesday for final show in university it would be much appreciated and I could even donate for the troubles.
I think I might try recreate the object tomorrow with less polygons to see if that helps

Re: render issues for printing large

Posted: Mon May 23, 2016 11:53 pm
by aoktar
I think just renderbuffer for this resolution will take about 1.5 Gb Vram on V2. Also OS+C4D takes some big amount VRAM. Consider that also geometry needs some VRAM.
If you use one GTX980 with 3 or 4gb you have not luck with V2. Try to use Software Opengl for C4D. Also try to use commandline rendering method.

Re: render issues for printing large

Posted: Mon May 23, 2016 11:58 pm
by calus
well I tried to unsubdivide your geometry but it doesn't work for it.
If you can generate it again with much less polygons (no more than 10% polygons you have now), this is the way to go anyway.
It will be way much easier to test some render and correct other problems after that.