OTOY Forums

can i get realistic interiors for architecture and realistic character models for vfx type work with 2 gtx 780tis?

sorry if this is a newbie question. what if they're in pci-e 3.0 x16 and x8 slots, will i suffer a lot in performance for using a x8 slot? please help. ty.

You can get realistic stuff from one. The number of cards you have will only effect how quickly you get it

As the preposter said, the number of cards does not influence the quality of the image, but the speed at which you're getting it. I use two 780Tis and am happy about the speed as a hobbyist, because I only render single small spaceships etc. If it was my source of money and my job, I'd personally invest in at least four 780 Tis if not even more with an external box .... I'm not very patient under pression and big archviz scenes use much more ressources than my little spaceships.

As for the x8 slot: PCI 3.0 is faster than PCIe 2.0 (more or less double the speed). 8x PCIe 3.0 is not even hindering gaming a lot, much less Octane .. you should have no problems at all with 8x slots (for Octane and rendering, probably even 1x is in theory enough). it's not like a game, where you have to update textures all the time as you progress in the game ... for rendering, you transfer textures etc once and then render for a long time usually.

Thanks for the replies guys. What kind of wait times would I be looking at with 2 780 ti's? It's all I can afford at the moment but if they are going to take 5+ hours or days I won't even bother haha.

Maybe I should also look into Thea, I do have a 4670K clocked at 4.4ghz and 32 gbs of ram. It's for work, but I'm just starting out in the 3d animation world.

How long it will take is far too big a question as it depends on too many things (how much geometry, how complex the materials, what kernel you use, what depth you go to, what lighting you use).

If you already have the programme, and a card, you should be able to get a comparison from someone based on the benchmark scene for example.

Dive, simply run some test with some free scenes available & You can do the math. Quality is very subjective & it depends on artist. Some are happy with DL renders that could be done in a matter of minutes with these two cards - some would render at least for half a day - that's subjective as I've said (it depends whether You are going to cheat or not calculating light, how You choose to optimise You models/scenes..)

Overall Octane is very fast, but one of the best things that it's very easy to use - taht saves a lot of time =)

interesting question... i got nearly the same

whats the difference between normal applications for example games and octane? usually its nearly useless if you use more than 2 cards, but it seems octane works different... is there any good faq about it somewhere? havent found anything

sorry, i am also a newbie who tries to learn and understand all that stuff

Hi gaazsi, Octane use the cards nearly 100% during render time.
Games and other Apps use it more variable between 0 and 100%.
Therefor you need a very good case, good cooling and a power supply that deliver 100% over hours.
And also important what Voon said:

it's not like a game, where you have to update textures all the time as you progress in the game ... for rendering, you transfer textures etc once and then render for a long time usually.

Cheers, mib

think this question is not hard to answer:
no.
octanes strength are the pathtracing and PMC modes and you will get old waiting for a complex interior getting noisefree in high-res.
of course you can render those images in the much faster DL Mode with AO but even otoy says that mode is not realy suited for final renders.
beside this you should notice that they dont say octane is up to 50 times faster than other renderengins but up to 50 times faster than other unbiased CPU-Rendersolutions. what they dont say is that only few would try to render complex interiors with a unbiased cpu renderer. the mainreason to use octane may be the easy usage and the amazing quality. but speed? not realy

another small question from a newbie

in my rig i have running three of those new 6gb gtx780 (zotac) cards now atm... so i could add a 4th card i guess (?)...
i have an old unused computer in my basement with a gtx 580 in it... is it possible to simply place this card also in my rig and it just "adds" his (compared few) cuda cores for rendering? or is it just slowing down the other cards then?