OTOY Forums

Do I kiwi123? well maybe I 'm just excited and waiting for 2.3 to arrive.
I must do this more often I'm usually slightly depressive

I dont know what type of stuff you render but there are gpu with up to 6gb out there.
Thats more than many pc have for cpu rendering.
I only have a 1gb card atm and I havent found it a limitation for my tasks but everyones need is different.
radiance has said he's extending the res to 8192x8192 shortly which is very big
so I dont really see that complaint holds up.
I am sure there are many clever ways to make use of memory, cpu and gpu that havent been explored yet.
I dont know where gpu are going but I would think they will continue to have more memory, and more cores for about the same cost. Not that long ago 64mb on a video card was big. 6gb would have been in lala land.
I think gpu apps and gpu will evolve together.
Octane might be remembered one day as being like the Pacman of renderers.

Yes, I'm excited about 2.3 as well, although I'm not sure why, as I probably won't use it much...lol

I have a lowly 9800GT with 512Mb ram, which is not a lot of memory you can imagine. I have 8Gb in my system, which is enough for my big images in non-GPU rendering though
I don't have the cash to splash out for a 480GTX or something like that, let alone a 6GB card! Plus the only thing I'd be using it for properly would be octane, which is not the case with a newer cpu (currently q6600 which could do with an upgrade also).

Well I think we had better try and fit you up with a nice GTX460/2gb as an upgrade and keep that stalwart old Q6600 going a just little longer in Octane nannying duty.
I think they will be about $450 in NZ. One of the reasons I am excited for 2.3 to come out is confirm more cores can be utilised in the 460 with radiances tweaks.
It could be 4-5x faster than the 9800...
That sounds like a good deal if you think of a decent cpu + mobo + ram costing ~3x as much.
and you could reuse the 460 in your next pc you get with your Lotto win
By sheer coincidence I have been looking at this size and I think a 2gb card will take an A3 size image @300dpi (thats very roughly 5000x3500) plus about 4-5m polys and quite a few textures - The actual use mix depends of course but I think 2gb is quite a handy size for many purposes.
If you did big renders all the time you might want a second 460 though.
Have i made a sale?? I've got to fill in some time somehow ah this w-a-i-t-i-n-g..

Wow this thread blew up. Its great to see everyone so passionate about this. Whether they were honest in their video or not, we have to understand that we are dealing with an unbiased engine versus a biased engine in lux (correct me if I am wrong on that). In doing that one already starts with a handicap. "Lets test blender's AO render against Vray. Look Blender was faster. Surprise surprise." Like it was said before, a listing of specific parameters would be best before making any claims. GPU rendering isn't all the way there yet, but it will be soon. I don't see this as a competitive type of thing, but still as a self boasting bit that might have some of their customers stop asking for gpu rendering in lux.

Personally I've used Octane on countless scenes and have had way better render quality than ever before. Render times seem to be roughly the same as with the other "internal" engines, but the quality is what is important. I also don't have to spend time doing photon mapping or adjusting a ton of parameters to get something usable. The interacivity also makes it much easier when I do have to adjust some things. Nothings worse then having to render a frame, adjust a setting, render again, adjust a setting, repeat, until I find the magic values. I know I've often spent more time doing that then the actual modeling in the past.

Which brings me to my last point. How much pre-processing needed to be done before that render? Yes the render may have taken ~20 seconds, but how many time did they have to test different settings in order to obtain that and was there any light pre-render processing done which may have helped that render time. There's a lot of scientific method stuff here that just seems lacking for a valid comparison study.

Correct me if I am wrong but isn't the Modo renderer biased? If so it's like comparing Maxwell Render to Mental Ray.

Edit: Ah havensole you beat me

The new upper res limit, is that with no supersampling or what ever it is called?
So the effective resolution is half the max?

Yes, not a direct comparision huh.
I think one needs to take into account the fact that unbiased renderers have been generally alot slower to resolve than biased ones. Now with the extra speed of the gpu, the render times are more comparable. Personally I have not used unbiased renderers before because of the time; now it is comparable to cpu/biased and I get the 'unbiased' look.

I own licenses for Modo, Octane and Maxwell, and I did a test on the Luxology forums a few weeks ago (before I know anything about Luxology conducting their own tests), you can see the results here:

http://forums.luxology.com/discussion/t ... x?id=48218

I approached the testing with a completely open mind, I was pretty excited about GPU rendering in fact, but I came to the exact same conclusion that Luxology did - the technology needs more time to mature and although GPU rendering is faster than CPU it isn't 10x faster, more like 2x faster at most IMO.

All my file are available for download and checking.

pixelrush wrote:... about $450 in NZ...

Hi Pixelrush.
So you are in NZ as well? If you were using ProE instead of Solidworks at work I'd be able to guess where that was
What blog were PTC fishing in?

gristle, about the supersampling I'm not sure.

PTC man is 'Unhappy face' see
http://www.dezignstuff.com/blog/?p=3060 ... mment-7311
He gets in trouble though see Alan Beliniak's post a few down