United3dArtists Archive

http://www.refractivesoftware.com/index.html"

I knew gpu rendering was on it's way, chaosgroup and others have dabbled with it, but I didn't know about this product.

Looks promising, and affordable.

recommendation removed.

Maybe wait the nvidia fermi...

Yes, the new nvidia cards are apparently purpose built for Cuda - multiple times better than current cards. They might give this renderer a boost as people go looking for reasons to use the extra power......

benji, you mean Octane or TS Dx9?

Marcel, the Fermi might be a little to pricey, might have to wait and see...
I just can't see spending more than $300 for a graphics card.

When It comes out, maybe I can get a better deal on an Nvidia
GTX 295 or something!

I was talking about octane. I wouldn't advise getting the gtx 295 for octane, as octane doesn't support SLI.

They say on the octane site that the gt 260 is the best bang for buck, but the new fermi cards are supposed to have double the memory of these cards (and much better cuda perfromance in general). This is important, as gpu-memory is all that is available for octane, cpu-memory isn't .

It's tricky, but if you can get your hands on a cheap gt 260, that may be your best bet for now. Otherwise, waiting until march when the new cards arrive may be wise. You could also upgrade to a cheap 9800gt or something in the short-term, but I have no idea what the performance with octane is like, and they generally have low memory.

Basically, for all these reasons, I probably won't be using octane until mid-year, when the fermi situation is clearer, and they've had a chance to iron out the bugs in octane.

What I found interesting is, if your motherboard can support it, you can have a dedicated card for Octane and another one for the OS. Thanks for the link.

seams that many Render companys jump on the GPU acceleration wagon

http://www.randomcontrol.com/arion

Not sure if it will be an update or an alternative to fryrender.
Cool is the combo from GPU+CPU

Very nice render engine. I like the node based texture system and the real-time results. The only problem I have with it is that it seems to be limited to one video card so It would be a good idea to wait until the GPU’s are available with more than just one gig of ram for larger scenes. Would be nice to have a card with about 16 gigs of memory or be able to use SLI enabled cards combine memory of multiple cards.

Very impressive hopefully it will not end up costing 3000.00 dollars plus hardware. Who knows we may end up with new workstations les than half the size of an average laptop that can render entire feature films in a matter of hours on one single unit. That would be nice as well.

recommendation removed.

Hypershot is a really nice renderer - just a bit expensive. I wouldn't write them off yet, they have a big head start on the gpu crowd, and they currently do their magic without cuda. Imagine what they will be able to do with what they've learned on the cpu, when they apply it to the gpu......?

The next twelve months are going to be interesting.

[edit] It seems bunkspeed have dumped hypershot (something to do with a licensing disagreement), and are making a new app. I guess the future is now...

Benji wrote:Hypershot is a really nice renderer - just a bit expensive. I wouldn't write them off yet, they have a big head start on the gpu crowd, and they currently do their magic without cuda. Imagine what they will be able to do with what they've learned on the cpu, when they apply it to the gpu......?

The next twelve months are going to be interesting.

[edit] It seems bunkspeed have dumped hypershot (something to do with a licensing disagreement), and are making a new app. I guess the future is now...

Oh wow..those poor hypershot buyers....

Luxrender is also going to go with openCL, but that will take a while since apparently openCL is
not as mature as cuda yet.

http://www.luxrender.net/wiki/index.php ... and_OpenCL"

http://www.blender3darchitect.com/2010/ ... he-future/"

Hypershot is still being sold, but by a different company owned by the original license holder (for the rendering tech behind HS) who has apparently had a hissy fit of some description. Personally, I think the guy's crazy, but you never know why it all went down. Bunkspeed are working on a new product to replace Hypershot (Hypernot?).

I personally think Cuda is going to cement itself as the premier option regardless of OpenCl, as Nvidia have boldly organized their new range of graphic cards around Cuda. This is important, as Ati seem more concerned with frame-rates and dx11 than stream processing. They are way behind in the processing stakes, and really it will be at least a generation before they catch up from what I've read.

Nvidia may have lost their luster with the gamer crowd, but they have bulldozed Ati in the processing market. I think this was good move on their behalf. My bet is that Ati will eventually license Cuda whatever happens with OpenCl - I guess we'll see.

recommendation removed.

Yeah, my gaming card is pretty weak, so the decision to get a new one isn't too hard. I suppose the decision to upgrade isn't as clear cut for those who bought the higher end ATI 4 or 5 generation cards.

You need a very good card for pretty normal scenes......
what you get is the render in about 6-8 seconds(to clear up) while you wait for a scene of that size for 40 seconds at 1600 px wide with radiosity on a standard render engine(chess)and 1minute 14sec on the spaceships.
I just tested the test scene that comes with the octane trial.
Its pretty nice if you want to visualise a product and make fast renders with it but for scenes with dozens of objects with tons of materials its getting a bit
overwhelming......
its a technology to keep an eye on but quite frankly I think we are not "there" yet...
Or to put it differently the scenes that take my render engin a day to render wouldnt load in octane...
Benji the so lowly respected Kerkythea rendered that scene in 17,9 seconds at 1200 x 1600 pixels with hdri scene 1 and photon mapping setting no 5(chess scene).
(22,3 sec for spaceship scene)

nice feature is the device memory useage gauge !

They've been talking about it, well the future or rendering over at the Thea forum and for them its way to early to take the plunge
I guess it comes down to how real do you need your images to be, how much money can you afford to spend, are you doing animation, how fast do you need to render and how physically correct does the scene need to be and how many polys are you stuffing in there

nigec wrote:They've been talking about it, well the future or rendering over at the Thea forum and for them its way to early to take the plunge
I guess it comes down to how real do you need your images to be, how much money can you afford to spend, are you doing animation, how fast do you need to render and how physically correct does the scene need to be and how many polys are you stuffing in there

your right i may ad that the standards are not out yet.... Cuda is just one way there is also open cl and caustic one and caustic Rt and Tesla and fermi and and and
dont get me wrong i have seen some awsome simulations almost in realtime but would I want to spend money now without any standard......
just to see my renders 10 seconds earlier.....
both my puter and the graphics card are good for another 2 years if they dont break.. !

Plus its all well and good having these sexy new toys, but they are only any good if you know how to use them
Marcel and Wiz can make Likeworks do stuff which is outstanding, there's Blender stuff that is as good as Vray, its all down to the guys using the tools not the tool itself

nigec wrote:Plus its all well and good having these sexy new toys, but they are only any good if you know how to use them
Marcel and Wiz can make Likeworks do stuff which is outstanding, there's Blender stuff that is as good as Vray, its all down to the guys using the tools not the tool itself

No question ......the render speed is pretty much different but the quality of the work mostly depends on the user..

Well, another rendering thread, another troll. The same as always. Epic

Benji wrote:Well, another rendering thread, another troll. The same as always. Epic

Indeed

interesting is how much impact the gpu seams to have on the rendering speed.


A GTX 285 costs currently approx 270euro,
not sure if it's worth the "saved" time but nice to see the power of a gpu
in something else as just games

parva wrote:

Benji wrote:
A GTX 285 costs currently approx 270euro,
not sure if it's worth the "saved" time but nice to see the power of a gpu
in something else as just games

The technology is cool no doubt but i think it will take another year or two for the market to find a common standard.

recommendation removed.

KILLER APP DETECTED!

That's amazing, but I'm facing problems with the demo install, how can I found manuals to do a first render, etc.

leonregis wrote: That's amazing, but I'm facing problems with the demo install, how can I found manuals to do a first render, etc.

it's a bit like working with the truespace LE editor

just open the macro in the graph editor and you see the nodes for camera etc.
now right click and insert the sample OBJ files from the octane demo scenes
and voila the prog begins to render.

my nvidia 8800 with 320 mb isn't exactly the kind of equipment
but still good results.
Think I will wait like Rayman said...

This gpu super-rendering is interesting. I'll bet it is much of the future of rendering. For now, even if I could afford it, I'd wait for it to mature. Some people need the cutting edge -- mostly professionals. Hobbyists like me can use this when it is cheaper and can work to a good hobby level on cheaper hardware.

The demo has which limitations or its more like a beta?

obj importer exporter from truesourceshop.I just purchased octane render and need to know whether I need to buy this product? I model in the modeller side but is there something in the workspace side I should be using before I export to the octane renderer? Thanks!

Wow! Octane is a good render engine. All I have is the demo but I like it very much and all I have is one Nvidia GTX 260. I plan to add another card soon as I can save out the cash but you can find a good card for around 150 bucks at new-egg or some other like it. But you must be careful and make sure the card will fit into your motherboard and inside your case. Just because you have a PCI Express slot don't mean you have enough room and power supply to use one. My single card requires 2 power cables so you cant just stick it into anything off of the shelf. Also my machine would need another card to make it faster than Thea.

As my workstation sits now Octane is a bit faster than Thea and Thea is a CPU based engine and you have much control over your project. So I would advise anyone to look over their situation before buying Octane and you should at least try the demo for Thea to compare.


No you don't need the OBJ plug-in for Octane if you use model side.

removed due to harassment from outside of these forums.

tylerzambori wrote:removed due to harassment from outside of these forums.

what? why?

simply removed.