[Josephus Holt]

BTW, talking about new technologies, I really like the idea of ArchiCAD's Virtual Building Explorer -
you can render the whole model/environment. Of course, ArchiCAD's render engine is nothing spectacular, but the greatest thing is, that after rendering is done, you can pack it into single .exe file, and then explore it (just like in PC games). Another cool thing is - it's just an .exe file - you don't need ArchiCAD, players, or libraries etc. to run it. It makes communication with a client VERY comfortable.
You can find some samples here:
http://www.graphisoft.com/products/virt ... -explorer/

Although I'm not an archicad user any longer, that is very cool indeed.

[zdeno]

http://www.archetyp.pl/orzeszek2.exe we did something like this in 2006 but it was very boring to set all lightmaps. So I am waiting for GI solutions in fly.
With this archicad solution it looks like more automatic.

[Michael Betke]

Guys please don't bash Octane so hard because of lacking features. It's offical still in Beta state and does never claim to be perfect or near done.
At the moment it doesn't support simple stuff like:
-IBL lighting, Sunrays through glass surfaces, SSS and other things which are, of course, present in an application which is on the market for years now and base on solid and well known technology.
You even get a lot of artefacts (fire flies) with the current version and som kind of MLT is missing.

As it comes to me I think I get a good feature set for the bucks I sold plus the free updates. I always was a fan of supporting developers and see software developing to a good state but this will not apply to all of us which is fully ok.
CGPersia got an Arion release some weeks ago and there were a lot of comments about the lack of quality and crashes. This could be either the reason because the people used a pirated/old version from there but also an indication why there is still no official demo.

I think GPU rendering is a nice new technology but maybe even a bit hyped because of it's speed. But lets take game-dev: Al people are hyping about thie new tesselation tech too.

But as individuals working in a fast paceing technology environment we should keep our eyes open and get an advantage if it arises.

[MetinSeven_com]

Hi guys,

This thread is too long to read in its entirety, but here are my few cents regarding the GPU rendering subject:

Every resource that would make Maxwell render faster is welcome to me, including the GPU. After all nearly every professional 3D artist owns a decent video card with fast GPU these days. Why not utilize it?

When Maxwell is going to support more rendering hardware resources, please make it hybrid CPU and GPU, not GPU-only, like some renderers do, and please make it easy to activate it.

When Maxwell is going to support hybrid rendering power, please keep the workflow in the 3D editors themselves as much as possible. I dislike the concept of having to export a scene from your 3D editor and then make realtime changes to the scene in external software, resulting in two versions of the same scene. Have a look at V-Ray RT GPU, which uses the ActiveShade rendering mode of 3ds Max to allow realtime changes to the scene.

Finally, when Maxwell is going for more speed, I would prefer a release supporting all Maxwell features, unlike Arion, which is a cut-down version of Fryrender right now.

Thanks and cheers.

[Fernando Tella]

Hi guys,

This thread is too long to read in its entirety, but here are my few cents regarding the GPU rendering subject:

Every resource that would make Maxwell render faster is welcome to me, including the GPU. After all nearly every professional 3D artist owns a decent video card with fast GPU these days. Why not utilize it?

When Maxwell is going to support more rendering hardware resources, please make it hybrid CPU and GPU, not GPU-only, like some renderers do, and please make it easy to activate it.

When Maxwell is going to support hybrid rendering power, please keep the workflow in the 3D editors themselves as much as possible. I dislike the concept of having to export a scene from your 3D editor and then make realtime changes to the scene in external software, resulting in two versions of the same scene. Have a look at V-Ray RT GPU, which uses the ActiveShade rendering mode of 3ds Max to allow realtime changes to the scene.

Finally, when Maxwell is going for more speed, I would prefer a release supporting all Maxwell features, unlike Arion, which is a cut-down version of Fryrender right now.

Thanks and cheers.

+1

[MetinSeven_com]

Muchas gracias!

[Cadhorn]

I think, like a lot of people here, I've got a critical eye and it's taking a good hard look at the GPU renderers. So far.... meh. I like the speed, but the quality just isn't there. Maybe in the future? (Certainly in the future! but in the *near* future?)

Do I expect CUDA/OpenCL to have any effect on NL software? I dunno, it depends. *If* they can off-load some of the processing from the CPU to the GPU without altering the way in which the unbiased calculations occur and their results, then "hell yes"! I'd bet you money right now that NL is working on that. *BUT* (and I know almost nothing about the nuts and bolts of porting code to CUDA/OpenCL) if moving some or all of the calculations to the GPU means adapting available hardware-optimized routines that are biased... that ain't gonna happen.

Is there anybody in this thread who knows much about programming for CUDA/OpenCL? I'm really curious to know how flexible it is. And what sort of calculations that are needed for unbiased renderings could be done with a GPU?

[VisualImpact]

I think, like a lot of people here, I've got a critical eye and it's taking a good hard look at the GPU renderers. So far.... meh. I like the speed, but the quality just isn't there. Maybe in the future? (Certainly in the future! but in the *near* future?)

Do I expect CUDA/OpenCL to have any effect on NL software? I dunno, it depends. *If* they can off-load some of the processing from the CPU to the GPU without altering the way in which the unbiased calculations occur and their results, then "hell yes"! I'd bet you money right now that NL is working on that. *BUT* (and I know almost nothing about the nuts and bolts of porting code to CUDA/OpenCL) if moving some or all of the calculations to the GPU means adapting available hardware-optimized routines that are biased... that ain't gonna happen.

Is there anybody in this thread who knows much about programming for CUDA/OpenCL? I'm really curious to know how flexible it is. And what sort of calculations that are needed for unbiased renderings could be done with a GPU?

Arion is unbiased, so yes it can and has been done.

[tom]

The word "unbiased" alone is not enough to explain what a render engine is capable of. Being unbiased is something, what you're able to do with being unbiased is another...

[VisualImpact]

I know what you mean, the ability to navigate your render in real time and edit materials in real time, dof, camera setup, lighting. Unbiased rendering has come a long way.

[Bubbaloo]

I don't think you know what he means...

[Hybaj]

Tom why are you suddenly so big on the rough glass thing? Let us turn clock back in time and see that when it comes to quality Maxwell was one of the last commercial and unbiased renderers which finally solved the brightness loss at steep angles Heh quality quality and nobody ever found it fishy when they had dark edges on rough materials or too dark rough dielectrics until I finally did a topic about it. So much for the quality and photo-reality sense of users too.

And quality quality.... it took very long time for Maxwell to address many many pitiful issues it had from start so why are some of you guys so hypocritical toward new renderers that utilise the "brand new GPU thang?" Now I don't want to be a serious asshole but everytime I peek into the V-ray gallery I find works that seem to me much more photo-realistic and interesting actually (ok ok we still have Tora here).

Again we say quality and the majority of really good works have been edited in Photoshop for sharpening, color corrections or better say alterations, chromatic abberation and etc.. shouldn't all this stuff be rather tweaked in the renderer? I still find it funny that there is such a lack of controls for example for super-sampling in Maxwell... or heck why there are not more controls for vignetting? Developers are ussualy not good in knowing which feature will really spice up the rendering or just make customers happy. Alot of the times they just want to patronize the user.

So let's be honest.. clients have usually the worst sense of photo-reality, so an temporary unavailable feature here and there is not a problem at all

That being said I don't want you to think about me that I have anything against Maxwell. Heck I don't and on the contrary I still love it. The only thing I'm scratching my head is the old-school conservative boneheads (well sorry but I had to) being hypocritical toward a new fresh and good path of development.

Even if CPU's which is highly unlikely (quite a few resons) will catch up to GPU's in term of speed in specific calculations it's still stupid not to use them. If a CPU is as fast as a GPU so why not have a 2x fast rendering??

Future = present = using every calculating capability in your computer... Period!! (OpenCL) )

[tom]

Tom why are you suddenly so big on the rough glass thing?

It's just one subtle technical example not successful in every other engine. Everybody asks for speed so I wanted to remind Maxwell is not wasting time while trying to simulate everything as correctly as possible.

Let us turn clock back in time and see that when it comes to quality Maxwell was one of the last commercial and unbiased renderers which finally solved the brightness loss at steep angles

True but last one? Don't be so sure as I know which ones still have this problem.

Heh quality quality and nobody ever found it fishy when they had dark edges on rough materials or too dark rough dielectrics until I finally did a topic about it. So much for the quality and photo-reality sense of users too.

I remember your post about it and very thankful to you for pointing those issues so, they are not here with us anymore.

And quality quality.... it took very long time for Maxwell to address many many pitiful issues it had from start so why are some of you guys so hypocritical toward new renderers that utilise the "brand new GPU thang?"

Seriously, it'd be stupid if we were against utilizing GPU. We just say we're not after quickly jumping the boat with sacrificing current quality. So, you shouldn't compare apples to oranges. It's not us comparing that...

Now I don't want to be a serious asshole but everytime I peek into the V-ray gallery I find works that seem to me much more photo-realistic and interesting actually (ok ok we still have Tora here).

Again Vray is another engine we have respect. It's around for many years and quite mature in many areas. But comparing engines by looking at galleries may be deceiving. It's the artist factor and invested time...

or heck why there are not more controls for vignetting?

Not sure I get this

Developers are ussualy not good in knowing which feature will really spice up the rendering or just make customers happy. Alot of the times they just want to patronize the user.

This is not true. I know it really looks like that but, I must admit NL is paying top attention to user requests as I'm the first one here complaining for adding other stuff to Maxwell but my wishes are behind the queue comparing to yours. That's how it should be....

So let's be honest.. clients have usually the worst sense of photo-reality, so an temporary unavailable feature here and there is not a problem at all

We hate features malfunctioning and we put huge efforts to fix them. As you say above, it took time but now they are fine. Our primary target is always terminating the bugs and then improving the features. Of course, sometimes 500 things can be fixed in 1 month but only 1 serious issue may stay alive for versions without remarkable improvement. It doesn't mean we're forgetting or abandoning the fix. Contrary, the fixing-agenda is always permanent and every fix is a must. Sooner or later... This is more important than dropping tens of working stuff away and rendering a few lambertian surfaces on GPU.

[Hybaj]

Tom, it was very comfortable to jump the boat with a buggy and slow software to get $$$ funding for it and make promises to people and not fulfill them, wasn't it? It's very hypocritical to say stuff about other companies when almost everyone did the same exact things and maybe even worse. So it's comparing oranges to oranges actually...

My only point is really for the people to be respectful about other softwares when they have little to no knowledge about why are graphics cards suddenly "usable" for rendering. No matter the quality of shading and tone-mapping, it's still the user who is either able to do good 3d models, diffuse maps, roughness maps, displacement maps to convince others that his rendering looks good or not. My rant is not about "Oh Why Next-limit you don't want to support GPU's anytime soon".. seriously I would never complain about that. It's all about the way people talk about other software which is not really rational or fair.

- A single GTX 480 is almost like having ~105 ghz of Intel's i7 processing power. 30 core intel cpu's with the power of i7 per core in the next few years? I don't really think so .. the architecture would have to be changed dramatically.
- Workarounds are popping quite fast for doing things with Gpu's which were previously thought to be too impractical or maybe even impossible. That's the beauty of it.
- Jumping the ship always seemed beneficial (seriously it's not a boat anyomore.. it's a damn ship!!)
- Path-tracing now is standard and there is now a lot of knowledge out there to about the features so that's the reason why actually all renderers are having this "standard unbiased look".
- The number of bugs decreases over time

[Half Life]

- A single GTX 480 is almost like having ~105 ghz of Intel's i7 processing power. 30 core intel cpu's with the power of i7 per core in the next few years? I don't really think so .. the architecture would have to be changed dramatically.

Here's a 16 core processor right now:

http://www.newegg.com/Product/Product.a ... 6819115063

Last year 8 cores was a big deal -- it doubled in roughly a year... and a dual socket motherboard would put you at 32 i7 cores on a single machine. So yes, it will happen whether you think so or not.

Intel isn't going to just roll over an die to leave the rendering market share to GPU without a serious fight -- and they REALLY don't like Nvidia (can't say I blame them).

Best,
Jason.