For anyone who hasn't seen the post by Muphasa over in the DAZ Carrara forums - development of the Carrara plugin for Octane Render is underway. More information can be found here:

http://render.otoy.com/forum/viewtopic.php?f=7&t=36332

I downloaded the Demo version of Octane and the demo Poser plugin today and all I can say is WOW!!!. I tried the last two Poser scenes that I did with Reality/Lux, and was easily getting results that would take 12-70 hours with Lux in less than 30 min. (yes, that's right, 70 hrs.) 

If you can live within the texture map and RAM constraints of your video card (Nvidia only), Octane is definitely worth a serious look!!! I have a Geforce 670M with 3Gb RAM in my laptop (think 560 in a desktop), and I was able the render and image with StoneMason's Castle in it with over 2Gb left over (but it did exceed the 32 gray scale map limit) with fantastic results in about 30 min. The same image (Castle Mage in my Gallery here) took 70 hours to render in Lux - plus my system didn't get as hot as it does when rendering with Lux.

The Carrara plugin looks quite nice so far. Of course the other real issue besides a good graphics card, is the cost (about $450 for the Octane/Poser plugin). But, if your using software that is truely aimed at the professional market like Octane, you really have to expect higher prices than what we typically pay for Carrara and related plugins (not that they aren't good, but it's what the market "expects").

If your interested in Octane, I'd definitely recommend testing the Demo version and the demo of the Poser plugin if you can. The results are so impressive that I'm seriously considering taking the plunge. It's really not to hard to see the value of getting lighning fast renders in your workflow (well maybe not lightning fast, but amazingly fast for sure). 

Great news! Thanks for sharing.

I really enjoyed reading this post. I congratulate you for the terrific job you've made. Great stuff, just simply amazing!

Thanks MsMillie!

All this 'over-excitment' about Octane for Carrara is similar to the cheering when Luxus for Carrara was released. Everybody was in tears. Finally a renderer that we can use to express our boundless creative talents!

And then it flopped. Luxus for Carrara didn't even reach version 1.0.1. I bought it and I never even installed it. There were a few images rendered with Luxus posted in RO's Carrara galleries, and then nothing.

Will the same thing happen with Octane plugin for Carrara? Will its development stop at version 1.0.1 or 1.1.x at most? Or will it live happily to reach version 2.x and 3.x and so on?

Considering what happened to Luxus, and considering that most plugins for Carrara don't get beyond version 1.1.x, I suspect that the same will happen to Octane for Carrara, so I'am not holding my breathe on it.

Carrara's native renderer is OK. It can produce acceptable results, but it is slow and, compared to Shade's jaw-dropping renderer, it is only good for amateur work.

The solution to Carrara's ailing renderer is not Lux or Octane. It is, in my opinion, Embree by Intel.

The Embree library can give Carrara's renderer a much needed speed boost, up to 30x by my estimate, while at the same time significantly improving the output quality.

Embree is now at its 2.0 release and, considering that it is part of Intel's core technologies, we can be sure that it will live to reach its 3.x and 4.x versions and beyond.

Intel is most likely more than willing to provide all the free consulting that a company like Daz needs to incorporate Embree in its products, thus actually absorbing some of the implementation costs.

Embree is a unique opportunity for Daz to rejuvenate Carrara relatively quickly, probably not more than 6 months of work for one knowledgeable developer.

The huge advantage of Embree over external renderers is that Embree will be integrated in Carrara's native renderer, thus it will support Carrara's shading and lighting systems. Therefore there will be no need to re-adjust the textures or redo the scene lighting.

While I do understand Embree being a major improvement, I would re-state that it would make sense to take the current rendering engine, install the Embree subroutines in it, and offer it as a plug-in.

This makes some sense for me because,

1.  I'm assuming the Embree subroutines don't work on a system which does not have an intel processor.  Therefore, it only would make sense for people who could use it to pay for it and the other people would not have to pay for something the you couldn't use.

2.  Some people may not want to spend the money anyway and would like the option instead of being sort of forced into it.

Something else I'm curious about is that if there is enough passion for Embree, why don't you organize the construction of the plug-in and provide it?  Carrara has the open SDK capability to do this.  There's nothing holding people back from making it happen.  It would require finding a few people who can write the code, test it, and sell it.  Funding for such a venture would also be required but that could be organized as well.

@Kixum

Q&A Intel Embree

Q8. Can Embree run on an AMD processor? Embree should run on any CPU that fully supports any of the following instruction sets: SSSE3, SSE4.1, SSE4.2 or AVX

List of CPU's that support the minimal SSSE3 instruction set necessary to run Embree:

AMD

Athlon 64 (since Venice Stepping E3 and San Diego Stepping E4) Athlon 64 X2 Athlon 64 FX (since San Diego Stepping E4) Opteron (since Stepping E4) Sempron (since Palermo. Stepping E3) Phenom Phenom II Athlon II Turion 64 Turion 64 X2 Turion X2 Turion X2 Ultra Turion II X2 Mobile Turion II X2 Ultra APU FX Series

Intel:

Celeron D Celeron (starting with Core microarchitecture) Pentium 4 (since Prescott) Pentium D Pentium Extreme Edition (but NOT Pentium 4 Extreme Edition) Pentium Dual-Core Pentium (starting with Core microarchitecture) Core Xeon (since Nocona) Atom

Some reading regarding Embree:

Embree in C4D R15 

Embree in Simlab Composer 2014

Simlab Composer 2014

Please read carefully the links above, then we will continue the discussion further.

Please note that I am not giving into passionate and cheap fanboyism. I am making a level-headed engineering case for Embree versus Octane/Lux, based on price, performance, survival ability in the future, support by a reputable company, and all the factors that an engineering team take into consideration when making a decision for one technology versus another.

dr_berniw wrote: "All this 'over-excitment' about Octane for Carrara is similar to the cheering when Luxus for Carrara was released. Everybody was in tears. Finally a renderer that we can use to express our boundless creative talents!"

I don't see a lot of "over-excitement" here over the Octane plugin, except for maybe me. My excitement comes from actually having used Octane with both the Poser demo plugin and the DS beta plugin. My excitement comes from having actually used Octane. I’ve experienced the workflow and quality of two plugins for Octane from two programs that I’m quite familiar with. I am extremely impressed with the performance of Octane, the quality of the plugins, and the quality of the renders produce by Octane from the plugins. I also understand the limitations with GPU rendering, and the cost of Octane relative to Carrara, and have been very upfront with both. I understand that Octane isn't a good fit for everyone, but for those that it is, it could make a huge positive impact on their productivity.

I really don't see where you’re getting "All this over-excitement" from, unless maybe your referring to the thread in the DAZ Carrara forum. If that's the case, then maybe you should post your query(s) there, where some long term and more knowledgeable Octane users could respond.

Now on to the Luxus/Lux Render issue. Having used Lux Render starting with Reality 1 for DS, I didn't quite understand the excitement over having a plugin for Lux with Carrara. Don't get me wrong, Lux is a fantastic unbiased renderer. It even has some features that are "better", more accurate, or easier to use than those available in Carrara (caustics, and sss for example). But, Lux isn't the elusive magic "Make Art" button. I think the excitement over Luxus was a combination of several things including; 1) there was finally an external renderer available for Carrara just like with the pro apps, 2) many people had wanted a "fully 100% unbiased" renderer in Carrara for years, 3) there was a bit of app envy - DS had 2 Lux plugins, Poser had 2 Lux plugins, pro apps have numerous external renderer plugins, and Carrara had ...... zero, 4) I think some people thought that Luxus/Lux was closer to a "make art button" than it really is.

I have Luxus for Carrara (and for DS), and play with it from time to time with Carrara, but I can typically achieve results that I like just as well using Carrara's internal renderer much faster, and with less shader work. Lux is much slower than Carrara!!! Just like Carrara, the Luxus plugin is just another tool in my tool box. 

dr_bernie wrote: "And then it flopped. Luxus for Carrara didn't even reach version 1.0.1. I bought it and I never even installed it. There were a few images rendered with Luxus posted in RO's Carrara galleries, and then nothing."

IMHO, I don't think anyone could say Luxus was a success or a flop from the existence of renders in the galleries here. My guess is that very few serious Carrara users post images here, but I could be wrong. Luxus not reaching version 1.0.1 isn't necessarily an indicator of success either, it could be an indication that Spheric Labs did an excellent job of coding, and there haven't been any major bugs reported or any need to add more features. Obviously, if you didn't install it, you wouldn't have a clue how good or bad the plugin was. The DS plugin hasn't received a update since before the Carrara plugin was released, does that mean it's a flop as well?

dr_bernie wrote: "Considering what happened to Luxus, and considering that most plugins for Carrara don't get beyond version 1.1.x, I suspect that the same will happen to Octane for Carrara, so I'am not holding my breathe on it."

**
**Who knows?? It may, or may not be a big success, only time will tell. But this can be said for any software or plugin. The developer takes a calculated risk, if the market isn't there, then development stops.

The way you are intermingling Lux and Octane, I think you may be a bit unclear that Lux Render and Octane Render, at this point, are not "competing" products, or even close to being competing products. The only similarity between the two is that they are both unbiased render engines. That is where the similarities end. Lux is a CPU based unbiased render engine, and Octane is a purely Cuda/GPU base unbiased render engine. Some of the confusion may be from the CPU/GPU hybrid renderer being developed for Lux, and the SLG (Simple Lux GPU)  GPU renderer that was a separate projuct, but over the past year was brought in as a part of Lux development. At this point, both the Hybrid and SLG render engines are in development, not recommended for production use, and not recommended for the average user with either Luxus or Reality.

Ocatne is targeting an entirely different market demographic than Lux. Therefor you can't really extrapolate the acceptance of a plugin for Lux (or in this case the perceived acceptance/use) as a proxy for how well a plugin for Octane will do. IMHO I'm really not worried. If it doesn't do well, that's OK, because the user interface is very consistent between the Octane plugins for different applications, so I can easily use DS, Poser, Blender, C4D, etc. to continue to reap the benefits of my investment in Octane. Unlike trying to render in Poser, DS, and Carrara, where you have to know, and be able to adjust for each applications own unique shader/lighting system, the Octane plugins use the same shader/lighting/rendering system for every application that their is a plugin for. So your shader/lighting/rendering knowledge/skills are fully transferable from one 3D app to the next.

dr_bernie wrote: "Carrara's native renderer is OK. It can produce acceptable results, but it is slow and, compared to Shade's jaw-dropping renderer, it is only good for amateur work."

I think there are professional users of Carrara that might take exception with you statement that Carrara's renderer is only good for armature work, but that would be for them to respond to. However, since this is a thread on Octane render, I’ll see your video and raise you three more. Can Shade (or Embree) do this? Keep in mind your seeing unbiased render results on screen - not faster biased approximations (well, the user may be running the direct lighting setting - which does use very very high quality ambient occlusion).

Octane 4 GTX Titans:
http://www.youtube.com/watch?v=mhqf1n2xq80&noredirect=1

Octane 8 GTX 580’s:
http://www.youtube.com/watch?v=CZ1IRQTqMMY&noredirect=1

Or, Octane with just one GTX 590:
http://www.youtube.com/watch?v=OmukImTkmHY&noredirect=1

Embree might be a nice inclusion to Carrara, unfortunately you're talking to people here that have no decision making power over the direction of Carrara. DAZ3D staff seldom visit the DAZ forums, and almost never visit the forums here. You need to either take you crusade directly to DAZ by submitting a feature request, or you're interest in Embree might possibly get noticed in the Carrara forums at DAZ, but they will never get noticed here.

While we are talking about Embree - what other applications use it besides C4D, Simlab (which is actually using biased rendering to get their amazing display quality/performance), and Vray? I haven't been able to find any others. If Embree is as easy to implement as you seem to think, then why aren't more render engines using it? Maybe because it isn't that easy to implement, or?????

Also, Embree doesn't have/support volumetrics or shader based hair (like Carrara has). This may be why it isn't "everywhere". In order to keep the volumetrics/hair available you would need to create a hybrid render engine that would use the "old" renderer for volumetrics and hair - this would require quite a bit of development/engineering time and resources.

Maybe the best thing to do would be to contact SphericLabs, or one of the Carrara plugin developers to see if there is any interest in creating Carrara plugin for Vray, which is an excellent render engine that has implemented Embree.

Thanks DustRider for your reply, and thanks for pointing out the 'over-excitment' issue.

You are right, I was actually referring to the 'over-excitment' in Daz forums, not here. But since what I am saying could be deemed as disruptive, I chose to post it here.

I was about to post another lengthy article about Embree when you posted your excellent points, which will take some time for all of us to ponder. So I will wait maybe a day or two before posting more about Embree.

Please note, I am not embarking into a Crusade. This is only a level-headed evaluation of several technologies, not a Holy War.

I will post here, because I like it better here than Daz forums. If it gets noticed by Daz people, it will be a proof that they are interested in what is said about Carrara elsewhere than in their own forums, and if it doesn't, it will be one more reason to blame Daz for neglecting Carrara.

Good news!

So let me respond,

First, thanks for posting the information concerning Embree compatibility as it would apply to a very large spectrum of users.  As I stated, it was my assumption that Embree would not work on non Intel systems.  I am glad that's not true.

Second, I would like to emphasize Dustrider's point which is similar to what I've been saying which is that getting somebody to develop a rendering plug-in outside of Daz is a completely viable option (like Vray) and at this point, I would take the stab that getting it done by an outside company has a better chance of actually happening.

I would also agree that it's difficult to say if Lux was a flop or not.  My gut says that the translation of shaders makes it just a tweence of pain in the neck for people to get excited about it.  Objectively, I just don't know if Lux was a good deal or not.

I am glad Lux was done though as expanding Carrara is usually a good thing.

I would like to add a few points to my posts above. As I mentioned before, this is a level-headed engineering evaluation of different technologies, not blind fanboyism.

First, Octane and Embree are not competitors, for one good reason: Embree is not a renderer, it's a set of fast raytracing routines that reside at the heart of every renderer.

Every renderer has raytracing routines at its core. They calculate how light interact with the scene. VRay uses raytracing routines, so do Octane, Maxwell, Lux, C4D, Lightwave, Modo or whatever. In Carrara's Render Room the first option is the  'Full Raytracing' checkbox. When you check it, you tell Carrara to use raytracing routines to perform its calculations.

The maths behind these calculations are complex, very complex (here and here). They involve billions of operations. The faster and more accurately these operations are performed, the faster and more accurate the renderer will be.

If you think of a renderer as a car, then the raytracing routines are the car's engine.

The engine is what makes a car go fast, the engine is, to a great extent, what defines the car's ride quality. Same with the raytracing routines. They are what make a renderer go fast, they are what define, to a great extent, a renderer picture quality.

Embree is a set of fast raytracing routines that uses Intel (or Intel compatible) CPU's built-in math instructions to achieve the highest speed. A renderer that uses Embree, like Vray or C4D R15 or Simlab and probably Modo very soon, is noticeably faster than a renderer that doesn't.

Carrara's raytracing routines are old, very old. They probably have not changed since the time of RDS 3.0.

Why do I think they have not changed since the days of RDS 3.0? Because, as I said above, raytracing routines are complex. They are too complex for a company like Daz to write. Writing raytracing routines is not within just anybody's reach. It's in the realm of very experienced math/physics PhD's. It's rocket science, literally. For this reason it is very unlikely that Daz had the expertise to improve the original raytracing code developed by Eovia.

The age of raytracing routines in Carrara's renderer is one reason why it is so slow and of such amateurish quality. They were developed by Eovia in a time when CPU's did not have buit-in advanced math instructions, therefore Eovia's enginners used slower routines and had to copromise over the final output quality, to get reasonable performances. They did an excellent job for that time, but now there are much better and much faster raytracing routines like Embree, and it's time for Carrara to use them.

I don't believe Embree can be incorporated in Carrara's renderer as a plugin

. Raytracing routines are too deep inside the renderer to be accessible via SDK. But someone like Fenric is a lot more qualified than me to express an opinion on this.

Regardless, it is not just an Embree issue. It is also the issue of the inaccurate way Carrara processes light, which, I believe, it does only over 8 bits per channel, for a total of 32 bits over 4 RGBA channels.

To get accurate lighting, one must ideally process light over 32 bits per channel, i.e. a total of 128 bits for RGBA. That is what a world-class renderer like Shade does, and it shows in the final results.

For this reason, to implement Embree inside Carrara, and correct its lighting inaccuracies, one must have access to Carrara's source code. You can't just write a plugin, and call it a day.

In other word it's only Daz who can do it. Unless Daz is open-minded enough to make Carrara's source code available to someone like,say Fenric, to incorporate Embree in it and modify the processing of light so it's done over 32 bits per channel, not just 8 bits..

OK, enough said for one post. In my next 2 posts I will look at Yafaray (here) and 3Delight (here) as 2 other rendering possibilities for Carrara. Yafaray has been Blender's default renderer for many years and is a highly regarded open-source renderer. 3Delight is an industrial strength renderer that has been used to render numerous Hollywood blockbuster movies (here). 3Delight is also Daz Studio's native renderer.

After having looked at all the possibilities, I will make a final cost versus benefit for each solution to find-out which one makes more sense for Carrara.

Couple of things...  INTERNALLY, light and shanding in Carrara is 32-bit per channel floating point RGBA (128 bits per pixel):  it is only the output that is clamped to 32-bpp.

Secondly, Embree does appear to provide the very things that are the most onerous as far as developing your own render plugin, actually.  It should be possible to take the SDK sample renderer, rip out the very inefficient calculations it is using, replace with Embree and add Embree's GI calculations.  As a plugin, you'd want to provide two modes: "legacy" that outputs 32bpp but is network render friendly, and "high quality" that outputs HDR, but has to run locally because it will be writing the file itself instead of using Carrara's built-in final image handling.  Hair and volumetrics would still be a challenge, since Embree handles neither of those things (though at least it does instancing).

Thanks Fenric for your reply. 2 questions.

1. When you say:

Quote - It should be possible to take the SDK sample renderer...

which SDK are you talking about? Carrara's SDK or Embree?

2. Do you have a rough estimate of the time it would take for a knowledgeable Daz developer familiar with the internals of Carrara's renderer to implement Embree inside Carrara (not a plugin), including hair and volumetrics?

What I mean is the rough time estimate for a carefully tested production-quality implementation of Embree, not some useless rushed-out-the-door implementation that crashes every step of the way.

I was speaking of the Carrara SDK, from teh point of view as a Carrara plugin developer.

The Embree "SDK" is not particularly polished, lacking even basic documentation.  The "tutorial" spends far more time on OpenGL than their render algorithms - I'm not impressed.  Business as usual for Intel, though: that's the way all of their stuff comes out.  I tried to use their math kernel library, once, and that was a nightmare.

I believe that one experienced developer as hypothesized, working full time on Carrara, WITH NOTHING ELSE TO DO, could probably do it in six months.

But I would put forward that there is no such person.  They've closed the bug tracker, so we can't inspect the tickets anymore, but I saw rather a lot of lengthy Eastern European names for a company whose only physical office is in Utah.  I'm really quite certain that DAZ has a pool of contractors at an outsourced company that get assigned to whatever project the management has given priority at that time.  If they have more than one or two permanent, full-time developers I would be utterly shocked.

Thanks Fenric for your feedbacks.

I guess we both agree that Intel has designed the hardest part of a renderer, the part that's in the realm of rocket science, i.e. the raytracing routines.

The fact that Embree is already implemented in some world-class apps like VRay, C4D and Simlab is a proof positive that Embree is itself a world-class technology.

Also since Embree is already in C4D, we can be sure that soon it will be in 3D apps in roughly the same price range, i.e. Lightwave and Modo.

And since Embree is also implemented in a $249 app like Simlab, we can be sure that soon even the cheapest of 3d apps will follow suite, including Project Messiah, Cheetah3D, Shade and Poser Pro 2016.

So where does that leave Carrara? Probably in the middle of nowhere wondering what's going-on.

I wanted to review Yafaray and 3Delight as possible rendering solutions too, before running some numbers. But I will first do a cost vs. benefit between Octane and Embree. I will get to the other 2 renderers later.

I will base all my price comparisons on non-discounted list prices, for all products involved.

First let me state that Embree has been designed to be integrated in a host application, not as a plugin, and that is how I will evaluate the costs.

We agreed that Embree can be incorporated into Carrara by a developer who knows what he/she is doing in six months. I add to this a part time art student to test as the work progresses, so the final result is a carefully tested industrial-strength production-quality product, not some useless bug-ridden junkware.

I think it is more than reasonable to say that the cost of a programmer in Utah, plus a part time student to do the tests will not be more than $100K for six months, including employer's overhead and extra-costs.

Carrara Pro's list price being $549.-, the Embree implementation costs will amount to roughly 200 copies of Carrara.

I believe that with Embree implemented, Carrara's sales could go through the roof, to the point that 200 copies of it will be sold in only few days.

The same Embree could impact Carrara's sales so positively that Carrara developments could become self-sustaining through the revenues generated by its own sales.

So for a mere 6 months/$100K investment, Daz will get a Carrara with a super-fast, state-of-the-art renderer, could become the preferred tool for some small production studios, could probably run even on a laptop, and could improve Daz's sales so it doen't have to continuously discount its prices to the bone to generate some cash-flow.

In my next post I will look at the cost versus benefits of Carrara + Octane + Octane plugin + A high-power NVidia card to run Octane.

Before getting into a cost versus benefits for Octane plugin for Carrara, I would like to add one final point regarding Embree's implementation in Carrara.

I seriously believe that Carrara augmented with Embree's performances will see its sales go through the roof.

I don't think that Carrara's current sales level is more than 10 copies/week. With Embree, its sales could easily reach 100 copies/week.

Assuming that Daz continues selling Carrara at a discounted price of $249.-, this amounts to $24,900/week or $1,294,800 per year for Carrara's gross income.

With this kind of revenues, Daz can bring-in a staff of 10 people, composed of 8 full-time programmers to work on Carrara, and 2 full-time testers to use Carrara and find bugs and quirks, instead of us users doing beta-testing.

Yearly cost for a staff of 8 programmers and 2 testers in Utah? Probably not more than $1M, so there will still be 294,800 left for promotional campaigns, ads, trade shows, etc, that will contribute in Carrara selling even more.

In other words for a $100K/6 months initial investment to implement Embree in Carrara, Daz will turn Carrara into a self-sustaining product, will make Carrara PA's happy because they too will see their sales go through the roof, and will make Carrara users happy because they will get a state-of-the-art product whose survival is guaranteed by its own sales.

Therefore Embree in Carrara will be win-win-win for everybody, Daz, Carrara PA's and Carrara users.

In my next post I will look at the costs versus benefits of Octane plugin for Carrara to see whether we can expect to achieve the same results.

Just a few comments.

I think your $549 price tag is a bit, shall we say .... ambitious?  Carrara is still at the $285 sale price for the "general public" (at this point that looks like a long term sale price). Of course anyone who knows about DAZ will get a PC membership and pay the PC club price, which is $171 right now.

There are also a couple of things to consider when comparing render engines/technologies and the type of processing used. First, Octane is an unbiased render engine. in order to get as close to as possible to equivalent comparison between Octane and Embree, you will need to use a renderer that is running in a fully unbiased mode. Only then will you be even close to apples to apples comparisons, otherwise your comparing apples to elephants. Once you compare the performance of Embree in fully unbiased rendering mode to Octane you will quickly realize that there is no way that CPU rendering speeds can compare to GPU rendering speeds.

How can I make such a bold statement (and be 100% confident that I am correct)? Easy, it's simply the math and physics. Let’s do two ultra-simplified hypothetical scenarios, one for CPU processing, and one for GPU processing, to illustrate the differences in the raw computing capacity between CPU and GPU processing. Keep in mind that Embree simply optimizes the processing of instructions and computations for the render, it can’t make the CPU run any faster.

First, let’s use a CPU rendering powerhouse with dual Intel i7-3930 Hex-core processors running at 3.8GHz, which gives you 24 logical processing cores with a theoretical capacity of processing 3,800,000,000 instructions/computations per second per core, or a total theoretic capacity of 91,200,000,000 instructions per second. That’s a really impressive number, and FAST.

Now, let’s use a GPU rendering powerhouse, dual Nvidia GTX Titans with 2,688 cuda cores each running at 837Mhz. This gives us a theoretic capacity of 2,249,856,000,000 instructions/computations per second per card, or a total of 4,499,712,000,000 instructions/computations per second, or 49.7 times more instructions/computations per second compared to the CPU rendering power house. True, this is using an ultra-simplified theoretic computation capacity scenario, but it provides a good basic comparison. In the real world, things aren’t even close to this simple, and there are hundreds, maybe thousands, of other factors that would feed into a good comparison. However, this lays some basic ground work for why at this point, unless you have a beast of a multiprocessor machine at your disposal (or a render farm), CPU rendering can’t compete with GPU rendering when using the same rendering algorithms (i.e. unbiased rendering) with the same geometry and shaders in the scene.

Of course, using biased rendering in CPU based render engines definitely levels the playing field quite a bit. It’s often difficult, if not impossible to visually tell the difference between a high quality biased render compared to the same image created with an unbiased renderer (Maxwell comes to mind). But, as you crank up the “quality” of a biased renderer, render times increase and come closer to those of an unbiased render engine. So it all comes back to personal preference.

It will be interesting to read your comparison of Embree and Octane, though I do wonder where/how you will get the data for comparison. Honestly, in order to make a true comparison and derive any sort of cost analysis you really need to do testing under equivalent conditions, with a targeted outcome. Reading forum posts, marketing material, and trying to extrapolate any true meaningful scientific or engineering data from it is close to impossible.

I would hope that you will include some form of real benchmarks in your comparisons of all the different rendering software you mentioned, including the work associated with using/exporting Carrara scenes to them. Also, providing sample renders is really helpful as well. I would also hope that you would include additional factors that might address things like enhanced or convoluted workflows. As an example of why this is important, few people use only one application. If you use DS and Carrara, if plugins are available for both you can easily go between the two, using the strengths of each, and have the same shaders, lights, and render engine in each, rather than trying to work with two different sets of shaders, lights, and renderers

Also, you need to keep in mind that the Octane plugin for Carrara is in development now, and will be available soon, regardless what DAZ does with the internal renderer in Carrara. Plugins to other render engines mentioned, and the integration of Embree into Carrara, at this point are only a wish and a prayer.

I would also like to stongly urge that if you, or anyone else reading this forum, really wants Embree in Carrara, you need to communicate your desires directly to DAZ3D. The notion that if they don’t read these forums simply indicates that they don’t care about Carrara is pretty short sighted. You say you want DAZ to focus their time, efforts, and $$$ on improving and fixing Carrara. Then you say that if they aren’t spending time, efforts, and $$$ reading forum posts all over the internet, they just don’t care. Seriously??????

I think DustRider's points make a lot of sense, and I really think Carrara users should be very excited about being able to use Carrara with Octane seamlessly, with an integrated bridge plugin.

As for Embree, I think the assumption that the addition of this feature in Carrara would increase interest of the package to the point of increasing users 10x what they are now, per week, is highly unlikely, and a very risky proposition for Daz to assume.

All Embree would do is make the current engine faster.  The current engine is quite a bit outdated (although a good solid raytracer) as it is, so I can't really see where there would be THAT many newly interested users of it, just because of the speed increase.  Heck, many Poser users are using Reality with Luxrender, which is the SLOWEST rendering combination I can think of today, and are quite happy with it, because the results are awesome, with minimal manual input.  Getting beautiful renders from Luxrender/Reality is far easier than getting beautiful renders from Firefly, you just have to wait longer, and people seem to be willing to do that.  So assuming there's thousands of folks out there just waiting to buy Carrara when it's current render engine is faster, is kind of an overambitious prediction.

Meanwhile, Octane, on the other hand, provides a truly different and modern engine.  One that uses not only unbiased Path Tracing algorythms, but also can be set to do biased raytracing, like Vray, and increase render speed on scenes where you may not need ultra-realistic lighting by a factor of 4x or more.  Octane has several built-in render mode kernels to pick from, and this versatility alone is like having 3 or 4 different render engines in one.  Of course, it's GPU based, so a good card is needed for optimum results, but on average, any Nvidia GTX card that is newer than a year old is more than enough to see incredible render speeds with unbiased results.  So you can spend say $200 or so on a card, and get very fast photorealistic results from Octane, but chances are that any GTX card will provide results that are impressive to the average user of many CPU render engines.

I really think if Daz is going to invest 6 months of dev time to something, it should be effort put toward improving the current features in Carrara, like the modelling tools, and other things, which may not be bullet point features, but would certainly create renewed interest in the package from people who may currently be using Blender, or some other packages that are overall confusing to hobbyists.

@DustRider:

The issue of CPU versus GPU based renderers will never be settled. Each technology has its supporters, and we can debate it for months without any tangible results.

What I am trying to do is to see which one of Embree or Octane make more sense for Carrara.

So far these are the results that, I think, I have established:

1. Embree provides that part of a renderer that is the hardest, most 'rocket science', most onerous to design, i.e. the raytracing routines.

2. Embree is already in world-class apps like VRay, C4D and Simlab. Therefore Embree itself is a world-class product. Being a world-class product means that most likely it will soon be in Lightwave, Modo, Shade and even Poser. So not being in Carrara, will take Carrara one step closer to a dying dinosaur.

3. Embree will run natively in Carrara and will support Carrara's lighting and shading systems, including hair shader and volumetrics, for a total estimated development cost of 6 months or $100K.-, asuming that the developer assigned to the project knows what he/she is doing.

4. Embree will run on any fairly recent Intel or AMD CPU, provided that it supports the SSE3 instruction set, i.e Embree will run on anything from laptops all the way to high-end workstations.

5. Embree could make Carrara's renderer go up to 30x faster by my estimate, while at the same time improving the render quality noticeably. This is more  than enough for most Carrara users. Actually if you look at this list (link) of all the movies rendered with 3Delight, which is a CPU based renderer, you will wonder  why the producers of these movies did not use a GPU based renderer. Certainly it is not a question of budget, since these guys have all the multi-million dollars budgets that they could wish for. Therefore there must be other reasons. Render quality maybe? Or speed? Or reliability in handling large projects?

6. Embree will positively contribute to Carrara sales going through the roof, i.e Embree could potentially turn Carrara into a self-sustaining product. This will be a huge benefit for Daz, for Carrara PA's, and for Carrara users, because nobody wants to drag along, develop contents, or work with a 3D app whose future is uncertain.

In other words with Embree implemented in Carrara everybody wins: Daz, PA's and users.

In my next post I will look at whether the same can be said about Octane and Octane plugin for Carrara.

Just a few "quick" responses.

**
dr_bernie wrote: "The issue of CPU versus GPU based renderers will never be settled. Each technology has its supporters, and we can debate it for months without any tangible results."**

Actually, my last post was more about demonstrating why you can't really consider Embree and GPU rendering to be equivalent technologies, and that no matter how much Embree improves perfomance, under equivalant conditions, there is no way that current common single box cpu configurations can even come close to GPU processing power/capabilities. This is simply a matter of available instruction/computation cycles per second on each respective processing platform.

If you don't want to use GPU based rendering, that's your choice. I just wanted to make it perfectly clear to anyone reading this thread (if they haven't fallen asleep half way through it) that even though Embree might make CPU rendering faster, it will still be slower than GPU rendering, and why.

dr_bernie wrote: "What I am trying to do is to see which one of Embree or Octane make more sense for Carrara."

Honestly, at this point this is kind of academic, and sort of a mute issue. This isn't an either/or proposition. The plugin for Carrara is being developed, regardless. When it comes out, you, and every other user of Carrara, will have the option to either purchase and use it, or not. If you don't want to use GPU rendering with Carrara, great, that's your choice. But I can tell you that most people I've communicated with who have actually tried Octane find some real value in using it - your mileage may vary.

dr_bernie wrote: "3. Embree will run natively in Carrara and will support Carrara's lighting and shading systems, including hair shader and volumetrics, for a total estimated development cost of 6 months or $100K.-, asuming that the developer assigned to the project knows what he/she is doing."

Really? Everything I've been able to find indicates that Embree 2.0 does not support shader based hair or volumetrics (including an Intel generated Q&A sheet on Embree 2.0). Where did you find that it does support these?

I think integrating Embree into Carrara could be a really great thing. The question arises though, why haven't more companies used it, or announced that they are integrating it? I think quite possibly some may have already integrated many of the SSE instructions into their code, and retooling everything to include Embree to get what might be marginal performance gains for them, simply isn't worth it. Or, they may be waiting to see if the next version of Embree will include support for features that are currently missing. Only time will tell how wide the use/acceptance of Embree will be.

Regardless how much it might cost, or how great it would be for Carrara, talking about it here will do nothing to promote the idea with the owners of Carrara. You really need to present your case to DAZ (I wouldn't say this if I didn't see merit in it).

dr_bernie wrote: "5. Embree could make Carrara's renderer go up to 30x faster by my estimate, while at the same time improving the render quality noticeably. This is more  than enough for most Carrara users. Actually if you look at this list (link) of all the movies rendered with 3Delight, which is a CPU based renderer, you will wonder  why the producers of these movies did not use a GPU based renderer. Certainly it is not a question of budget, since these guys have all the multi-million dollars budgets that they could wish for. Therefore there must be other reasons. Render quality maybe? Or speed? Or reliability in handling large projects?

First, how are you generating your speed improvement estimates? Have you run actual tests that indicate this, or just guessing based on what you've read. If I were to send a feature request to DAZ, this would be an excellent point to include, however without real data to validate my claims, I could only say that based on what I’ve read, there could be significant speed improvements with the implementation of Embree.

I don't have to wonder why the big studios aren't using GPU based rendering, I know why. Here are the obvious reasons. First, typically their scenes won't fit into video RAM, and it's not uncommon to have scenes in the 16-32Gb range. Second they have huge investments in projects that run over multiple years (i.e. Pixar usually has four films in their pipeline at the same time, each one taking about four years to complete), they can't afford to change render tech mid stream as this could require re-working everything done prior to changing to the new tech. Third they need to have a stable, proven environment for their massive render farms, GPU rendering isn't there yet. Fourth, they use biased render engines to improve render speed, and create special effects in the render rather than in post - the focus for GPU renders has been unbiased, as that is where the speed increases are most needed.

Also, keep in mind that even if GPU rendering made sense for them (and at this point I honestly don’t believe it does), it would be a huge undertaking to move to a GPU render farm. They would have to either purchase, build, or rent additional facilities to build the new GPU render farm, because it is highly doubtful they would transfer any current productions to GPU rendering – there is just too great a chance that there would be inconsistencies in renders that would be unacceptable. Second, it would require a whole new team to manage the second render farm, the purchase of thousands (yes thousands) of computers and vedio cards for the new render farm, and no doubt re-engineering workflows.

I don't think anyone using Carrara has requirements similar to the big Hollywood studios. In fact, I doubt very many users even do network rendering. Octanes target demographic is more the small graphics/3D studio, architects, freelancers, and serious 3D enthusiasts. They definitely aren't targeting even medium to large studios, as Octane has no multi-computer rendering controls built into the software (render nodes).

Now really, in you last sentence your actually questioning the quality of unbiased GPU renders??? You do know what unbiased is don't you? But seriously now, there can be some background flickering issues using unbiased renders for animation However a quick look through the videos posted on the Otoy site shows that this probably isn't an issue with Octane (Si Fi Funk is already working on moving his workflow for animations to Octane)

While we're on the subject of the big studios, you are aware that Pixar has incorporated GPU rendering in their workflow aren't you? If not, take a look at this video, pretty cool tech. Set up all your lighting near real time using Nvidia Optix GPU rendering tech, then us Katana to consistently transfer the exact same lighting setup to PRman for final production. So, even though GPU rendering isn't being used for final production work, it is definitely moving into, and accelerating the workflow of the big studios.

dr_bernie wrote: 6. Embree will positively contribute to Carrara sales going through the roof, i.e Embree could potentially turn Carrara into a self-sustaining product. This will be a huge benefit for Daz, for Carrara PA's, and for Carrara users, because nobody wants to drag along, develop contents, or work with a 3D app whose future is uncertain.

I would like to know where you are getting your estimates for these sales figures. Did C4D, Vray, or Simlab sales go through the roof because of the addition of Embree? I'm all for something that will improve Carrara's sales and use. I ask because you state this as fact, so you must have some information on another similar situation. I have no doubt that it would be a great upgrade bullet point, and that it very well could increase sales. But, if I sent an enhancement request to DAZ, before I would include a claim like this, I would want to be able to include real data to back up my claim.

I agree 100% with DustRider's points in this case.  It should also be known that given some time, Octane WILL make it's way to the major studios.  It's a very young render engine, and just in it's first few stable releases out of beta.  However, it's ALREADY found it's way big-time into the arch/viz market, and is being used to some capacity there already.  Just browse their forums.  90% of the professional users there are working for major arch/viz studios, and using Octane for that purpose.  Why?  Because it's lightning fast, and you don't need to invest tens of thousands into a huge CPU render farm to get real time photorealistic walk throughs.  You can invest in a much smaller amount in some networked GPU Titan cards, and have literally thousands of cores to render with, for much cheaper.

The film studios are a different animal.  They are already heavily invested in million dollar render farms that are CPU based, and have their chosen engines already working in the pipeline.  They can't just uproot that to switch.  It's a much bigger deal then you might think.  However, it WILL happen. You will see movies being done 100% rendered with GPU farms soon enough, probably using unbiased engines like Octane.

In fact, Octane was already used to make a couple good shorts.  One in particular by a studio pro that stand out would be the Sanderson Camera short by Daniel Ahrens, where he averaged between 30 seconds and 5 minutes per frame using Octane on his GPU, and it's photorealistic all the way.  It was mainly a personal project of his, but there are others too.  So it's going to be in studios soon enough.  Just look at the quality of the short...

http://render.otoy.com/forum/viewtopic.php?f=5&t=35710&start=0

So obviously, the speed and quality are there, and GPU rendering is the future.

Speaking of real time GPU rendering, check out this demonstration of Octane being used for real time, photorealistic Arch Viz, on a RenderStream GPU render farm with 8 GTX 580 GPUs.  You have blurry reflections, blurry refractions, fresnel reflections, DOF, daylight system, and SSS, and unbiased path tracing, all at play in this example, with TONS of geometry in the scene, being done in very near real time.  Amazing technology and quality.

http://www.youtube.com/watch?v=CZ1IRQTqMMY

So don't think film studios aren't going to see the potential here to get photorealistic FX rendering at almost real time.  They are going to jump on this eventually too.

EVGA EVGA GeForce GTX TITAN SuperClocked 6GB GDDR5 384bit, Dual-Link DVI-I, DVI-D, HDMI,DP, SLI Ready Graphics Card Graphics Cards 06G-P4-2791-KR $1,115.03

Quote -  

EVGA EVGA GeForce GTX TITAN SuperClocked 6GB GDDR5 384bit, Dual-Link DVI-I, DVI-D, HDMI,DP, SLI Ready Graphics Card Graphics Cards 06G-P4-2791-KR $1,115.03

 

Thanks ManStan to talk some sense into this thread. I was preparing my next post with some numbers, but it's good that you are bringing-in yours.

You forgot to add the cost of a cooling system, a hefty power supply, an extension chassis/motherboard in case you want to bridge 2 or more of these monsters, because no motherboard that I know of can fit more than one of them, and you get yourself a system nearing $7500 to $10K, to run Carrara as a hobbyist!

Man o man! This is becoming an expensive hobby! I better start thinking of some good explanations to my other half as to why I need to spend this kind of money.

But hopefully DustRider and Maxxxmodelz will help me on that too!

BTW, this scene (link) takes over 8 GB in RAM, as seen by the Task Manager. So I would need at least 2 of the toys you mentioned to handle it.

Seems that I should have done a bit more research before posting, but I was trying to keep the CPU configuration cost as low as possible, to keep the $$$ more or less equivalent. As it turns out, the CPU I used in my example isn't supported in a dual socket configuration (looks like all i7's are single processor only). Attached is a screen grab from a basic configuration with 2 Xeon processors running ar 3.4GHz - it's only $6,769.00. You might be able to build your own for $500-$1,500 less. Going to 8-16 Gb of RAM and a real low end graphics card might get you in the neighborhood of $5,000 - $5,500. Of course the CPU I used in my example was running at 3.8GHz, and this config is only running at 3.4Ghz, you can do the math, but it’s a fair reduction in theoretical performance (a combined .8Ghz less). A system with 2 GTX Titans will actually cost less than this dual 6 core Intel processor system. In fact, you could add in the cost of Octane with the dual Titan system, and still save money, while getting over 50X more instruction/processing cycles per second.

The point of your post is??? (not sure where you got your quote from, but right now on Newegg you can get a GTX Titan (6Gb) for $999.99-$1,019.99)

If you’re trying to say it’s too expensive to get great performance with GPU rendering, that simply doesn’t hold true. I just checked at NewEgg, and you can get an EVGA Geforce 660 with 3Gb of RAM about $225.00. It has 960 cuda cores running at 980MHz – I let you do the math. It will blow the doors off of the GPU I'm running, and I'm extremely happy with my system performance.

@ DustRider:

Regarding single vs dual CPU's. I happen to have both, so I can give you some hands-on bechmarks:

1st System: single i7 920 (quad core, 8 threads) overclocked to 3.6 GHz, 12 GB DDR3 RAM, NVidia GTX 285

2nd System: Dual Xeon @ 2.6 GHz (8 cores, 16 threads), non-overclocked, 24 GB DDR3 RAM and NVidia GTX 295.

In all the test that I have done with Carrara renders and the Torque Game Engine, the single overclocked i7 CPU system has always been consistently 25% to 30% faster than the dual Xeon CPU system.

So to get fast CPU based rendering, you don't need a dual Xeon system, an overclocked single CPU is actually faster.

Besides, you don't buy a cheap $1,500 system to fit it with a $1,000 graphics card. You buy an expensive system, and you put in it a $1,000 grahics board.

The point ManStan is making is correct then: The high cost of the GPU is an added cost to the price of an already expensive system.

Quote - BTW, this scene (link) takes over 8 GB in RAM, as seen by the Task Manager. So I would need at least 2 of the toys you mentioned to handle it.

 

Well, that's 8 gigs in Carrara.  Exporting that scene through a translator to Octane might optimize things a bit.  Do we know for sure that Carrara's management of scene assets is as efficient as it could be?  Looks like some geometry in that scene could be instanced to save mem if you really wanted to, but all in all, you'd also get a higher quality of render than you have there, in less time.  I'm not knocking your render quality, it's a nice work, but there's a lot of occlusion shadows and atmosphere missing there, which I'm assuming was done to speed up render time.  Also, if you exported that geometry as an OBJ, is the size of the OBJ 8gigs?  Couldn't you reduce bitmap size to help memory too?

Quote - Well, that's 8 gigs in Carrara.  Exporting that scene through a translator to Octane might optimize things a bit.  Do we know for sure that Carrara's management of scene assets is as efficient as it could be?  Looks like some geometry in that scene could be instanced to save mem if you really wanted to, but all in all, you'd also get a higher quality of render than you have there, in less time.  I'm not knocking your render quality, it's a nice work, but there's a lot of occlusion shadows and atmosphere missing there, which I'm assuming was done to speed up render time.  Also, if you exported that geometry as an OBJ, is the size of the OBJ 8gigs?  Couldn't you reduce bitmap size to help memory too?

OK, so I spend 2 to 3 days figuring-out how to reduce the size of the scene so it renders in Octane?

At this rate I'd much rather use full GI in Carrara, which will 'only' take 24 hours or so to render.

And then what if I want to create a 100 frames animation with this scene? Do I have to adjust all of the frames, one by one? It could literally take me weeks to do it.

If I have Embree implemented in Carrara, this same scene could render in about 1 hour/frame with full GI and much better quality, so in about 100 hours, maybe 150 hours, I will get my animation done.

I understand your conerns, dr_bernie, but I have to say, seriously, if it's going to take you 2 or 3 days to optimize that scene in Carrara, then that's just further reinforcing my belief that Carrara needs a lot more done to it's modelling and workflow tools, and the render engine is something that can, and should, wait.  Especially when there's plugin technology to render in superior engines out there.

Keep in mind, that the Embree implimentation in C4D and Vray aren't major bullet points in either package.  It's almost a side-note in Vray.  Plus they don't work perfectly either.  You are restricted to use legacy GI methods with it.  The benefits are still good, of course, but even the pro implimentations have limitations and shortcomings, so it's highly unlikely to become the holy grail you are expecting for Carrara's user base.

I'm not against adding Embree, but as has been pointed out by a Carrara developer in an earlier post, it's not a simple thing, and Daz likely contracts out for all it's dev.  So that time and investment is better spent improving other aspects of the software for now.

@dr_bernie

This is simply getting ridiculous. I've been very upfront about everything I've posted, including the limitations and advantages of GPU rendering. At this point, if you want to actually learn about GPU rendering, fine there's plenty of good information on the net. If you don't, that's fine as well. But please don't post more misleadeing information about it in this thread.

I started this thread to inform Carrara user that a plugin to Octane for Carrara was being developed. I realize that for whatever reason, you don't want that to happen, and you don't want Carrara users to adopt it. I don't know if you have a grudge against Otoy, or you think that this will make the inclusion of Embree less attractive to DAZ, or ?????

I do get it (as does anyone else who looks at this forum), you want Embree in Carrara. I would even consider supporting your quest, if I have the proper real, hard data to put forth with a request to DAZ. But in all honesty, it is not a big enough deal to me to do all the homework myself. Bring me good hard data (preferably not in this thread - as it was created to discuss the Octane plugin), actually put forth the effort and make a formal request to DAZ yourself, and I will do so as well. 

You talk like money is a big issue for you, then you post the specs for your systems. They are definitely more than I can afford for computer hardware for just my hobby. You even have cards capable of using Octane, so why such a negative attitude.

I really don't get what your issue is. You start off in this thread talking about all the over excitement regarding the Octane plugin, and say that Luxus was a flop, and that the Octane plugin will be a flop as well. Then you try to hijack the thread to continue your Embree crusade here. You've tried to discourage the use of Octane, often with poor information and conjecture. I just don't get it.

I really don't understand your comment regarding your 8Gb scene either. I've been up front about the limitations of GPU rendering. You do realize that with Octane two cards with 6Gb ram doesn't equate to a total of 12Gb for your scene don't you? The max available ram is limited to the card with the least amount available. Again, if you had really done your homework, you would know that Octane includes a lot of tools and automatic functions for optimizing your scene, but as I have told you before, you really shouldn't consider using Ocatain for the big scenes you create (though an 8Gb scene in Carrara no doubt would fit in 6Gb of RAM). 

Regardless, I'm done, I have better things to do, like rendering with Ocatne.

OK. I'll start a new thread regarding CPU based renderers that could be used as rendering solutions for Carrara.

I'm done here too. Sorry for the interruption.

Dustrider, quick question for you, just to give me an idea of the current filesize limitations in Octane.  One of my favorite renders you did I think was set in Faverals Medieval Docks (I could be wrong, but it was definitely a medieval city setting) with 2 or 3 M3/V3 characters in the distant background, fully clothed, and I think 3 different V4's in the foreground, 1 of the V4 characters was captured by the 2 others who were flanking her, and I remember the captured one was topless (lol) but other than that again I think they were fully clothed with weapons and props, etc.  I'm at work so I can't scan your gallery to identify the exact image, but I'm betting you recognize the one I'm talking about (great render, by the way).

I'm wondering if a scene of that size and complexity could have been rendered in Octane or not, or whether it would have been too big.  My curiosity stems from the fact that that's about the size and complexity of scenes that I currently top out at (I don't do anything with 100 M4s for example, but it's not terribly unusual for me to have 2 or 3 V4/M4 characters with some lower-poly background people in the shot).  I'm just trying to get an idea of whether a scene like that could easily be done in Octane at the moment.

Hello,

thank you for the information. I certainly think the plugin will be relevant and not be abandoned quickly. I have used Carrara a few years ago, and Poser, and since the Octane Poser plugin is there, I'm almost only using that for rendering. 

I have a couple of Howie Farkes' environments for Carrara, and would love to use them in Octane. 

Best regards,

   Michael

Quote -
I'm wondering if a scene of that size and complexity could have been rendered in Octane or not, or whether it would have been too big.  My curiosity stems from the fact that that's about the size and complexity of scenes that I currently top out at (I don't do anything with 100 M4s for example, but it's not terribly unusual for me to have 2 or 3 V4/M4 characters with some lower-poly background people in the shot).  I'm just trying to get an idea of whether a scene like that could easily be done in Octane at the moment.

Hi,

with complex scenes, you could indeed run into problems. The limiting factor at the moment is the number of different textures. (144 color + 68 greyscale). I never hit the video card RAM limit so far with my current graphics card (GTX 680).

So if you either use the same textures, or spend some time tweaking (or use tools like texture atlas) you could go quite far. Without tuning, I usually hit the limits at about 3 or 4 M4s/V4s.

Best regards,

   Michael

Good news on the GPU memory size front.

With one time effort you can set up MASSIVE scenes in carrara which fit into 4GB. I will be proving this shortly when I release episode 8 of sci fi funk, and you can pick up a  Gigabyte Nvidia GTX760 4GB DDR5 Triangle Cool PCI-E Graphics Card for not a lot (I won't quote a price as I'm in the UK and we pay way more for everything than you guys in the USA).

Most commercial models are very wasteful in object size, and for large scenes you can usually drop the medium range textures to 256x256 without noticing a fall off in quality. Far away objects don't even need textures half the time.

Ontop of a massive scene, I've had 32 distinct characters in the scene at any one time and with replication this can go as high as 132 on an 8GB machine (CPU ram not GPU here).

I am currently waiting for the above card to arrive, suviving on a 1GB card exporting out of Carrara to DAZ and then into Octane. I've already squeezed a lot into that space (by using those cut down models).

So if you are prepared to do the prep work, and prepared to spend some time in the standalone version of Octane (to learn the complete system), all this can be yours now, with todays medium range technology.

As regards OTOYs stayability, just look at the range of 3d software it already works with, some high end stuff in there, as well as Blender, and poser based. The facebook page has 1.1million likes. This is here for the forseeable future.

Personally I think the game has changed. I'm not looking back. Time to get practising via the existing DAZ interface. The future is now.

Thanks, Michael314, that gives me an idea of what to expect.

SciFiFunk, I am eagerly looking forward to that episode, very good news that there are ways to make even a massive scene work with Octane  :)

As far as CPU vs GPU:

I don't think CPU vs GPU has to be an either/or proposition anymore.  Arion is an unbiased renderer that uses a combination of CPU + GPU working together, and I think going forward that it's likely CPU + GPU (rather than CPU vs GPU) will be the new paradigm that more and more render engines move towards.  I know that TheaRender (another unbiased renderer) will be releasing a CPU + GPU around Christmas time, and I have no doubt that Octane probably has something in the works for this as well, as focused as Octane is on speed anyway that just makes sense.

I was blown away back when Octane first came out and some of the renders (and the blazing fast render times) started getting posted on the DAZ Carrara forum.  The only reason I never got Octane was because I've never (yet) had a computer that could run it  :)    Near the same time, Reality was just starting to make it's first impact and I started seeing Lux (via Reality) renders appear in the daz forums as well, via those users of Reality that were using Studio.  Again, the quality of the unbiased renders was just freaking excellent, but the render times people were talking about was just unreal... 20 hours to a week in some cases.

So I knew I wanted to get an unbiased renderer, but found Lux way too slow and didn't have the hardware to use Octane.  Eventually I tried Thea out and found it was a pretty fast CPU unbiased renderer, so I went with that, though it did mean exporting my Carrara scenes in .obj and retexturing once I opened them in Thea again.

By the way, I don't think Luxus is at a stop; just a couple of weeks ago Spheric Labs put out a new release for it, fixing a few bugs.  I picked up Luxus a couple months back and have had some good results with it, though I have yet to master the texturing the way I would like.  Lux itself is still just so slow to render, though I'm sure I might have a different opinion if I had a computer that could run the GPU version of Lux  :)

But I'm already trying to plan out what my next rendering computer will be, to ensure that it can handle GPU rendering, and I really do think that CPU + GPU is the best model to plan for.

I think it's excellent news that Octane is planning to develop a plugin for Carrara, and I really do think it's a good thing for Carrara.  This in no way pulls resources or time from the actual DAZ team developing Carrara, so I certainly don't see any downside, and a lot of upside.  

I think this thread may have wandered a bit into the realm of theorizing what the Daz Carrara development team should be spending time/effort on, and since the Octane plugin has nothing to do with that at all it really is a bonus to Carrara users (those who can use it), but doesn't subtract anything from what the DAZ team may (or may not) be developing.

I do think that if Embree really can speed the native Carrara renderer by up to 30X, that that would be a huge selling point, and the DAZ team would be well-served to put some development muscle into phasing it in.  Among biased render engines, my personal opinion is that Carrara is pretty damn good (admittedly I use full GI nearly all the time as it doesn't seem to add prohibitive render times, at least for the type of renders I tend to do), and anything at all that could substantially speed the render times for the native engine, even 2x much less 30x faster, would be a big selling point I would think.

Jon - I don't think you need really great equipment to use Octane.

I'm running a bog standard i7 machine with an old graphics card, and getting acceptable results very quickly (whilst rendering with carrara on the CPU taking 99% of the CPU, plus an extra carrara session as an editor).

So that's two lots of rendering simultaneously plus an editing session on a mid range i7 with an older 1GB graphics card.

Well currently I have an i5 (generation 2) laptop which is several years old (3 I think) now and wasn't top of the line even when I first picked it up.  I'm pretty sure it doesn't have the graphics card capability for Octane; I know when I first picked up Luxus I downloaded the Lux version that was for GPU (by mistake) and got an error message that I couldn't run it and had to then download the CPU Lux version, so I'm fairly certain I couldn't run Octane.  Just on the off chance though, I probably ought to try the demo version of Octane to see if it's workable...

atm. I'm not sure how important the speed of the cpu is. I believe all that matters is the GPU speed (and of course its needs to be CUDA and Nvidia).

I will test this at some point with my two older machines (pre i5 what ever that series was). I will swap out my current 1GB card when the 4GB card arrives.

I suspect they will be fine.

DAZ can't even get its own features to work right in carrara, I have no faith they can do anything but ruin the carrara render engine.

The comp I'm on now cost me less then $400, meaning I could have built a render farm for $1600. And it could be done for even less if you bought DIY kits.{i5 and 6gig}

Now this may be an option for some one with the right card, or money to get the right card, or a comp that can drive it once they get it, then the cost of octane. But please take in to account how badly render engine plugins for carrara have done, which is not at all well. If some one was to take the time to develope the plugin they would never get their man hours/$ back out of it unless they charged a greavious amount for it.

Still, in the end, this would be a rather elitest render engine plugin.

Quote - DAZ can't even get its own features to work right in carrara, I have no faith they can do anything but ruin the carrara render engine.

The comp I'm on now cost me less then $400, meaning I could have built a render farm for $1600. And it could be done for even less if you bought DIY kits.{i5 and 6gig}

Now this may be an option for some one with the right card, or money to get the right card, or a comp that can drive it once they get it, then the cost of octane. But please take in to account how badly render engine plugins for carrara have done, which is not at all well. If some one was to take the time to develope the plugin they would never get their man hours/$ back out of it unless they charged a greavious amount for it.

Still, in the end, this would be a rather elitest render engine plugin.

I have to disagree with you for once here.

My 2 posts above show that Octane will run on modest cards. If you do the homework of scene reduction you can fit a lot in.

As to things not working, just try out the demo. It works and it is here now. Can you try the Daz demo for free too? If so please spend an afternoon with it before saying it won't work.

Sure there are crashes, but there are crashes in Daz and Carrara anyway. You just have to save a lot (like a nervous flinch) and progress can be made. I've lived like this on both pieces of software for 4 years before Octane came along. It's a way of life.

Please give it a try.

At present my graphics is the onboard intel 2500 HD. I have a 1gig card but it is ATI; I need to put it in but I have to swap the power unit as well, no onboard power for the card{I think}.  

So.. ;)

Still one of my CG dreams is a render farm.

Ah ok. You are stuck for now then.

Here's to a new Nvidia card!

p.s. Did you know you can get breakout boxes for housing a "farm" of graphics cards, on a single machine. Would be a cheaper way of getting a farm together.

I haven't looked into this in detail yet (having splurged on a new card and Octane), but seems to hold some good possibilities.

New card update. Just FYI if you are planning ahead for how much GPU memory you'll need.

My 4GB (4096MB) GTX 760 gives me 3786MB of usable VRAM with an empty DAZ scene. So not too bad. I try to run as little other aps as possible on a Windows Vista system.

Attached Link: http://www.youtube.com/watch?v=-GBgQspeOsg

Horror show after I followed standard advice (from DAZ, Octane, the world), of upgrading to the latest drivers.

Sony vegas and DAZ stopped working. I couldn't solve the problem, so I forced and uninstall and used the drivers that came with the card on the CD.

Solved.

Well at least your card works. After I shoehorned in the new power supply; more then this comp will ever draw{2 fans noisier as well}, the card still insta kills the comp. I can only assume that even though the card drops right in, it isn't compatable with this comp. :(

Guess I'll have to check the comp specs again and order a new card; when I come up with the $. 

Couldn't you just do a driver rollback?

I hope you can find a quick and cheap solution. Are the card and the motherboard from different eras? They obviously connect. How strange.

I should have thought of a driver rollback, but to be honest in my panic I thought I'd never get the thing working again, so I went off in all sorts of directions.

Ah well, 4 hours "invested" in driver hell is not the end of the world. Onwards!

MOB is 5 years newer then the card. The card was old enough it needed replaceing anyway.

I've checked the card and slot are both PCI Express 2.0 x16. Only difference is the card I have is DDR2 and they cards they show for this comp are DDR3, but that shouldn't be an issue. The card came out of a comp with DDR3 ram. I'm stumpped and kind of hesitent to order a new card.

Regarding the above comments.....

1. Many users have reported issues with the latest Nvidia drivers version 331.  Looks like they introduced some bugs which effect Octane, so go with the previous version.

2. Use your on-board graphics card if you can.  To do this, plug your monitor into the on-board card.  You will probably need to change a BIOS setting too - which enables the on-board card whilst you have another graphics card plugged in.

Paul