There is a thread in this forum about the GPU based Octane and its outstanding performances (link).

Since I promised to stay out of that thread in order to not disrupt, I said I will start a new thread to talk about CPU based rendering solutions and their costs versus benefits when compared to Octane and other GPU based renderers.

I will start with Embree (link).

First please read through these two links: here and here.

One of them is from the Modo forums, the other from Lightwave's. Look at how positively they talk about Embree. None of them is actually even mentioning Octane or any other GPU based renderers as an alternative to embree or CPU based renderers.

These guys are professionals. Some of them could be working on multi-million dollars projects, some of them could be award-winning artists with the potential of making it to the cover of 3D World Magazine.

In this kind of business you're either quick or you're dead.

The fact that they don't mention Octane could mean one of 2 things: Either Octane is only marginally faster than CPU based renderers when you use it on actual scenes, not just carefully crafted demos, OR that the loss of quality resulting from reducing 2000x2000 hi-res textures into 256x256 super-low-res textures, possibly even leaving parts of your scene untextured, to make them fit in the GPU's RAM, takes such a toll on the render quality that the resulting picture or animation won't even make it into the ads of your local Penny-Saver or into the commercials of your neighborhood's burger-shop closed circuit TV.

True professionals don't seem to be impressed by Octane. Only hobbyists seem to think that Octane's gimmicky demos are eye-popping.

With this brief introduction, I will go into some cost versus benefits of CPU base renderers like Embree, Yafaray and 3Delight in my next posts.

You can search for Octane in the gallery, here at Renderosity, in order to get a glimpse of the capabilities of this renderer.

Ramon

I think we have to be careful not mix terms.  There are 2 different rendering engine types, Biased and Unbiased. (there may be more I'm not aware of, btw, not claiming to be an expert here)

There are 3 different methods that render engines can use to actually do the software computations on your computer to render an image.  It can use the computer's CPU to render, or it can (if you have a good enough video card) use the computer's GPU to render, or in the case of some render engines it is possible it can use both the CPU and the GPU working together to render the image.

Some Unbiased render engines (not all) and probably some biased render engines too can render using the computer's GPU.  Most render engines whether biased or unbiased (though not all) can use the computer's CPU.  There is at least one render engine which claims it can use both the CPU and the GPU to render (Arion 2, which is an unbiased renderer).

So what I'm saying is there are 2 different conversations/debates to be explored here.  Whether to use a Biased or an Unbiased renderer is a different conversation than whether it is more advantageous to use a render engine that uses a computers GPU to render rather than it's CPU, or whether it's better to use a render engine that combines GPU and CPU to render.

Since Embree is a library of ray tracing kernels that can be integrated into existing biased render engines to increase their speed and quality of their performance (I have no idea whether Embree could be integrated into an unbiased renderer too, does anyone know? My impression is no, but I could be wrong), and since Octane is an unbiased renderer that renders using a computer's GPU, to compare/contrast the two would (sort of) be comparing the biased vs unbiased, and (sort of) be comparing the CPU vs GPU approach. 

At least that's my understanding (and I don't claim to be an expert).

Thanks Jon, you made some excellent points.

I personally don't care how many plugins there are for external renderers for Carrara, biased or unbiased.

Over the years Daz has shown itself utterly uncapable of handling 3 excellent software products: Bryce, Hexagon and Carrara. Daz has literally ruined these remarkable 3D apps. Daz's own cherished baby, i.e. Studio, still doesn't look like a finished product after 6 or 7 years of development.

My goal is to not give Daz the impression that with Octane they have a good excuse to neglect Carrara even more than they already did.

If Daz cannot implement Embree in Carrara, that would be a further proof of their management incompetence, because putting Embree in Carrara is the easiest and fastest way to improve Carrara in a very noticeable manner.

By contrast revamping Carrara's aging modeler would be a lot more difficult and a lot more time-consuming.

If one expects Daz to do anything at all with the three aforementioned software programs, one should not hold ones breath. LOL.

Laurie

Quote - If one expects Daz to do anything at all with the three aforementioned software programs, one should not hold ones breath. LOL.

Laurie

Huff. Turning blue....Can't hold it...

whew! I think Laurie is right.

I'm just hoping daz gets around to fixing Genesis, AutoFit and the bugs.

"Look at how positively they talk about Embree. None of them is actually even mentioning Octane or any other GPU based renderers as an alternative to embree or CPU based renderers."

"These guys are professionals. Some of them could be working on multi-million dollars projects, some of them could be award-winning artists with the potential of making it to the cover of 3D World Magazine."

These statements ALONE are enough to make me highly suspect of how fair and balanced your comparison of CPU and GPU would be.  You are suggesting that because they work with Modo or Lightwave that they are professionals working on million dollar projects and award-winning artists?  Seriously?

Just because there's no mention of Octane in those forums, you are saying it's NOT a professional package?

What about the dozens of professional, highly regarded artists in the Otoy forums who work with Maya, 3dsmax, and C4D, who DO believe in GPU rendering as the technology of the future, and that have already used it to produce award-winning shorts, and acclaimed renders?

The cover of 3DArtist magazine, which is the UK equivalent of 3DWorld,  has featured work that was rendered in Octane, by award-winning artists, and has reviewed the software, and offered the trial of it in their magazine, as part of their included CD.  So what does that say FOR the software, if your implications that people who "could" make the cover of 3DWorld aren't talking about it?

I don't know, I just think you really need to focus more on the positive, and less on the negative comparisons and somewhat misleading information if you want to win people's opinions and get Daz's ear.

Here are some numbers to evaluate if Octane for Carrara makes any sense. I will use the list price or discounted price from manufacturer's web site (when available), or from Amazon.

1. Carrara ($285 from Daz web site) + Octane & Octane plugin ($470 at today's exchange rate from manufacturer's web site) + A hefty NVidia GTX 660 ($209 from Amazon) = $964.- OR a top of the line NVidia GPU that could run you around $1,000.- for a toal of $1,755.-

So an installation for Carrara and Octane renderer could run you anything between $964 to $1,755.-, give or take a few bucks.

At these prices you get a dying dinosaur (Carrara), Octane and its plugin, and close to nil job prospects for the Carrara expertise.

2. Shade Professional ($499 from manufacturer's web site) + Poser Pro 2012 to be used with Poser Fusion ($138 from Amazon web site) = $637.- .

At $637.- Shade Professional + PP2012 offers everything that one could wish for in a midrange 3D app , including Daz/Poser conternt support and a world-class renderer. The job prospects for Shade expertise are not great, at least not in the US, but the price is very good for a hobbyist.

1. Lightwave ($1,495 from manufacturer's web site) + Poser Pro 2012 to be used with Poser Fusion ($138 from Amazon) = $1633.-

Lightwave+PP 2012 is actually cheaper than the $1,755.- Carrara using Octane on high-end NVidia board. Lightwave is a world class 3D app, used by prominent production studios worldwide, and the job prospects for Lightwave experts in many parts of the world are very good.

1. Cinema 4D Broadcast Edition $1695 + Poser Pro 2012 to be used with Poser Fusion ($138 from Amazon web site) = $1,833.

Cinema 4D Broadcast edition (which has Embree integrated) + PP2012 is only marginally more expensive than the $1,755 Carrara using Octane on a high-end NVidia GPU.

Job prospects for Cinema 4D? Please don't argue with me on that.

So from where I am standing, a dying dinosaur called Carrara + Octane & plugin are more expensive or about the same price as world class applications like Shade, Lightwave or Cinema 4D Broadcast edition, all of them used in conjunction with PP2012 for PoserFusion functionality.

There are still other issues to be considered:

1. With Octane, in order to fit your scene into the GPU RAM, you may have to reduce your textures resolution from 2500x2500 to 256x256, eventually leaving parts of your scene untextured. You may also need to decimate your meshes to reduce the poly count, so it all fits in your GPU RAM. So what's the point of using high-res textures and meshes to get photorealistic renders in the first place?

2. How likely is Octane which, after several years of development, is still at its 1.2 version, to survive and make it into version 2.0 and beyond? How likely is the Octane plugin for Carrara to ever mature into a version 2.0 and beyond? What if, G-d forbid, something happens to the lone plugin developer. What if he is suddenly unable or unwilling to do any more work on the plugin? Sure, the plugin will be perfect in its 1.0 release and will work flawlessly everafter. Sarcasm is intentional.

By comparison Embree's survival prospects are very good since it is part of Intel's core technologies, it is already in its 2.0 version, and it is open-source anyway.

From where I'm standing, Octane for Carrara may be a cute toy to play with when I have nothing better to do, but as a useful tool in my workflow that I can use on a consistent basis day-in day-out for some serious projects? Hmmmm, I'll think about it.

This just in:

Lightwave 11.6 is priced at $995.- during the Holiday Season (link).

In other words a state-of-the-art Lightwave installation + Poser Pro 2012 will only cost you $995+$138 = $1,133, i.e. actually a lot cheaper than Carrara + Octane + a high-end NVidia board to run Octane ($1,755, see previous post)

At this low price Lightwave comes with PDF manuals, training materials and contents.

I'm sorry but the case for an Octane renderer/plugin for Carrara simply can't be made.

dr_bernie, I think you're forgetting to factor a few things into your comparison. 

If you just choose Lightwave by itself, you're stuck with the native Lightwave render engine.  I believe it's a biased engine, and uses CPU to render.  And this part is subjective, but from what I've seen of Lightwave renders... meh, not very good IMO, at that price level I'd much more likely be shopping for 3ds, C4d, Modo, or Maya.  (honestly, if there was one rendering software out there that I think deserves the 'dying dinosaur' moniker, I would pick Lightwave - just my opinion, not trying to hurt any feelings for any Lightwave lovers out there.  I took a modeling course in which the instructor used Hexagon, Wings, Blender, Carrara, and Lightwave, and watching the videos even though the instructor was in love with Lightwave and praised it to high heaven, all I could think was "thank god I'm not saddled with that piece of..."  I was very underimpressed with Lightwave, but I know it's on of the industry-standard apps, though I can't think why that might be other than force of habit).

Now if someone is content with Lightwave's biased render engine results than that might fit perfectly, but if they are weighing the decision against (theoretical) Carrara + Octane, then it stands to reason they are probably interested in an unbiased render engine, a render engine that can run on GPU, or both.  Lightwave as-is will not fit that bill.

A quick glance on Octane's site shows they do offer a plugin for Lightwave (which demonstrates that there must be a market of Lightwave users who are not satisfied simply with the Lightwave native renderer and instead are willing to pay for Octane), so to get to the same results you have theorized above, you have to pay for Lightwave + Poser Pro 2012 + Octane + Octane plugin for Lightwave + a high-end Nvidia card to run Octane.

So really, apples to apples, it seems to me the Carrara + Octane would be the much less expensive way to get to this result.

Also factor in that most of us discussing this in the Carrara forum already have Carrara and some familiarity with how to use it, so jumping into Lightwave also means devoting hundreds of hours learning how to use a new app (time/frustration cost), and I think it becomes a little more clear why the developers at Octane are willing to spend time/effort developing a plugin for Carrara rather than simply saying 'you Carrara guys should just go buy Lightwave and learn it, and then use our Lightwave plugin'  (not as likely they'd get as many takers from the Carrara userbase).

Jon,

In another thread I believe you mentioned that you are running Carrara on an i5 laptop. Can you please explain how you plan to run Octane on it?

Lol, I oh don't have any plans to run Octane on my current rig  :)  As far as I know, it can't be done (maybe someone with tech expertise to build laptops from scratch could somehow do it but it's beyond my current skill level).

On the other hand, I am planning at some point in the future I'll need a new rig anyway (my laptop is 2 -3 years old now, and works just fine, but nothing lasts forever).  Next rig I get, I'll be factoring in the possibility of GPU rendering into the equasion (I don't necessary feel the need to get a Titan lol, but it would be nice to have one of the lower priced Nvidia cards so I have the option).

Assuming my current laptop doesn't unexpectedly die on me, I probably won't be looking at getting together the new rig until middle of next year at the soonest, and who knows the whole landscape may have changed by then. 

When I first started seeing renders from unbiased renderers I decided I had to have an unbiased render solution.  Octane was too much for my laptop and Lux was too slow, so I've picked up Thea, and I really like the quality of the renders.  I'm not saying Unbiased is better than Biased because I feel they both have their strengths, and for most circumstances I can render full GI using Carrara's biased renderer and it works for what I need, but there are other circumstances where only unbiased will do, so I think it's good to have both a biased and an unbiased renderer in my stable so I can call upon the 'right man for the job' when needed.

The Luxus for Carrara plugin works pretty well, even better with the newest release, so I'm going to keep refining and learning how best to texture with it, and I'm aware there is a GPU version of Lux now.  There's already a GPU version of Thea, and soon to be a CPU + GPU version, so even if I never get Octane, I can still safely say that at some point in my future I'll want to have a computer that could do GPU rendering, which is why this topic is of such interest to me.

Now the laptop I have works fine, and the renders in Thea don't take very long, even though it's CPU rendering, neither to the renders in Carrara, and of course I could simply say 'good enough'.  But I think it's only human nature that we all want our renders to be faster, as fast as possible :)  This is why I'm 100% behind the idea that DAZ should incorporate Embree into Carrara to speed the rendering, and why I'm also eager at some point to upgrade my machine to be able to take advantage of GPU rendering technology.  Hell, I think even if we all got to the 'rendering singularity' where every program could render even the most complicated image in only 1 second, we'd all be complaining about how we wished we could get our render speed down to half a second  :)

erm. Just wanted to point out that Octane gets a mention on the links provided at the start of this conversation. On the last of the links by this person, at this point in time.

09-08-2013, 06:16 PM #11

erikals

Lightwave Cop

I hope Embree is put into Carrara. Any thing to speed it all up.

As to GPU vs CPU. Check out the progress in GPU speed over the last 2 years vs CPU. I am doubling my speed and quadrupling my memory (to 4GB) for slightly more than the card I bought back in 2011. CPU wise I looked at an upgrade and could only get a 30% increase for the same outlay as before (which incidently was 2x the cost of changing my GPU).

As to whether Octane will survive, not sure. There are lots of competing unbiased renderers out there. Thea for example, at slightly more cost. There is room for a few pieces of software to survive short to medium term imho.

I guess eventually one becomes the standard.

re: carraras future. It gets further behind. 8.5 was a bit of a joke, apart from the content package that came with it. Now that was value!

Jon Stark wrote:

*Quote -*This is why I'm 100% behind the idea that DAZ should incorporate Embree into Carrara to speed the rendering

SciFi Funk wrote:

Quote - I hope Embree is put into Carrara. Any thing to speed it all up.

Thank you guys. Now we're getting somewhere. As I mentioned above, I have nothing against Octane or those who like it.

My concern is that Daz will use the enthusiasm that is being built-up around Octane's availability for Carrara as another lousy excuse to not do anything about improving Carrara.

A problem with a stictly based GPU renderor like Octane is it does not support displacement maps because NVIDIA does not yet. That is the reason that true professionals shy away from it. Displacement maps are key to a professional workflow.

Quote - A problem with a stictly based GPU renderor like Octane is it does not support displacement maps because NVIDIA does not yet. That is the reason that true professionals shy away from it. Displacement maps are key to a professional workflow.

This isn't entirely accurate information.  First, Octane is not being shyed away from by professionals.  Octane is being looked at very closely by professionals, and Otoy has partnered with Autodesk, and many big names in the industry, who have a close eye on the development.  There's many professionals waiting to use Octane in their pipelines (just read their forums) but have to wait.  It's not because of the lack of displacement mapping either.  Otoy has stated the software is not ready for production YET, and recommend it not be employed in a studio workflow at this time.  This is really why it's not been used yet in pro pipelines.  It can do Normal mapping, which is used more in professional gaming studios for real time dev than displacement mapping, so that's not the reason.  The reason is maturity.  It's not ready yet, but it soon will be.  For reference, Vray, which is now the most highly regarded and utilized professional rendering engine, was not used widely in production houses until well into it's first major point release, so this isn't uncommon.

Otoy is working on preparing Octane for studio pipelines.  They are working on cloud streaming technology, and network rendering on a cluster.  Given their public partnerships with major software companies, there's going to be a lot of studios interested in this when it's ready.

Also, modern GPU are already capable of getting displacement mapping using the geometry shaders for auto tessellation, but the industry is still developing games for the Xbox360/PS3. So perhaps the next gen consoles, like Xbox One and PS4 will most likely introduce the use of displacement mapping to the average games.  This will also probably be introduced to Octane in the future, but that's not the most important thing right now.

The issue of CPU vs. GPU or unbiased vs. biased seems to be a very personal one.

When I look at the best of Lightwave renders (link), I like them better than what I see in Octane or Lux galleries.

But it's all a matter of personal preferences. At the end of the day, beauty is in the eyes of the beholder.

.. plus in the ability of the artist!

Ah well back to practising, practising, practising.

Right now at this point in history I think having your best set up for biased and unbiased is a good option, as they have different strengths.

Personally I'm going to push on with unbiased to get those skills up to speed with what I can do in the biased world.

Good luck to all which ever path you chose.

I think this all depends on the user, and their goals.

If you were planning on getting "into" the industry, Lightwave is acceptable as a choice, even without the fancy rendering, after all, if you are a modeller, its best to get proficient with the tools, let they guys doing the rendering worry about the shaders and lights.

You won't get into the industry using Carrara. You will probably end up being a professional hobby Illustrator or animator (or like me, get an occasional NASA render published or book Illustration gig).

Hobby isn't that bad but you need to do everything, Model, Texture, Shade, Light, animate if needed, sound, FX's... its harder, as you end up as good as your weakest skill.

In the industry, you will likely only do one thing from the above list, and you need to be really good at that one thing, others will be taking on their specific skill and taking your work to their professional standards.

So go Lightwave if you plan on being a professional modeler at that price.

The fact that they don't mention Octane could mean one of 2 things:

If  you check the galleries at http://render.otoy.com/forum/viewforum.php?f=5 I think you'll agree that a) there are a lot of professional visualisation artists using Octane, and b) the quality of their work is absolutely stunning.

One of them is from the Modo forums, the other from Lightwave's. Look at how positively they talk about Embree. None of them is actually even mentioning Octane or any other GPU based renderers as an alternative to embree or CPU based renderers.

I am currently developing the Modo plugin for Octane, and the forum post about it at the Foundry is 8 pages long (as opposed to the 1/2 page Embree post), and there is no mention of Embree in the Octane thread.

1. With Octane, in order to fit your scene into the GPU RAM, you may have to reduce your textures resolution from 2500x2500 to 256x256

I have never heard of a user needing to reduce from 2500x2500 to 256x256.  VRAM limitations are very rare - and the new 700 series cards come with between 2GB and 6GB of VRAM, and there will be a 12GB card available shortly.

2. How likely is Octane which, after several years of development, is still at its 1.2 version, to survive and make it into version 2.0 and beyond?

Otoy have announced the feature list for 1.5 and that it's release is immiennt.  There are also threads on the Otoy forums regarding 2.0. 

In another thread I believe you mentioned that you are running Carrara on an i5 laptop. Can you please explain how you plan to run Octane on it?

Your CPU and RAM as of little concern when running Octane (you just need enough RAM to be able to load the scene into the graphics card).  I run a laptop, use the on-board graphics as the display adapter, and using the GT 640M for rendering.  When rendering in Octane, there is virtually no rendering load on the CPU.  In fact, with the Modo plugin, I can simultanteously have the OpenGL viewport, Modo render viewport and Octane viewport all running at the same time and move through the scene etc with minimal lag.

A problem with a stictly based GPU renderor like Octane is it does not support displacement maps because NVIDIA does not yet. That is the reason that true professionals shy away from it. Displacement maps are key to a professional workflow.

Not sure what a "true professional" is!  I would say the majority of the users of my Octane plugins are people who make a living from rendering - does that make them a "profession" or "true professional"?  Most users who have to use a displacement map convert it to vertices or use a normal map - because "true professionals" know how to do that :-).  But I agree - displacement maps would be a big bonus, and Otoy stated this week the displacement mapping was definitely happening (in the "Features Poll" post).

Paul

@ face-off,

You have a vested financial interest in Octane, since you are developing a plugin for Modo.

Oops! My post got truncated for some unknown reason. So I repost it.

@ face-off,

You have a vested financial interest in Octane, since you are developing a plugin for Modo. Your opinion could therefore be biased (No relation to biased/unbiased renderers!).

The point I am trying to make is that Carrara ($285) + Octane and its plugin ($470) and a hefty NVidia board to run Octane ($1,000) will cost $1,755.- This is an awfully high price for a Carrara setup.

Lightwave at $995.- is a lot better deal, considering all the professional doors that it opens.

If Maxon follows Lightwave's example and drops C4D's price by $500 for the Holiday Season, then C4D Broadcast Edition will cost $1,195.-, with Embree integration. At this price you get a world-class, highly regarded 3D app used by some of most prominent production studios.

No matter how I look at it, I can't find a justification for using Octane with Carrara, based on sheer numbers.

I have tested Octane, Thea, Lux on my Intel i7 930 cpu machine.

I have a GeForce GTX 660.

I have 4 core for 8 threads with 6 gig of ram. (need to double that)

My graphic GTX660 as 5 cores 960 threads with 2 gig of ram.

Depending on my render settings of certain scene, the GPU render is faster and on some others i don't bother and stay with CPU rendering.

But the cost to upgrade the CPU for something more recent compare to upgrade the video card, cheaper changing and adding video card until i can really upgrade for a new system.

To be honest, i prefer a rendering that would give both options to run one of the other type of rendering, or both at the sametime.

AS always, depends on the work, the scene and the load in your 3d graphics software.

Each as strong and low point.