Midphase
I kinda wish that SideFX would have bought/licensed their technology instead of pursuing building a new renderer from scratch. I can only imagine just how powerful 3Delight would be if it had the advantage of being a Houdini native renderer.

Gotta agree with this. 3Delight is really, really good, and it's already working very well in Solaris. It's a much smoother experience than Karma at the moment.

Buying 3Delight makes no sense, SideFX already has a production CPU engine in Mantra - and SideFX clearly accepts that CPU engines are a legacy product - as SideFX has literally stated, they are not investing in their Mantra codebase anymore.

So, along with their investment in USD, they invested in a GPU engine. Making a GPU engine takes a very long time, just ask ChaosGroup. Pixar is already a year behind on their initial Beta projection for their GPU offering - my point is, SideFX is going for the long game, future proof design here.

GPU is the future and AI has just begun its disruption in the 3D/VFX industry.

Daryl Dunlap
Buying 3Delight makes no sense, SideFX already has a production CPU engine in Mantra - and SideFX clearly accepts that CPU engines are a legacy product - as SideFX has literally stated, they are not investing in their Mantra codebase anymore.

Exactly. A primary reason for karma will have been to lay down new foundations for a CPU/GPU renderer which would be challenging and uneconomical with the aging mantra architecture. If side fx hadn't invented a new name and instead called karma something like “Mantra XPU”* everyone would be waiting patiently and not worrying about it . It takes years to get a renderer into a refined place.

The only thing that's slightly irregular about the situation is that karma is tied to lops while mantra isn't. But a robust scene import into lops would make the UX for karma feel very similar to mantra, modulo some deeper differences like ifd, style-sheets etc.

* disclaimer - I have no idea if karma will one day be a hybrid CPU/GPU renderer!

Daryl Dunlap
GPU is the future and AI has just begun its disruption in the 3D/VFX industry.

GPU rendering has limitations which are not easy to overcome. GPU renderers are used by practically nobody in the high end studios for a reason – just ask anyone who works at WETA, Method, The Mill, ILM, etc.

They are well suited for freelancers and smaller boutique outfits who mostly work on advertising and mograph. Nothing wrong with that obviously, but not exactly the Houdini crowd.

The GPU/hybrid technology will obviously keep getting better, but so will CPU, and the larger outfits will continue to gravitate toward robust CPU rendering solutions as I'm sure SideFX is very aware of.

Regardless, Karma GPU is likely at least a year or two away from being available and solid enough for production, and with current GPU options like Octane and Redshift available in Houdini/Solaris today, I don't really see the point of SideFX investing a whole lot of energy into Karma which will be constantly trying to catch up to everyone else.

Midphase

Daryl Dunlap
GPU is the future and AI has just begun its disruption in the 3D/VFX industry.

GPU rendering has limitations which are not easy to overcome. GPU renderers are used by practically nobody in the high end studios for a reason – just ask anyone who works at WETA, Method, The Mill, ILM, etc.

They are well suited for freelancers and smaller boutique outfits who mostly work on advertising and mograph. Nothing wrong with that obviously, but not exactly the Houdini crowd.

That's an accurate assessment at the present time. But going back to karma for a moment, sidefx will want karma to be relevant 10 years from now. Even if large studios continue to run CPU render farms for economic reasons, its likely they'll want their artists leveraging GPU for interactive work if/when the productivity gains are proven and the required quality is obtainable.

Sure…why not? I mean it's customary for DCC's to offer at least a built-in render engine and Karma will likely be it in the future.

Maya adopted Arnold, C4D is making Redshift the official native renderer, personally I think those were good decisions on their part.

There is no doubt that Karma is here to stay, I just think it'll take a while before it becomes widely adopted and by then who knows what other 3rd party options will be available?

The CG/DCC industry is moving at an incredible rate of innovation, far more than any other software. 10 years from now we might be all working in Unity for all I know. I mean seriously, the way technology has swung from one side to the other in recent years is unprecedented. I'm placing my bet with Blender as taking up a serious place in the pantheon of industry adopted apps. It's not there yet, but it's getting there fast and has an incredible amount of momentum and support from some major players at the moment.

Midphase
They are well suited for freelancers and smaller boutique outfits who mostly work on advertising and mograph. Nothing wrong with that obviously, but not exactly the Houdini crowd.

Quite a large crowd of Houdini indie users might disagree here

And big studio vfx movies, globbaly it's maybe a hundred orso a year that are being produced while there's millions of ads and smaller TV productions that just have a bit of vfx where GPU is more then adequate.

Daryl Dunlap
Making a GPU engine takes a very long time, just ask ChaosGroup.

And Pixar.

SideFX so far has chosen to not buy or be bought by any company, but rather just invest into the tech themselves. This brought them into the position they are currently in. This position led why so many of us are here in the forum. So I guess we all kind of agree with a lot of their choices.

The immediate need for cutting edge render tech/times is currently not provided by SideFX themselves, but we have couple of the best Renderer integrations out there. And Houdini Indie + any Renderer costs less than the C4D + RS bundle.
So you can chose whatever Renderer you want with great integration (due to the great HDK) and have one built in production proven renderer together with one USD based renderer that aims to compete with the big boys that is currently in beta. I think we are in a good spot.

Do you not believe in SideFX being able to provide a competetive alternative, or is it the *current* lack of a fast, feature rich, stable renderer that you are worried about? Or is it the fact that you have to import your scene into Solaris to use it?

Personally I'm not worried about either of the options. Sure the current situation is awkward, but why not use a 3rd party renderer in the meantime? Give SideFX some time.

shadesoforange
Do you not believe in SideFX being able to provide a competetive alternative, or is it the *current* lack of a fast, feature rich, stable renderer that you are worried about? Or is it the fact that you have to import your scene into Solaris to use it?

All of the above!

Midphase
Note that pretty much all 3rd party render engines including Redshift, 3Delight, Octane, Arnold, V-Ray, Renderman work both as Hydra Delegates as well as OBJ context renderers with no apparent hit in performance, so the question remains as to why Karma is any different?

Correct me if Im wrong, but all of them pre-solaris existed as /out context engines, are you sure they will keep supporting two houdini scene translations in the future?

shadesoforange
Or is it the fact that you have to import your scene into Solaris to use it?

It does seem cumbersome to me. I like the viewport IPR and the easy control of multiple lights. But it feels to me like Solaris was designed for a company with departments: the modelling people finish their stuff, the animation people finish theirs, etc, and then it's off to the Solaris people to light and texture, without ever altering anything else about the scene. That workflow doesn't make as much sense when you're just one dude bouncing around between all stages of the process.

I like the viewport IPR and the easy control of multiple lights

For me it's the opposite…I like the Stage context because having multiple lights is much easier to manage and ‘Tinker’.

Mantra code base is old, and has a lot of moving parts, it was not a GPU move that forced it's fate.
Pitching Solaris meant having a working renderer to demo it. Karma makes sense that in that it's fresh, so might
as well be USD focused/centric. But OUT context is not going anywhere soon, Solaris and 3rd party renderer
final frame output is not happening pretty much anywhere yet.
Karma will continue to improve, but it's a long way off competing with what is already out there.

If you are happy in Mantra, stay there a while. Or just pick a nice 3rd party engine and concentrate your efforts
there. Karma will steadily improve no doubt, but whether it's competitive time/feature wise by then, who knows.

Regarding them licensing 3dl, well they could have used the free and open source NSI as the framework for Karma.
That would've meant already saving much dev time, and using the most modern API for DCC > renderer interfacing.

L

Daryl Dunlap
Buying 3Delight makes no sense, SideFX already has a production CPU engine in Mantra - and SideFX clearly accepts that CPU engines are a legacy product - as SideFX has literally stated, they are not investing in their Mantra codebase anymore.

So, along with their investment in USD, they invested in a GPU engine. Making a GPU engine takes a very long time, just ask ChaosGroup. Pixar is already a year behind on their initial Beta projection for their GPU offering - my point is, SideFX is going for the long game, future proof design here.

GPU is the future and AI has just begun its disruption in the 3D/VFX industry.

Oh gosh… Uncle Einstein there must be embarrassed

Just offering you a tiny bit of perspective here: at the present state CPU renderers offer more flexibility and performance, especially in complex scenarios. 3Delight has proven to be as fast – if not faster – than GPU-only renderers, it scales with complexity with zero limitations, including algorithmic ones. So if you think CPU rendering will disappear in the short term, you really need to reframe your thoughts. Stopping to blindly believe in marketing is usually a good start: think critically.

Karma, in the form it is distributed, is currently a CPU-only engine, and there is a lot of work to be done to just put it “on par” with Mantra and any other established renderer. Sidefx will surely do that, it will just take time. Karma will also use the GPU when they will feel it works well enough to be introduced / previewed, and it will most likely be not complete for quite some time as making a renderer is a damn hard job. I honestly doubt Sidefx will make Karma a GPU-only renderer, it will be most likely similar to Cycles in the way it will use the GPU, but I could be wrong on this and maybe they may really go for GPU-only to make it “stand out” as it certainly has a lot of appeal to some people. BTW, food for thought: guess who gives the data to the GPU renderer? The CPU.

CPU and GPU will continuously evolve, they both solve different problems very well. Personally, and currently, for complex and high quality output, I think nothing can beat a CPU renderer in terms of efficiency, scalability and cost per image. Nevertheless if (when) GPU power becomes accessible in a more transparent way for programmers, e.g. when you don't need to use some special compiler / some totally hardware vendor dependent API / some special 3rd party library, and also when a considerable amount of memory will be available (with nowadays scene complexity 64GB is the bare minimum, 128/256GB of RAM are needed FYI…), CPU renderer developers will be more comfortable to call GPU compute for the tasks that make sense. And when this happens a program will easily ALSO use a GPU if present on that hardware unit.

Last but not least – and more in general – the end user should NOT care whether the CPU or the GPU (or both) are behind the computation of an image. Do you care what hardware delivers to you an iCloud photo image or a Netflix video? No. All you should focus is the amount of time you spend to produce and receive an image / sequence of images (important: this includes the setup time, so complex render engines with tons of obscure parameters lose points here), at what quality, and at what actual cost. As a user you should also not be limited in what you can render: this means you should enable yourself to render either a sphere or a full frame of a super complex full CG movie, ideally with the same renderer and without any hardware nor parametric/algorithmic worries. Locally computing a render will become obsolete, big expensive boxes with many cpus and gpus become old quickly, they were (in the 90s) and are already anachronistic again (history repeats), since all non-preview rendering will be streamed as a service to the user by remote computing (the cloud), which is the true source of unlimited compute power at low cost.

BrianHanke

Midphase
I kinda wish that SideFX would have bought/licensed their technology instead of pursuing building a new renderer from scratch. I can only imagine just how powerful 3Delight would be if it had the advantage of being a Houdini native renderer.

Gotta agree with this. 3Delight is really, really good, and it's already working very well in Solaris. It's a much smoother experience than Karma at the moment.

Thank you. We are working hard to make 3Delight the fastest and most powerful rendering solution in Houdini.

I don't think Karma is that bad. I kinda like it. The noise looks actually pretty pleasing, I get less fireflies as with other renderers. USD is a useful thing for studios. Features are still missing but you have all the good Houdini bits. Some of the defaults are a little weird like the principled shader has always been slower than the legacy one and the pixel samples are a little high. I think if you play around with another shader, pixel samples and variance, and bucket sizes (why is that still a thing?) you can speed it up a lot, like literally half the rendertime in my setups. I also think that for bigger studios speed doesn't quite matter as much as for indies. You throw it on the farm overnight anyway. Who cares, it's ready in the morning. License costs are cheap and if Solaris can replace the clunky overpriced Katana even better. As for gpu support, it would be nice but those cards suck up a lot of power compared to cpus and they're basically one trick ponies.

Soothsayer
. The noise looks actually pretty pleasing, I get less fireflies as with other renderers.

Soothsayer
I think if you play around with another shader, pixel samples and variance, and bucket sizes (why is that still a thing?) you can speed it up a lot, like literally half the rendertime in my setups.

Great to hear. Would you mind sharing more details about this setup? Because I've been playing around more and more with Karma but I usually get a cleaner image with Mantra in the same time. I'm pretty sure this is mostly due to not being familiar with Karma compared to Mantra which I've used the last few years.

I'd say play mostly with pixel samples and variance. What I usually do is switch variance to uniform and convergence to path traced. Then I'll use pixel samples to find out the max samples I need. Once that is done make the variance varied and find out that maximum threshold you can deal with before the image quality breaks down. I wish Karma would make it easier to find that threshold. Redshift lets you visualize the sampling and I think with Renderman you can see the buckets dance around the relevant places if I remember correctly. Maybe I just haven't found a good way to judge this in Karma yet. I'd be nice if we get some help in that area from sesi.

I dunno, there's no secret sauce here it's just that I think the defaults aren't terribly useful.

Hi. What I am finding is I need to set pixel samples to either 512 or 1024 to get a nice clean edge on a normal pass.

With Pixel samples this high I can keep diff samples etc all to 1 (in distributed mode) Getting the best visual match so far is around 15-20 mins in Arnold and 90 mins in Karma. One very handy feature is exporting meta data from Karma in the exr. It is pretty easy to read the meta data as text in Nuke so I do not forget the setting I am using. So far I have mainly been testing diffuse indirect. I am going to try Karma against 3Delight with my scene when I get a bit of time as well. Best