If CUDA is a proprietary hell, then Metal or anything mac specific would be the deepest circle of hell. IMO it would make far more sense to support GPU rendering with AMD GPUs for all operating systems (win, mac, linux) or even better something like Vulkan, which could potentially use the upcoming Intel GPUs as well.
MartinTS, I agree with you about Vulkan. But the Khronos folks — open source people — have noted Apple’s refusal to embrace it with macOS and iOS.
Apple is pushing Metal to the Pedal. Hellbent on getting developers onboard using it as they believe it optimizes their hardware…oh and the little fact devs will make Apple-only products with it.
Apple has been detached from other Khronos initiatives as well.
I too have dabbled with Blender 2.8 and the fact is that there are some compelling things there. It’s inevitable it will siphon off some c4d, mood, maya, max, Houdini users.
All the commercial vendors are going to need to continue to monitor their value proposition.
Yeah, you’re right. It was just wishful thinking one my side. An open source, platform agnostic, modern 3D graphics and computing API, it would make too much sense
I invested some evenings into trying out Blender 2.8 because as a hobby user, the Prime MSA is still a lot of money. Sure, Blender 2.8 has loads of features and the new Interface makes some sense at last. But some simple things (especially in the field of modelling and scene layout) are still so cumbersome that I still can’t imagine using it on a more regular basis.
Regarding ProRender: maybe it is not dead yet. Its platform agnostic nature has a lot going for it and they also showed improvements at NAB (https://community.amd.com/community/radeon-pro-graphics/blog/2019/04/09/amd-radeon-prorender-nab-2019-updates-include-plug-in-for-blender-280-beta-ai-accelerated-denoising-and-more). I guess R21 will show if Maxon can push this towards a valid solution.
I just hope Maxon can clean up their render business a bit. With Standard, Physical, ProRender and Redshift they now have 4 options and I wonder how much overhead this produces. Also, as a Prime user I wonder how much this (two additional engines to develop) will reduce Invetsments into the Standard render even further. I guess one Engine for all Editions but with restricted feature sets in Prime / Boradcast would make the most sense. However, for the time being, Redshift could not be that engine. But I could imagine that they are working on AMD and even CPU support (which would mean a move towards OpenCL). Not only would that make it an option for THE internal C4D engine, but would also lead to an even larger user base.
Well no… Cuda only works on Nvidia. Nothing else. Despite the fact that Nivida cards are better this is not great for anyone.
Metal, while it only works on Macs is the abstraction layer, like Vulkan, D3D, OpenGL, and Should with whatever GPU you put in there ( so long as there are GPU drivers in the first place )
Vulkan and Metal are similar enough Low overhead APIs that it’s Not impossible to develop them concurrently.
MoltenVK is infact Vulkan that runs on/with Metal.
Anyway this is all very boring and hopefully shouldn’t matter in the long run - Octane has set the ball rolling with proper platform independent rendering and I am 100% sure that Redshift and other will develop the same to compete.
Hence I said this would mean a move towards OpenCL. Not sure that Vulkan alone would be the answer as this is more a replacement of OpenGL and is more focused on rasterization (Vulkan Compute does not replace OpenCL afaik). You still need OpenCL for GPGPU similar to CUDA, if I am not missing something. Unfortunately, OpenCL lacks quite a lot compared to CUDA from what I hear…
Regarding Octane: from what I remember, this is also CUDA based, but Otoy developed some transpiler thingy that creates OpenCL code… Anyone knows how well that works?
Octane are completely rebuilding the core
“As well as unifying development, the RNDR framework includes new backends for Direct3D, x86 CPUs, and the Vulkan and Metal APIs.”
however… the bottom line says it all about octane
“However, it’s worth noting that the company has been over-ambitious with its timescales in the past: a number of the features on the roadmap for OctaneRender 3.x and 4.0 it released at GTC 2016, then scheduled for 2016 and 2017, still aren’t available in commercial builds of the software.”
OpenCL (along with OpenGL)has been deprecated on Mac OS, and may not even be in Mac OS 10.15. No further development. The writing is on the wall for it, I’m afraid. It’s an ignominious end for what was once a promising technology. Even though they invented it and shepherded it into the industry standard it was, Apple has never been one to coddle technologies for the sake of backwards compatibility, for better or worse.
Apple is all in on Metal, and it makes sense (for them, not saying I endorse this) that they focus on a technology they can control and ensure is performant on their hardware (especially the GPUs they are developing for their iOS devices.
Some apparent signs of sanity in Maxon in last times. But i am afraid this was more a move to deny an opponent than a move to build something. The resources spend by Maxon in rendering in last years are huge for what they have to show for it: ZERO
As an ex Softimager - mostly forced to use 3Ds Max - that always liked Cinema 4D appearance and weaving between C4D and Blender i am moving more and more towards a Blender solution . This might be shocking but with animation nodes and many other tools it seems much more straightforward, Cinema 4D lost it appeal of being that. C4D Fields is very powerful but its use is cumbersome and not clear at all. Then there are all kinds of “tricks” one have to get and takes a lot of time to to not go into a dead end - i see them many times with tutorials by real C4D “aces” : don’t do this , don’t do that because might happen this…
Cost obvious is also a component of this. If i want to do 3D printing i have to have sculpt, then i have to buy Studio which is the most expensive solution… Just does not make sense.
Edit to add: Prorender is also a case in question, AMD just released last version for Blender . If Blender always get it first where is the point for Maxon to invest in it? they will get always after Blender.
I honestly disagree. Have you tried ZBrush? It’ s a very unforgiving piece of software. Miss a step, move the wrong slider and you are in for total
I really wish ProRender would get more updates in Cinema, and not be limited to the C4d yearly release. I saw the latest version get released for 3ds Max (I think) and it had a huge library of 300 shaders, Substance and Mixer support, and AI accelerated denoising. All would be welcome in C4d pro render.
Now I wonder where ProRender is going given the purchase of Redshift. The optimist in me says Maxon and Redshift devs know RTX drivers and CUDA support is coming in Mac OS 10.15, but they can’t say anything yet (perhaps tied to Mac Pro release?)
The pessimist in me says that if Apple doesn’t straighten out nVidia drivers and Cuda on 10.15, I think they are truly backing many Mac 3d artists into a corner where they simply must use Windows.
Ah, it’s a shame. I love using Macs, but dang.
Yes. one of the reasons i don’t use it.
I don’t know why you disagree when your example is one of most extremes. For me if Blender 2.8 would not have appeared i probably would not even consider it.
Not real “options”,
Standard and Physical are able but unfortunately derelict and terminally slow.
Prorender is immature with a long list of can’t do, not particularly quick and noisy.
Only Redshift is mature, production ready, and close to be an industry standard in GPU rendering which is immensely valuable for Maxon.
Yes, unless you are just doing stills or some simple stuff the money should also include an external render engine: Redshift, Vray, Cycles, Arnold etc…
Utterly depends what you are doing. Standard and physical have a lot of uses. I use the cel render a lot and sketch.
I’m hoping this amounts to a solution other than Cycles4D for XP rendering. Cycles 4D seems quite powerful in this regard, but for some reason the demo and the demo content didn’t inspire me enough to want to learn it. I find it very slow - even on a multi GPU setup, even with 40 cores on a dual Xeon Gold; it crashed constantly on my own tests and made me want to turn my attention; should I be in the mood for a steep learning curve, toward Houdini and vex.
Give me the same power with Cycles4D on a fast gpu engine like Redshift, throw in point based particle rendering with motion blur, give us the capabilities to really tap into XP, mograph, fields; things that the ugly C4D native render engine can “kinda” do; with all the beauty pass power of Redshift, and I for one, will be more than happy to throw money at them.
Good point on sketch/cel, i forgot it.
I used Cycles on something recent and was very happy with it. I don’t have any GPU facility so it rendered on the CPU. Averaged about 2mins per 1920x1080 frame
But Cycles recent announcement and the Redshift acquisition have really got me thinking seriously about a pc . I think it’s highly unlikely Apple will come up with an Nvidia friendly machine.