I hope Maxon watched this and learned a lesson.
In Cinema 4d it is impossible to navigate in scene with 1 000 000 polygons separated in few thousand objects, I’m not even talking about any deformations ( displacer, skin, xpresso, hypernurbs, pretty much anything). If Maxon want to compete with Luxology R15 has to be few times faster in almost every aspect and few hundred times faster in some, otherwise price/performance ratio will be very very bad on maxon side.
I hope Maxon watched this and learned a lesson.
Viewport speed, or lack of it, is just another nostalgia-feature of Maxon. We are “happy” to pay a yearly increasing price of tools that never get updated, be it SnT or Pyro or TP or viewport speed or Cloth or whatever. For example, over the years I must’ve paid way over 1000 for Pyro, and quite frankly I can’t afford to see it updated.
A new viewport would presumably double the price of C4D…
Been banging that drum for a while (not in comparison to Modo though). I’d take a performance overhaul over quite a few of the latest R14 features (sculpting and the phys render for starters - I expect to take some flak for that :)).
My guess is that performance enhancements are not glamorous enough to attract new buyers, so Maxon has to balance what’s best for the business and pleasing existing users.
I’d be interested to know what the technical challenges are involved in giving a reasonable viewport speed with large object counts, I’m sure it’s a resource heavy update.
Object handling has become a major workflow hurdle, you end up spending a lot of time optimising a scene to run realtime in the viewport. Mainly an issue for me when working on previs, but it applies eslewhere too.
A bit unfair this as previous modo versions have been terrible with complex scenes, and there’s been plenty of progress from Maxon - render instances make scenes that used to be grinding become fairly fluid.
But the live, rendered preview that lets you navigate through it easily is awfully compelling.
Does anybody know how big is Luxology team compared to Maxon? It seams like Luxology is substantially bigger.
maxon got 16 dev members (according to “who is maxon” video quick tip)
luxology got 8 (http://www.luxology.com/company/key_staff.aspx)
btw. im seriously impressed by the modo update :drool: …maxon have to do so with next update :bounce:
That is a very impressive video, but I do have to pull you up on the numbers of your claim. If I make 3500 objects in c4d with a total of a million polys and apply a 2k texture to them, im able to navigate at a healthy 36fps on a 2560x1600 screen. This is on a 5 year old i7 920 cpu and geforce 580
Bump this to 4.5 million polys and I get 33 fps
17 million, 31 fps
64 million polys 10 fps
These are perfectly healthy and usable numbers. Whilst the video shown is impressive, does it serve any real world purpose? Nobody has projects with 2.7 billion polys in them, if you have such a scene then you’re just incredibly bad at your job.
Don’t take this as me saying I think the performance in c4d is good enough, I dont think it is, but regardless of how fast it gets, I dont think anyone will ever be satisfied.
It appears like opposite.
I was afraid of this situation after maxon rewriten some core functionality like getting rid of modularity and adding double precision stuff. It was obvious that if in this update we won’t see any major speed improvements we will have to wait to another major core redesign for this. Personaly I think we have to wait at least 2-3 releases to expirience any significant ( like deforming 2 000 000 poly mesh while previewing it in GPU accelerated view with Gi and SSS) improvements in performance/functionality field.
Ok, ok, maybe I’m exaggerating a bit but most of the time when your viewport is struggling to keep up CPU is only 25-30% loaded.
I don’t get this complaint at all. What hardware are you using?
In R14, I’m currently recording (w/camtasia) while I work and navigate smoothly in a scene with 37 million polygons which is full textured.
Keep in mind, if youre on a quad cpu machine, that 25% is 100% of 1 cpu core. Most actions in most apps are still single core bound. One of the worst choices I see some people make when picking hardware is to go for 8, 12, 16… core machines, but then pick a slow speed like 2GHz to achieve this. If you want the fastest most responsive machine, go for GHz, then overclock them some more. Theres little reason, at least for windows users, to not be running a good 3.5, 4GHz+ machine.
c4ds viewport speed is sufficient. if you have problems with your viewport speed
you do most likely need more ram. given the facts that previous modo versions are
well known for their really bad viewport performance, the short film ruin relied heavly
on instancing and that cloning that ship took almost 10 seconds, the whole video
seems pretty pointless to me. it is just some cached heavily instanced polygon data.
don’t get me wrong, i am neither a luxology hater, nor a maxon fanboy, but the luxology
hooray videos are just stupid. would have been somewhat more believeable if they
had shown something uncached and not used such nonsense numer as 3e9 polygons.
That’s what I did when I switched to Windows for C4d. I got the latest Core i7 at the time and overclocked it to 4.2 Ghz (4.5 when it’s in turbo). It’s fast as hell.
Rendering will NEVER be fast enough, though! I basically want Unreal Engine or something real time to work with to be happy.
Back on topic - I am also watching Modo very closely. Next time Maxon needs to work on particles. I want to be blown away with R15’s particles. The dynamics are there. I want to see it happen!
It’s old Intel i5 with 8 gigs of ram and GTX 275.
I did some tests and I’m able to navigate in scene with 70-80 thousand objects, total 6 mil polys, quite smoothly, but there are some places where cinema is extremly slow.
This is some xpresso fun, play it and watch cpu usage.
I’m pretty new to C4D, coming from 3ds Max. The biggest thing I noticed with C4D is just the horrible viewport speed. I can’t even get realtime playback to check my animation when I add some grid cloners in there.
What are some tips/optimizations I need to do in order to improve playback? I basically have to hide/solo layers, switch to Low level of detail, etc… but it’s just SO SLOW. I’m running Xeon e5 2.5GHz with 64GB ram and quadro 6000 card.
If you have a look at the latest release of Modo and all it’s updates and features and pricing you start to wonder why Maxon charges such insane amounts for their flag ship. Clearly, Modo has outrun C4D in many areas and I think most people predicted that not too long ago.
Don’t get me wrong Cinema 4D is capable of producing great things, but it desperately need some boost. Xpresso is first in a row in my opinion, if you look at xsi ICE, with speedy snappy multi-threaded nodal system deeply integrated into software core making custom tools become possible. Also OpenSubdiv is open, gpu rendering is common and life is short, so give it some boost maxon!
the market can charge huge amounts for thier flag ships because Autodesk keeps the prices high.
just check their website for product prices and/or bundles.
Same thing I keep ranting about, made a similar post here just a couple of days ago.
I can not only navigate a 25 million polygon object just fine, but sculpt a 25 million polygon object, of course I also have the 64 gb of ram to allow that large of a subdiv.
In terms of performance theres also the balance of number of objects to resolution of each object. C4D’s always been terrible in terms of number of objects. I’ve been cleaning up the old blog to start posting again and the post I’ve been working on is some comparisons about cpu performance in C4D. With overclocking tool being so easy and flexible nowadays it’s finally easy to compare the exact same system at 3.2ghz and at 4.4ghz. So I’ll be comparing those as well as huge numbers of clones as is, as render instances, and individual instances and render instances, as polygon objects, and combined into a single polygon object.
My key rule though is combine, combine objects whenever possible. Ideally with UV tiles but if necessary with polygon selections as well. Yes this can make it a little more difficult to deal with the individual components of an object, but honestly fill selection makes it super easy to select individual model shells within a single object. The file size and viewport performances difference between thousands of individual objects and one single object is pretty substantial.
That all said, I agree the comparisons a little harsh as it’s a brand new update to modo and that if even emphasizes the difference between 701 and 601 being pretty substantial, so a comparison of 601 and C4d would make a little more sense. It is however a huge release for modo, as this is the version where maxon can no long ignore them because they are only a partial app. There isn’t a major feature in my mind that modo doesn’t also have. Things like the character builder, or certain mograph objects, pyrocluster maybe, but they model, animate, rig, render, sculpt, UV layout and texture, dynamics, Particles, expressions, hair, they got it finally. If they have some better stability, then this is a serious contender and with a linux version being ported by some of the best linux graphics software developers, Maxon’s a definite step behind in terms of higher end pipelines ever adopting it. I’ll definitely be giving modo another try.
Yeah might be time to jump ship if c4d R15 doesn’t address these speed issues in a big way.