Particle Problems on a Mac
jrm21
Posts: 140
I seem to recall posting about this a while ago (probably the old forums). I also recall entering it in the bug tracker, which also seems to be gone.
Am running a mac (2x2.4 Ghz Dual Core) with 20GB RAM.
In C7, particles worked fine. I think it was somewhere early in C8 that things started to get very slow. While it is marginally better now, it is still unusable for most things.
"Simulating Particles" often takes a very long time. I have turned down settings, but that doesn't help. Turning off the 3d calculation in the Assembly room eliminate the problem, but of course it prevents you from previewing the effect.
Render times are insanely long. If I want to render a single frame, the "simulating particles" process can take up to an hour. (Once done, the render itself only takes seconds.)
DimensionTheory has a very good "object destruction" tutorial on YouTube using particle emitters. In that video he has three emitters each generating 100k particles. The video shows him doing things in real-time or near real-time that takes me several minutes (or more) using lower settings and only one emitter.
Is there some some other setting I am overlooking? Is this a "know" Mac problem?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Using C8.5 (also on Mac) I experience none of what you are describing so, it can't be a problem across all Macs.
Simulating particles takes a fraction of a second, not an hour. Render times are swift and speedy.
I can't imagine why it's working so poorly for you.
How many particles are you working with in your emitter(s)?
I can get things to work fairly well with a one emitter and the default setting. Once i start upping the particle count or add forces, things slow to a crawl.
In the DT tutorial on YouTube, he is using 3 emitters on a modified character mesh. Each one has 100k particles yet the display seems very quick. I can't come close to that with one 100k emitter. Are you using anything on that scale with success?
This may be neither here nor there, but as I recall, certain functions in Carrara are not multi-threaded. I don't think Physics calculations are multi-threaded, as well as the post effects such aura and DOF. I do not know for sure about particle calculations. If they are not multi-threaded, they are not on the PC side either.
I don't recall what the issue was, but I know that something in C8 broke many of the particle presets. Smoke is the one that tended to come up most often in threads about the issue. The fix was really simple and I don't know why it worked exactly. Basically, if someone with Carrara 7 saved the smoke particle preset in a new .car scene file, then it could be opened in C8.
Lack of multi-threading for particles would explain the poor performance.
Regarding presets - I also recall the issue with smoke from one version to the next. In my case, I am not using any presets, just a brand new emitter.
Again, I point to the DT tutorial video on object destruction. That is 3 years old and I think it was done in C7. I couldn't even consider trying that on my Mac with C8.5, although I recall doing similar things in C7 without much issue. That's what makes me think there is a problem.
GigaShadow is using particles on a Mac without the problems I am seeing, so I am now wondering if it isn't an issue with my system. I recently did a clean install of C8.5 so that isn't the issue. If others are not seeing a similar problem with 100k particle emitters, then I need to track down what is wrong on my side.
I'm on a Mac as well, but it's an old one and I still use C7.2 Pro, so I can't offer direct insights into C8 or later I'm afraid. It's been a long time since I saw the video and I have to wait until this weekend for my satellite bandwidth to reset to be able to watch it again.
Ouch. I feel for you.
I had broadband cable for many years and now have FIOS. Both are really fast and have no caps. I am able to work from home and deal with some large files.
On the other hand, I also have a part time residence in a rural area. Lucky to have POTS phone service there. No broadband, no DSL. Satellite is the only option. With all the problems, I can't justify the cost of it.
I have an iPad with Verizon 3G service that works (marginally) at that location. When I have to get work done, I knock it out on my laptop. Then I use a program to transfer the finished file (fortunately, I normally only have to email low-res PDFs of the project) to the iPad where I can then email it. A real PITA workaround, but it gets the job done. Still have to watch the bandwidth usage, although it isn't too bad considering it is usually work related.
For any really large files, I wait until I go into "town" and can find a place with wi-fi.
I fondly remember the days of 300 baud modems and long downloads for a *gasp* 32k file. Calling into services like CompuServe or Genie or someone's BBS and tying up the home phone line in the process. Amazing what we take for granted now.
Someone _really_ needs to get the _entire_ country wired up for high speed access. With the push for "cloud" services, online video and rich media, it really bothers me that they all overlook the small but important percentage of people without any real access to these services.
After reading your initial post I placed three emitters into a scene and set them all to 100,000 for max number of particles just to see how it would work. I'm not sure if that matched the tutorial your referring to exactly but it worked quite smoothly. If you want to provide the settings that give you trouble (or a screen shot of them) I'd be happy to match it, give it a try and let you know how it goes.
I'm wondering if it is because of particles+scene forces? Have you tried the particles without the forces?
I will test without forces.
I will also see if I can either assemble settings or upload a simple scene file.
Just recreated a scene and am trying to render. Render speed seems decent when starting from the beginning of the animation (frame 0). The problem seems to be if I try to render just one frame from later in the animation or just try to jump to a later frame in the assembly room.
Once the render is completed I will get back with more data.
Thanks, both, for your help.
Made me think of this:
http://www.collegehumor.com/video/3052195/24-the-unaired-1994-pilot
Even if you weren't a fan of 24 it's still funny if you remember "those days".
(of course if you were a fan of 24 it's even better)
Made me think of this:
http://www.collegehumor.com/video/3052195/24-the-unaired-1994-pilot
Even if you weren't a fan of 24 it's still funny if you remember "those days".
(of course if you were a fan of 24 it's even better)
Never actually seen an episode of 24, but know enough about it to get the reference. Hilarious video and I can relate to all of it. While I never had to disarm a bomb, most everything else happened to me with work related items trying to meet deadlines. Thanks for sharing.
I think that is normal. As far as I know Carrara does not "bake" the particles into the timeline. It has to calculate the frames between when you jump far ahead in the timeline...
If you saw a tutorial video, his timeline may not have been as long as yours? If that were the case his "clicks" may be faster because it is actually fewer frames...?
I do not know if particles would be any faster in the preview with a better render card. I think there is a setting under one of the tabs in the assembly for what percentage of particles you show in the preview... That could also be a big factor.
Calculation speed does not seem to be affected no matter where I jump to in the timeline. I just tried moving forwards and backwards, with five emitters set to 100,000. It calculates quickly every time.
makes sense, and is the case on the original scene I was working with. However, I tried it on a default 4 second scene and am seeing the same problem.
Render card could be the issue. I do have display percentage turned down quite a bit.
I just tried some different things and believe that scene forces is a big factor in my issue.
New empty scene, medium scale.
Create one particle emitter.
Set max particles to 100,000
Emission cube set to 5ft x, .5ft y, 5ft z (a 5 x 5 x 1/2 ft wall)
100,000 particles/sec, no deviation
.1 size, no deviation
1 mass, .02 deviation
0 velocity
0 dispersion angle
Emission time range from 0 to 1 second
Lifetime 10 sec
No changes to advanced tab
In other words, I want to fill the cube with 100,000 particles in the first second
This seems to go fine. The "simulating particles" isn't instant if I jump to the 3 second point, but it isn't bad. I can also crank up display percentage to 100% without a big speed hit.
Note that particle emitter default for "local gravity" is 10, so particles drop to ground as timeline progresses.
Next, I set local gravity in the advance tab to 0. Everything still seems to work fine. Particles are created and fill cube. I can jump to varying points in the timeline and things are okay. Spot render/render times are reasonable, regardless of timeline position.
Calcuate 3d view at 1 second mark takes about 1 second. At three second mark it takes about 3 seconds.
Here's where things get strange.
Next I add a directional force. It is pointed straight down. From frame 0 to frame 23 (24fps animation), the strength is 0. At frame 24, the strength goes to 10.
Strangeness #1: Although there is no force active for the first second (directional force is at 0), the particles expand upward. It is as if the emission cube grows on the x axis (y and z are not affected). Oddly enough, this problem is cured by changing the force tweener to "linear." Even though it goes from 0 at frame 0 to 0 at frame 23, it yields a strange effect on the particles when set to bezier. Easy enough to fix.
Now I jump to the 3 second mark. Click "calculate 3d view" with 5% of particles set to display. "Simulating" time jumps to about 45 seconds (about 10x longer). Spot render time jumps to over 14 minutes (of simulating. the render itself is quick). Thats up from about 4 seconds in the non-force versions.
Strangeness #2 (although I think this is something I am doing wrong): although the particles fall to the ground as expected, they never leave the emission cube. At the end of the animation, particles have a footprint the size and shape of the emission cube (5 ft wide, 1/2 ft deep). I would expect them to spill out across the ground as they fall.
So the problem definitely has to do with scene forces. No forces or just using the particle emitter local gravity works well. I guess the question is: Do the above times seem reasonable when using scene forces? Can a render card (such as whatever DT was using in the tutorial video) increase "simulating" time by such a drastic amount? (4 seconds to 14 minutes is 210x longer - that's a lot).
I think if you choose to use scene forces, then that overrides the gravity that is set in the emitter. You may have to use scene gravity in the scene's physics tab.
I generally set scene gravity and emitter gravity to 0. Neither can be animated/changed over time. Force objects can be animated and offer the necessary control.
I have never enjoyed the scene forces... I go to elaborate efforts to group the whole scene (cameras lights and all) and moving the whole universe around, rather than using scene forces. Doesn't always work of course depending on what kind of force you need...
Back when DimensionTheory was active here, he posted a lot of animations using particles and forces. That's what originally got me started using particles and seeing their potential. He did some amazing stuff.
If you group everything - cameras, lights, etc. and move it together, wouldn't it seem to the camera that there is no motion at all? I guess movement could be perceived if there was a background image which didn't move with the scene. Or are you relying on physics/scene gravity when moving things around? Just trying to get an idea of what you are doing as it seems a force object would be much simpler.
...
If you group everything - cameras, lights, etc. and move it together, wouldn't it seem to the camera that there is no motion at all? I guess movement could be perceived if there was a background image which didn't move with the scene. Or are you relying on physics/scene gravity when moving things around? Just trying to get an idea of what you are doing as it seems a force object would be much simpler.
Here is an example:
the first link is me trying to use scene forces to act like "wind" on soft body.... It is a creepy disaster so I put it up on line to compare:
http://www.youtube.com/watch?v=epCvVYfwOeU
Here is after I gave up on scene forces, grouped everything and dragged the "universe" through space.
http://www.youtube.com/watch?v=Wt9XezoCKsg
Interesting. Big difference between forces and dragging everything through space. I have never really played with the soft body stuff (I can't get hair to work the way I would like, so I didn't see the point in playing with something as complicated as soft body). I will think about the "universe drag" concept and see how I can apply to achieve the effects I am looking for. It does present some interesting possibilities.
Thanks for sharing.