View Full Version : layout hangs loading settings

08-26-2003, 06:08 AM
I need some help with a layout problem. I have, an admittedly, complex scene to render and layout hangs on loading the settings when I try to reload. The same thing happens in screamernet when the first cpu never gets past attempting to load the settings. The scene consists of an anatomical model of the human heart (126333 polygons, all triples) which morphs through the various phases of a single heartbeat. There are five separate partigon emitters to simulate fluid bloodflow within the chambers of the heart. Each of the emitters does not generate more than 200 particles. The particles have hypervoxels (volume) applied. The heart object has semi-transparent surfaces and is set up as an inside collision object so that the "fluid" flows are contained within the appropriate chambers. Am I asking too much of LW or is this a known issue. Any help much appreciated.

Specs Dual 800 G4, 768 Mb OSX 10.2.6, Stock NVDA GeForce twin view. LW7.5

08-27-2003, 02:42 AM
doesn't seem like too heavy a scene for LW, maybe try a blank scene and "load items from scene" from the original. There might be something wrong with the file - you might also try reloading the plug-ins, a bad one could stall the loading of the scene

08-28-2003, 05:35 AM
Thanks Toby, I tried importing to a new scene and got the same problem. Trashed the prefs and reloaded the plugins when prompted. Now the scene loads but takes about an hour to do so. (I think that's to do with the collision object properties being calculated). Start particle FX and it runs for about 2 hours. Then I hit F9 for a test render and Lightwave quits. I think I'll start from scratch.

08-28-2003, 09:18 PM
ouch. that blows

come to think of it, how many particles did you have? That always slowed down the load-time if they had any dynamics applied to them - how about rendering the particles separate and compositing?

09-01-2003, 06:52 PM
thanks again Toby. I've learned a few thing in the past couple of days like once LW has calculated the paths for your particles you can save them and remove the collision object properties which will speed up reloading of the scene. At least thats what I thought. Here's where I'm up to now. I recreated the scene and sequentially added the HVemitters with volume hypervoxels. Now with 2 threads set in the render options I could test render with no problems. The process viewer shows the CPU usage approaching 200% .. all very appropriate. Then I applied the collision property to the model and it completed after about 90 minutes. I now have 7 emitters with a total of about 2400 particles at the end frame (now 90). Then I hit start_FX to calculate the paths so I can save them. For the last 36 hours I've been staring at the spinning beach ball ! (not really, I went home in the intervening time.) The proces viewer shows LW hovering around 95-105 % with no other process showing any significant activity. Now I guess that the number of calculations increases exponentially with the addition of more emitters and particles so i'm a bit reluctant to force a quit but I have no way of knowing whether LW is progressing or stuck in a loop. So here's a suggestion/request for NewTek.. can we have some sort of progress indicator from which we can estimate the length of time required to complete these sort of calculations. In addition, the particle FX system does not seem to be using as much of the CPU as the renderer but it must be mulitthreaded or I wouldn't get >100% ??

Thaks for the suggestion about compostiting. I'm an amateur; can you advise me on compositing programs. will FinalCut Pro do ?

09-01-2003, 07:44 PM
I use After Effects, I haven't used FinalCut but it's for video editing anyway.

It only takes a little time to learn and now that I know it I wouldn't do it any other way

I haven't done much with collision, it's pretty complicated as you know - LW never updates while there is a spinning beachball (of death!) - I think it means it's interacting with OSX so it can't update

Do the particle emitters interact with each other or could you calculate the motion with only one emitter in the scene at a time?

09-01-2003, 08:26 PM
thanks again for the quick reply. The particles do interact with each other, I'm simulating fluid streams within the heart. What I'll do is give another 24 hours and then if there's no solution I'll split them into groups on the left and right sides to try and speed up the process. In any case, I'll post a reply with the result. Whatever else I've found the process instructive and It may be useful for others.