View Full Version : Rendering time increasing with every pass

12-19-2003, 10:16 AM
Ok, I have had this happen now with two different scenes on two different computers on two different OS's, all with the same result. When I render a scene with radiosity the first frame will take about 2 minutes to render, then the next one will take 2 minutes and 2 seconds and then the third one will take 2 minutes and 8 seconds and the rendering times just increase from there. This is just an example of what is happening and not actually times. I started a scene last night, rendering to an .mov file and it started at 2 minutes 18 seconds and when I came back this morning it was up to 19 minutes. The scene is not getting more complex as the camera moves, and in fact, if I quit the render and then pick it up at the frame that was taking forever to do (say 19 minutes) it will take 2 minutes 18 seconds to render. It more or less looks like some sort of memory leak but I'm not sure what could be going on. This only seems to happen with interpolated radiosity. If I use Monte Carlo it works fine. No matter how low or high of radiosity settings I make it the time increases. I have tried it on a Dual 500 G4 and an 800 mHz G4 Powerbook, both with the same result. It's odd because this happens with different scenes also. This is quite frustrating. It seems inherrent to radiosity because when I turn it off it renders fine and when I use Monte Carlo it renders fine, but takes forever from the get go. Thanks in advance for any help anyone can provide.

12-19-2003, 11:10 AM
i concur.
the exact same issue has occurred to me, this should be put into the bug thread.

12-19-2003, 12:10 PM
I wasn't sure whether or not it was a bug or whether it was Mac inherent or what exactly the problem was so I just posted it here. If somebody wants to move it, feel free, but it could be a Mac only problem. Oh yea, it happens with both 7.5b and c. Weird.

12-28-2003, 07:33 PM
Top, anyone have a solution to this problem or notice it? I hate to have to render everything with Monte Carlo radiosity because it takes too long, but it actually takes even longer to do it on even the lowest interpolated settings. Any help would be appreciated. I've had a pretty simple animation with radiosity rendering on a dual 500 mhz G4 and an 800 mhz powerbook G4 for about 6 days now and it's only rendered about 1400 frames. Not too much considering it's a simple model of a farely simple machine. No background. I've got a gig of ram in the Dual g4 and 512 in the powerbook. I'm also rendering it with screamernet but that doesn't make a difference. This would have been done days ago if I could have used interpolated radiosity but I was forced to use Monte Carlo. Anyway, any and all help would be appreciated. Thanks in advance.

12-29-2003, 01:13 AM
Never seen this bug, but do you know about the motion blur trick? It could save you bunch of render time.


Oh yea, do you realize you can bake the radiosity onto your objects? - then turn Radiosity off -

12-29-2003, 09:56 AM
Yep... same thing has happened to me in the past... actually, I noticed it first on a PC, so I posted it in the PC thread a long time ago, but no one had any idea what it was. Now I've noticed it on my Mac, too. The thing is, it seems to be random... sometimes it works fine, other times I get the slow-down bug. I don't have time to troubleshoot and find out what exactly is causing this... maybe someone from Newtek can shed some light on this?

12-30-2003, 08:41 PM
How are you saving the sequence? If you are saving to a QT movie, try saving to images and see if you experience the same problem. Just a stab in the dark, it could be something to do with the way it adds the frames to the end of the movie. Also, does it take the same time to render a certain frame if you start at that frame instead of starting at the beginning?

01-01-2004, 01:23 PM
I have tried both QT and .psd output and both give the same effect. Also, if I were to start at frame 1 it would be pretty slow by frame 150, but if I were to stop the render and start it again at frame 150 it would render as fast as it were the first frame. Basically, it comes down to the length of time that it renders and the frame that I start at doesn't change anything. I'm still completely puzzled by this but I'm glad to know that I'm not the only one with the problem. Thanks for the help.

01-01-2004, 08:49 PM
This is obviously a bug with the software.

01-02-2004, 03:04 PM
what about breaking an animation sequence into smaller bits, ie, instead of rendering 300 frames all at once, breaking that into 6 groups of 50 frames.

would the increase rendering time still happen, or would it be limited over the course of 50 frames..smaller than it would be during a 300-frame render?


01-13-2004, 09:36 AM
I've had the exact same problem, try turning off the cached options for radiosity and caustics, that worked for me

01-30-2004, 07:16 AM
This increasing render time problem is a real deal-breaker...I find that 30-60 frame renders don't take too long, but when I hit 120 or more frames, the render times start to get silly, ie, 3 times the number of frames taking more than 3 times as long.

at 30 fps, 60 frames is a scant 2 seconds, and not typical of all the shots I need to create.

The only workaround I can see to this problem, short of a fix from NT, is to break all my sequence into separate files of no more than 60 frames, have SN blast away at them, then stitch them together in FCP.

Is this how people are dealing with this problem? I know that there are folks here paying their rent by doing animation, and I can't imagine anyone putting up with renders taking longer than they have to, or having to explain to clients what's taking so long.

thanks for any tips!


01-30-2004, 07:33 AM
i send the renders to our in house farm of PCs.

my mac is not used for rendering long sequences, I suspect many many people do it this way.

01-30-2004, 07:44 AM
do those pc exhibit the same problem, or since you have a farm, doesn't it matter that much?


01-30-2004, 08:08 AM
johnny they are PCs, so they don't matter to me. my co-worker might have problems with em, but he's clever, he can work it out. We don't do really long animations, we usually stay around 10 seconds (300 frms).

In addition to the original problem, I've also noticed that if i let my computer render over night, my render times Drastically slow down, with whatever options. I don't let anything but my monitor go to sleep, but for some reason this basically stalls renders. the more complex the scene, the slower the stall. were talking about a 4 minute-a-frame-render turning into a 33 hour-a-frame-render (that happened over one weekend last month). Generally its something like double the original time, but the longer you let it go, the worse it gets. And then when you come in and wiggle the mouse, the render goes right back to its original speed.

I might be crazy, but I think LW renders faster when I stay up all night surfing the web! Is it a feature or a bug? you decide.

01-30-2004, 08:12 AM
that's an interesting observation...wonder if there's a mouse-wiggling shareware program out there which would tickle lightwave every now and then?

Wonder if the overnight slowdown could have to do with that housekeeping that OSX does?


01-30-2004, 08:24 AM
i doubt it.
in os 9 when sherlock would do its thing in the middle of the night, maybe. but only because sherlock would use all of the system resources to do that. I have never seen any system level janitorial process do anything that even comes close to that level of disruption. And even if it did slow LW down, then LW should be able to recover as soon as the system is done cleaning up.

I've seen similar stalls in the last version of After effects, it would just stop rendering if the screensaver came up. that didn't happen all of the time, but there were a few overnight renders that had to be restarted.

01-30-2004, 08:44 AM
Hmmm, this is weird speculation from my side, BUT when i render large stills that took 30 to 40 hours i had the feeling that the first passes render at normal speed and that overnight it slows down. Looks to me like one of Lightwaves memory problems....

01-30-2004, 09:24 AM
ingo, my thought exactly. LW "forgets" what its doing.