PDA

View Full Version : Particles not simulating the same at rendertime


blakboks
04-04-2006, 07:51 PM
Hey all,
I'm having issues lately where my particles don't seem to be simulating the same at rendertime as they are in the viewport--which, of course, makes absolutely no sense. There doesn't seem to be much rhyme or reason to it. I have a few script tests on my particles, all doing rougly the same thing, just to different groups of particles (events)--and this is where the problem seems to be occurring. I'm also having the issue of getting different results if I step through one frame at a time vs. jumping to a frame (i.e. I get the proper results when going a frame at a time). I would think that both would get the same result, since even if you skip to another frame, then it recalculates up to that frame. Ideally, I'd like to cache the results, but will the renderer use the cache? Is there such a thing as a cache player (like a .bin loader in realflow)?

On a related note, what's the best way to net render particles?
Much appreciated,
-Chris

Ls3D
04-04-2006, 08:04 PM
Well there is a cache operator and you can tell it save with the max file. Particles are driven by linear events so jumping around can be trouble. I guess one could even convert particles to verticies and point cache them. As for the scripting stuff I'm no pro.

-Shea

blakboks
04-04-2006, 08:13 PM
Yeah, I was using the cache operator (in the same event as the render operator, placed below the render operator), but the renderer doesn't seem to use the cache.

-Chris

OlegB
04-04-2006, 09:19 PM
Cache operator has options: 'Viewport', 'Render' and 'Viewport/Render'. For your scenario you have to use the last one.

Thanks,
Oleg B.

blakboks
04-04-2006, 09:47 PM
Too bad there isn't a smiley for 'kicks self in teeth'.

I think I actually saw the viewport/render option there when I first put the cache operator into the flow. I really feel like an idiot...that should've been the first place to look...

I just think I'm starting to get too dependent on the forums...

Thanks for your patience.

-Chris

blakboks
04-11-2006, 08:20 PM
Ok, well, now that I've got the cache working for rendering and everything, I seem to be having a problem of a different sort.

My cache is enormous. It seems like it's abnormally enormous. I only have about 300 or so particles and the cache looks like it could easily go up over 4GB (only 2GB on the machine). My scene is 2400 frames long and I have geometry instanced on 2/3 of the particles (only 556 faces--nothing too high-poly).

I tried caching only every 5th frame, but when it renders I get some 'skipping' with the particles (basically due to the problems described in 1st post). I also tried rendering in chunks--i.e. cache frames 1-600 and render, cache 601-1200 and render, etc.--however, since the cache operator skips right to frame 601, it doesn't get the correct state at that point.

Does anyone have any tips on getting the cache down or how to render and still get the correct results?

(hope this makes sense)
-Chris

Ls3D
04-11-2006, 08:59 PM
Remember you have to kill your particles by age, collision etc. Do you have a delete operator anywhere in the graph?

-S

blakboks
04-11-2006, 09:06 PM
No, I don't.
I emit the 300 particles in the first frame and stop emitting. All of the particles are persistent throughout the entire 2400 frames, they just move from event to event through script tests in a cyclic manner.
Would it be better if whenever I move a particle from one event to another to kill it and create a new particle?
-Chris

OlegB
04-11-2006, 09:08 PM
You can either try figuring out what is wrong with your scripting (if scripting is set up properly then render = viewport), or use Cache Disk operator from Box#3 (http://www.orbaz.com/forum/viewtopic.php?t=464) - it allows to cache everything (or some selective channels) to files.

Thanks,
Oleg B.

blakboks
04-11-2006, 10:16 PM
Does the disk cache operator in Box #3 support network rendering and everything? How does that work--would I need to copy the cache files onto each computer in the network, or would they all look to one location (provided it was shared)?

Is it even released yet?
-Chris

OlegB
04-11-2006, 10:34 PM
Yes, the Cache Disk operator supports network rendering - you can set up the cache to write/read files on a shared drive, then it can be used for the renderfarm. It creates cache on a-file-per-frame basis.

The official release is not out yet - right now it's in pre-release phase where you can buy it at discount price and start using it in the production. Comparing with the current pre-release build, the official release will have more tutorials and examples. The tutorials are oriented at the usage of the numerous sub-operators. The Cache Disk and Cache Selective operators are quite straight forward.

The details are in the link that I mentioned earlier: http://www.orbaz.com/forum/viewtopic.php?t=464

Thanks,
Oleg B.

pete
04-28-2006, 12:55 AM
Hey there,
I don't know if you are still having issues with PFlow but I had something that sounds pretty similar and the way I fixed it was to make sure the emitter object was unhidden when I set it off to render.

pete
04-28-2006, 02:00 AM
ahh crap. I spoke to soon, now mines doing it too... That sucks, looks cool in the viewport but misses a few operators in the render. yeay.

Here's what it looks like in the viewport
http://www.peterleary.net/forumfiles/In_Veiwport.jpg


And in the render it looks like this

http://www.peterleary.net/forumfiles/In_Render.jpg

It's GOING nuts I tells ya...
If anyone has the time, here's a max file demonstrating the problem. Download (http://www.peterleary.net/forumfiles/Pflow_pain.max)

OlegB
04-28-2006, 10:58 AM
You have different integration step for viewport and render (see the parameters for PF Source 01).

Thanks,
Oleg B.

pete
04-29-2006, 03:26 AM
Thanks Oleg!
Sorted! :)
blakboks, you should check that out for sure.

CGTalk Moderation
04-29-2006, 03:26 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.