Each frame renders longer than previous. All frames similar.
Moderators: ChrisHekman, aoktar
What about baking out the particles to alembic? Does that help to improve the efficiency?
CaseLabs Mercury S8 / ASUS Z10PE-D8 WS / Crucial 64GB 2133 DDR4 / 2 XEON E5-2687W v3 3.1 GHz / EVGA 1600 P2 / 2 EVGA RTX 2080Ti FTW3 Hybrid/ Cinema 4D
Is it fast? Oh, yeah!
Is it fast? Oh, yeah!
-
Shamefestival
- Posts: 14
- Joined: Fri Mar 10, 2017 5:18 am
I have this exact issue. I have rain cached x-particles and and a whole scene of assets that are optimized. Anyway, times start at 1min 30sec on frame 1. At frame 103 it raises up to 4m 30secs. HOWEVER------ If i stop and start the render again at frame 103 it renders back at 1min 30sec. Voila! i say render every 12 frames or so and restart frame last frame. You will save tons of time! This seems like a memory buffer type error that can absolutely be fixed. I have no idea how but if I found this method to "fix" it then maybe octane can implement and memory re-buffer thingy. I don't know. Hope this helps. I've attached a pic. Enjoy!
- Attachments
-
-
Shamefestival
- Posts: 14
- Joined: Fri Mar 10, 2017 5:18 am
jayroth wrote:What about baking out the particles to alembic? Does that help to improve the efficiency?
I have also tried my rain as an alembic, however Octane doesn't read my vertex speeds X-Y-Z from X-particles from the Octane tag which I would place on the alembic. Without proper motion blur rain looks really poopy. I keep seeing that realfow methods work but nothing at all on X-particles. All in due time I suppose.
I think I found a suitable solution! In the xpCache object, under 'Record', I selected 'Custom.'
Then under 'Particle Data', I checked only the fields I needed, which were 'Velocity' and 'Radius'. (My simulation was for a waterfall, but I imagine it would be similar for rainfall).
I haven't tested this exhaustively, but the result was functional (linear) motion blur, lower render times overall, a ~60% smaller cache, and most importantly, no per-frame render time increase! Hopefully this works for you too!
Then under 'Particle Data', I checked only the fields I needed, which were 'Velocity' and 'Radius'. (My simulation was for a waterfall, but I imagine it would be similar for rainfall).
I haven't tested this exhaustively, but the result was functional (linear) motion blur, lower render times overall, a ~60% smaller cache, and most importantly, no per-frame render time increase! Hopefully this works for you too!
i thought this solution was brilliant!
Now my render frames actually start relatively quickly..... but each frame is actually taking longer to render...... it is quite bizzare.
its 100% the motion blur, as if you disable blur, it's fine. I've also noticed switching to just "camera motion blur" instead of "full motion blur" has absolutely helped.
I guess its just managing what's right for your scene, hopefully you can get away with one of these 'hacks'.
Now my render frames actually start relatively quickly..... but each frame is actually taking longer to render...... it is quite bizzare.
its 100% the motion blur, as if you disable blur, it's fine. I've also noticed switching to just "camera motion blur" instead of "full motion blur" has absolutely helped.
I guess its just managing what's right for your scene, hopefully you can get away with one of these 'hacks'.