OctaneRender™ Standalone 3.00 alpha 4

[abstrax]

When rendering 3500 pixel wide image I can see different parts of the image refine at different times because of this new tiling mechanism in Octane 3. But the central tile is always much more refined then the surrounding ones and this does not go away if I let it cook longer. You can see the border between two different levels of refinement in the attached image.

previewImage2.jpg

Please read new features of OctaneRender™ 3.00, especially the part about the work distribution in network rendering.

[abstrax]

Sorry if I am bugging again but I got no answer before..

Will we see motion blur for liquids with changing topology? Why can vray, i-ray, etc etc have it and we cannot have it?

How can professionals use Octane if they cannot have motion blur for liquids

Thank you!

It is possible if the plugin can read the vertex speeds via the API of the CFD plugin/module. I'm not sure if or how that is possible, you would need to ask the plugin developer.

[abstrax]

Marcus, I wanted to update you that I now have 2 users - Boris and TomGlimpse (both Windows) who have successfully rendered my cam baking red car scene that was crashing at a res of 3840x2160 without freezing or crashing, using PCI 1x connections. (I had sent you an ORBX, however I sent Boris and Tom G an OCS, which was why Boris hadn't gotten an image, so I sent ORBX to both when Tom G informed me that the OCS wouldn't work).

Can you think, in your experience, of any troubleshooting us crashers can do to expose what may be a culprit when using V3?
I for example, tried different recent GEForce drivers (including most recent) with no success (all same result).
Thanks!
Notiusweb

UPDATE:
I tried with no success:
-running PCI lanes at Gen 1, 2, and 3 speeds (changed in Bios)
-rolled back BIOS
-rolled back NVidia drivers, including as far as initial release CUDA 7
-swapped out RAM sticks
-unhooked USB 3.0 riser 1x cards and only used Amfeltes expansion cluster
-disabled Internet and anti virus

So strange, sometimes crash occurs after render is completed. Most of time it gets caught during the little pauses occuring between the render sample points/steps, like when it goes from 0-100, then pauses then goes from 101 to 200.

Wondering also if some other Windows driver is mucking it up, something that never communicated with v2 but communicates with v3. Like us Crashers have a conflicting version while no crashes have an older or newer one that is non-conflicting.

It's most likely a driver issue, but don't ask me where. When it crashes, how does Octane crash? And what are the exact steps you are taking? I may add some more logging to figure out when exactly it happens.

[roeland]

Hi! I've found an issue with Dirt texture. It doesn't seem to be working as emisson texture.

As of now, dirt will not work as either an emission texture, or as an alpha texture for shadow rays. We can probably fix it for emitters. Maybe, to make sure any fix ma

--
Roeland

[Notiusweb]

Marcus, I wanted to update you that I now have 2 users - Boris and TomGlimpse (both Windows) who have successfully rendered my cam baking red car scene that was crashing at a res of 3840x2160 without freezing or crashing, using PCI 1x connections. (I had sent you an ORBX, however I sent Boris and Tom G an OCS, which was why Boris hadn't gotten an image, so I sent ORBX to both when Tom G informed me that the OCS wouldn't work).

Can you think, in your experience, of any troubleshooting us crashers can do to expose what may be a culprit when using V3?
I for example, tried different recent GEForce drivers (including most recent) with no success (all same result).
Thanks!
Notiusweb

UPDATE:
I tried with no success:
-running PCI lanes at Gen 1, 2, and 3 speeds (changed in Bios)
-rolled back BIOS
-rolled back NVidia drivers, including as far as initial release CUDA 7
-swapped out RAM sticks
-unhooked USB 3.0 riser 1x cards and only used Amfeltes expansion cluster
-disabled Internet and anti virus

So strange, sometimes crash occurs after render is completed. Most of time it gets caught during the little pauses occuring between the render sample points/steps, like when it goes from 0-100, then pauses then goes from 101 to 200.

Wondering also if some other Windows driver is mucking it up, something that never communicated with v2 but communicates with v3. Like us Crashers have a conflicting version while no crashes have an older or newer one that is non-conflicting.

It's most likely a driver issue, but don't ask me where. When it crashes, how does Octane crash? And what are the exact steps you are taking? I may add some more logging to figure out when exactly it happens.

Thanks Marcus. In that red car ORBX scene I sent you:
(1) I open scene, has res at 1280 x 720, and let it finish render
(2) I switch res to 3840 x 2160, and let it finish render
(3) I set to baking camera, and PC will freeze, to point of either self-reboot, switch to Blue Screen of Death, or permanent hang (can't move mouse or end program).
(sometimes it progresses a few seconds into render, other times it is almost instantaneous with selecting the back cam from drop down.)

With regard to driver issue, whatever conflict it is, I wish PC would hold out a little longer while the render can sort itself out and move forward to next segment. Almost like a timeout issue exists somehow.
Also, V2 works perfectly, so cause appears to be uniquely related to the way my PC has the GPUs speaking to V3.

Thanks! And good work with everything else, by the way!

[face_off]

With regard to driver issue, whatever conflict it is, I wish PC would hold out a little longer while the render can sort itself out and move forward to next segment. Almost like a timeout issue exists somehow.

I have had a few users report issues with NVidia driver 361.x. 358.50 seems to be the most stable driver at the moment.

Paul

[Notiusweb]

With regard to driver issue, whatever conflict it is, I wish PC would hold out a little longer while the render can sort itself out and move forward to next segment. Almost like a timeout issue exists somehow.

I have had a few users report issues with NVidia driver 361.x. 358.50 seems to be the most stable driver at the moment.

Paul

Paul, thanks so much. I rushed to try this when I saw it, but issue persisted even moving to that driver, at least in my case.

Marcus, maybe certain hardware splitters and risers connecting PCI 1x are not 'sturdy' enough to now handle elements of the new added-Kernel communication, even if they were able to handle the Kernel arrangement on V2?
It seems like when a V3 scene freezes, the render may even stutter for a little bit while it is trying to push through (you can see it by moving the mouse around), it but it can't clear the info exchange in time, and then PC crash.
Kind of like a stampede out of a building when it's on fire. If the doorway, or amount of time, could somehow be widened, maybe that would help? I know we have Parallel samples and max tiling, which should give us the option to proactively alleviate burden, but it doesn't seem to matter which setting I use if a jam happens, and it freezes on all parallel and max tile settings. Maybe the program could tell itself more specifically in the event there is a bottleneck, it could wait or something, I don't know....

[abstrax]

With regard to driver issue, whatever conflict it is, I wish PC would hold out a little longer while the render can sort itself out and move forward to next segment. Almost like a timeout issue exists somehow.

I have had a few users report issues with NVidia driver 361.x. 358.50 seems to be the most stable driver at the moment.

Paul

Paul, thanks so much. I rushed to try this when I saw it, but issue persisted even moving to that driver, at least in my case.

Marcus, maybe certain hardware splitters and risers connecting PCI 1x are not 'sturdy' enough to now handle elements of the new added-Kernel communication, even if they were able to handle the Kernel arrangement on V2?
It seems like when a V3 scene freezes, the render may even stutter for a little bit while it is trying to push through (you can see it by moving the mouse around), it but it can't clear the info exchange in time, and then PC crash.
Kind of like a stampede out of a building when it's on fire. If the doorway, or amount of time, could somehow be widened, maybe that would help? I know we have Parallel samples and max tiling, which should give us the option to proactively alleviate burden, but it doesn't seem to matter which setting I use if a jam happens, and it freezes on all parallel and max tile settings. Maybe the program could tell itself more specifically in the event there is a bottleneck, it could wait or something, I don't know....

It's really hard to tell, but my guess it has to do with the way how page-locked memory is handled over these splitters (i.e the memory that resides in system RAM, but is accessed from the GPU via DMA). We will do an alpha 5 release fairly soon, after that I will make a test build with more logging enabled to (hopefully) figure out, which step exactly is causing the problem. Please send me a PM and remind me in case I forget.

[FrankPooleFloating]

Marcus, did you figure out yet if you guys will be able to put render priority back? Again, myself and others really, really like being able to render at low or medium and shave off 10-20° from max temps... Pretty please?

[DominionXX]

I am also having this issue... the Photoshop exchange plugin keeps telling me that an update has taken place.. when I already have the correct version.
The post on page two, explains that we should tell you if this still happens with a fresh install of Windows and Photoshop. I can verify that it does still happen. I have talked to Adobe at great length, and they can not find any fault on their end. The asked us to talk to you the plugin creator.

Can you please help us with this issue?

Best Regards,

Todd Manus

Hi Todd, could you point me to the Adobe contact you've talked already about this to follow up with this matter?

Thank you.

I am sending you the case number in a PM.