OctaneRender™ 2020.2 Tweet

[Kalua]

Good work, thanks!

But when will there be the promised bugfix to avoid seeing the polygons on curved surfaces if using anisotropy or noise in the bump channel?
Waited so long for 2020.2 and hoped for a fix

kind regards

Looks like it didn't make it in time of 2020.2 release. It will be on 2021.
Check the 2021 features list.

[PolderAnimation]

Thanks for this release, glad to see 'Fixed denoiser output to be slightly brighter when the render layer is enabled' being fixed.

I already posted this before. But I will post it again. Motion blur on deforming characters is linear (that is of course understandable). But it would be great if we could use substeps (on vertex cached objects) to have more accurate motion blur on deforming objects (like characters). At the moment we have a project where we made a character that sometimes moves extremely fast an you see only his motion blur. In a part of the animation he circles around an object. If I render this, the motion blur is just a straight line instead of having a nice arc. If substeps (on vertex cached objects) where supported we could get this nice arc.

Below a test with a simple object.

A sphere that rotates 180 degrees in one frame.
The first one has rotation keys on the object is self, the second one has a vertex position cache (like characters would have) (both have substeps).

As you can see the one with the vertex position cache has wrong motion blur. So for all animated characters with subframes caches, the motion blur is wrong.

I have attached an orbx, if you slide the shutter time you can see where things go wrong.

is this something that is very heavy to calculate? Or is there another reason why nobody is responding to this issue?

[thanulee]

Hey,

Cinema4D R20 and sss materials issue. When i enable nested dielectrics from the kernel, the material changes a lot. It becomes kinda emissive.
I was told to post here since it seems like a core issue.

See images and scene below:
https://www.dropbox.com/sh/mgt2unuh5i54 ... 8tsGa?dl=0

[funk]

See images and scene below:
https://www.dropbox.com/sh/mgt2unuh5i54 ... 8tsGa?dl=0

Your link is broken

[thanulee]

See images and scene below:
https://www.dropbox.com/sh/mgt2unuh5i54 ... 8tsGa?dl=0

Your link is broken

fixed

[wallace]

Hey,

Cinema4D R20 and sss materials issue. When i enable nested dielectrics from the kernel, the material changes a lot. It becomes kinda emissive.
I was told to post here since it seems like a core issue.

See images and scene below:
https://www.dropbox.com/sh/mgt2unuh5i54 ... 8tsGa?dl=0

Thanks for the report, found the problem and a fix will be in place for 2020.2.1

[mojave]

Thanks for this release, glad to see 'Fixed denoiser output to be slightly brighter when the render layer is enabled' being fixed.

I already posted this before. But I will post it again. Motion blur on deforming characters is linear (that is of course understandable). But it would be great if we could use substeps (on vertex cached objects) to have more accurate motion blur on deforming objects (like characters). At the moment we have a project where we made a character that sometimes moves extremely fast an you see only his motion blur. In a part of the animation he circles around an object. If I render this, the motion blur is just a straight line instead of having a nice arc. If substeps (on vertex cached objects) where supported we could get this nice arc.

Below a test with a simple object.

A sphere that rotates 180 degrees in one frame.
The first one has rotation keys on the object is self, the second one has a vertex position cache (like characters would have) (both have substeps).

As you can see the one with the vertex position cache has wrong motion blur. So for all animated characters with subframes caches, the motion blur is wrong.

I have attached an orbx, if you slide the shutter time you can see where things go wrong.

Hi PolderAnimation,

Thank you for your feedback. Instance motion blur currently supports multiple motion keys within the current frame, however in the case of triangles just one motion key is supported at the moment. This is the reason why using object motion you can observe the whole trajectory interpolated using the various keys while in the second case the vertex positions are interpolated between the beginning and the end of the frame only.

Since we will be adding support for moving triangles to leverage Ampere HW in an incoming release, which provides support for multiple motion keys per frame, we may take the chance to also support it on older devices via software using CUDA the usual way.

I hope that helps.

[enricocerica]

Hi,
After a while and thanks to some free time I decided to get back on cooking.
I have 4 x GTX 1070 8Gb
I'm working on a big scene closed to the memory edges but some memory is still available.
I have the last drivers 461.40
If I use two of them, whatever are the devices, it works fine. As soon as I activate a third one I get the message that one card is failed.
Two devices are connected to a display but even if I choose those cards it works fine.
Any idea what could be the issue ?
Thanks

[abstrax]

Thanks for this release, glad to see 'Fixed denoiser output to be slightly brighter when the render layer is enabled' being fixed.

I already posted this before. But I will post it again. Motion blur on deforming characters is linear (that is of course understandable). But it would be great if we could use substeps (on vertex cached objects) to have more accurate motion blur on deforming objects (like characters). At the moment we have a project where we made a character that sometimes moves extremely fast an you see only his motion blur. In a part of the animation he circles around an object. If I render this, the motion blur is just a straight line instead of having a nice arc. If substeps (on vertex cached objects) where supported we could get this nice arc.

Below a test with a simple object.

A sphere that rotates 180 degrees in one frame.
The first one has rotation keys on the object is self, the second one has a vertex position cache (like characters would have) (both have substeps).

As you can see the one with the vertex position cache has wrong motion blur. So for all animated characters with subframes caches, the motion blur is wrong.

I have attached an orbx, if you slide the shutter time you can see where things go wrong.

is this something that is very heavy to calculate? Or is there another reason why nobody is responding to this issue?

It is possible, but fairly non-trivial. The problem is allowing more than a fixed number (currently 1 or 2) of vertex positions per shutter interval, without affecting ray tracing performance and without introducing too much overhead in the data structures. This is why we haven't tried to implement it yet. The highest priority in that area is currently to allow larger and more complex scenes since we got plenty of complaints about that.

[thanulee]

Another issue (i think). I have many specular objects and i need em all to have same material.

With nested dielectrics on, I need to go to each object and assign doublicates of the same material and set priorities. Either that, or I disable this feature..

But same time if i need bubbles inside the honey? Either I dont understand this feature yet, or this way of working isnt optimal.

Take a look at the video below:
https://www.dropbox.com/s/jyaeb9nmhclg4 ... .webm?dl=0

Many thanks!