OctaneRender® 2023.1.2

[abstrax]

Hi all,

We are happy to bring you an update to 2023.1.

This version supports all current NVIDIA GPUs with compute model 3.5 or higher. Please make sure to use a NVIDIA Studio driver with version at least 456.38 on Windows or 455.23 on Linux. For macOS please ensure you are running 13.3 or preferably newer.


Changes since 2023.1.1

Improvements

• Added new Light mixer output AOV layer node. This has the same functionality as the old Light mixer output AOV node, which is now hidden.
• Made the Legacy output AOV layer node visible to allow output AOVs of older scenes to be used as layers in the new output AOV node.
• Added "Subtract" and "Divide" blend modes to the Legacy output AOV layer node.
• Added pin "Affect occlusion only", which controls whether only the alpha channel is modified, or whether the (premultiplied) RGB channels are also modified, to the following output AOV layer nodes:
  • Adjust opacity
  • Mask with layer group
  • Mask with Cryptomatte
• Added the following new output AOV layer nodes, which can be used to store an image during compositing and recall it in a higher layer and make it easier to build reusable effects, and can improve performance by preventing repeated work:
  • Save checkpoint
  • Load checkpoint
  • Discard checkpoint
• Added new Sharpen output AOV layer node.
• The "Camera imager" node type is now called "Imager", because it's not related to cameras.
• The Imager node's "Saturate to white" pin is now called "Clip to white".
• Added option to switch between shading and smooth normals in the falloff texture node. This also fixes the problem that the mode "Normal vs. eye ray" did not take bump mapping into account.
• Refactored how we allocate memory for the denoiser buffers: Buffers required by the denoiser will be allocated before we upload the scene data to make sure that we have the required memory available to run the denoiser when needed. This should solve cases where the denoiser failed due to an insufficient amount of available memory.
• OSL: Added ray type "direct" to the OSL function raytype() to match direct light rays.

Fixes

• The combined slider step size of value nodes that have multiple destination pins is now the smallest step size of the destination pins instead of picking the stp size of the last destination pin.
• Fixed render failure if using an analytic sphere light with photon mapping.
• Fixed a net render node failure due to missing RTX data for meshes after an incremental render data update.
• Fixed network rendering on macOS when using certain OSL textures, and when using these built-in nodes:
  • Channel inverter
  • Image adjustment
  • Tile patterns
  • Chainmail
  • Angular field
• The bump map offset on macOS is now exactly the same as on Windows/Linux.

Downloads

Studio+ Subscription Downloads

OctaneRender Studio+ 2023.1.2 Standalone for Windows (installer)
OctaneRender Studio+ 2023.1.2 Standalone for Windows (ZIP archive)
OctaneRender Studio+ 2023.1.2 Standalone for Linux

OctaneRender Studio+ 2023.1.2 Node for Windows (installer)
OctaneRender Studio+ 2023.1.2 Node for Windows (ZIP archive)
OctaneRender Studio+ 2023.1.2 Node for Linux

Prime Downloads

OctaneRender Prime 2023.1.2 Standalone for macOS

Demo Downloads

OctaneRender Demo 2023.1.2 Standalone for Windows (installer)
OctaneRender Demo 2023.1.2 Standalone for Windows (ZIP archive)
OctaneRender Demo 2023.1.2 Standalone for Linux


Happy rendering
Your OTOY team

[nuno1980]

I'm super confusion for still no number of spp increased 1M to 100M due to the extreme delay as 4 months despite to be EASY.

[frankmci]

I'm super confusion for still no number of spp increased 1M to 100M due to the extreme delay as 4 months despite to be EASY.

If you are the only person requesting this double-order-of-magnitude increase in a core parameter of an insanely complex tool, it might just not be worth changing, no mater how easy it may or may not be (it's not a wise assumption to make), not to mention the potential unforeseen consequences it could have across twenty-two different hosting programs.

What is your primary argument for requesting such high per-pixel samples? Is it a niche situation, or does it impact a significant fraction of Octane user's work? Are there work-arounds for your situation that provide reasonable results with a mere 1,000,000 samples?

[abstrax]

I'm super confusion for still no number of spp increased 1M to 100M due to the extreme delay as 4 months despite to be EASY.

We won't change that for the reasons I explained to you multiple times already.

[nuno1980]

We won't change that for the reasons I explained to you multiple times already.

Ok but TOO BAD! I don't agree your reasons. I may abort to use OR and I may back to use the Cycles renderer.

[frankmci]

We won't change that for the reasons I explained to you multiple times already.

Ok but TOO BAD! I don't agree your reasons. I may abort to use OR and I may back to use the Cycles renderer.

I'm curious to know what situations actually need more than a million samples per pixel. Could you share an example scene file?

[nuno1980]

I'm curious to know what situations actually need more than a million samples per pixel. Could you share an example scene file?

Ok, I need more than 1M spp at "path tracing" kernel for very good (fixed + variable) caustics on our sun or on (very) small light source - here. I'll upload new film tomorrow about "how to increase the number of laps of 1M spp to two or more from one"...

[nuno1980]

Happy holidays, @frankmci and @guys!

How to increase to more than 1M spp for region picked only

[SSmolak]

I noticed that there is no Exposure control for camera in AOV. We can control near all settings from Camera Imager, Post but no exposure.
Beauty AOV doesn't include Camera Exposure.

Cryptomate doesn't work in PMC kernel.

Glare, Bloom AOV nodes still works different from non AOV Imager versions.

[elsksa]

I noticed that there is no Exposure control for camera in AOV.
Beauty AOV doesn't include Camera Exposure.

Could you elaborate? I am confused since Camera Exposure at frame buffer level is present in the Camera Imager and an EXR export doesn't expectedly bake (nearly all) the Camera Imager parameters. Not to mention that AOVs are purposed for post-comp where exposure controls are available in a linear workflow/floating point environment.