OctaneRender™ Standalone 3.06 TEST 2

[vijay]

Noticed the interface of the AS has now changed into only 2 Values BUT with AS ON I now have this kind of artifacts that I never had before .. Min Adaptive Samples 128 Noise Treshold 0.020

Increase the Min adaptive sample value to either 256 or 512.

[calus]

Noticed the interface of the AS has now changed into only 2 Values BUT with AS ON I now have this kind of artifacts that I never had before .. Min Adaptive Samples 128 Noise Treshold 0.020

Increase the Min adaptive sample value to either 256 or 512.

Hey Vijey,

Indeed I also noticed this artifacts in previous version with group pixel 2x2 or 4x4 , even with high minimum sample,
depending on the scene we can actually see the 2x2 square or the 4x4 square,

So then I used the "none" option, but I had hard time to understand how the noise detection was done in this mode.

So then I thouhgt maybe that was the point of the adaptive strengh attibute, mixing back a little of noise to make the squares disapear,

But anyway both of these options are gone now so I'm confused ...
(also didn't have time yet to test seriously adaptive testV2 version)

[vijay]

Noticed the interface of the AS has now changed into only 2 Values BUT with AS ON I now have this kind of artifacts that I never had before .. Min Adaptive Samples 128 Noise Treshold 0.020

Increase the Min adaptive sample value to either 256 or 512.

Hey Vijey,

Indeed I also noticed this artifacts in previous version with group pixel 2x2 or 4x4 , even with high minimum sample,
depending on the scene we can actually see the 2x2 square or the 4x4 square,

So then I used the "none" option, but I had hard time to understand how the noise detection was done in this mode.

So then I thouhgt maybe that was the point of the adaptive strengh attibute, mixing back a little of noise to make the squares disapear,

But anyway both of these options are gone now so I'm confused ...
(also didn't have time yet to test seriously adaptive testV2 version)

We fixed a bug in calculating noise estimation, then after, in all our tests results, group pixel modes don't make any big difference. so we decided to have 2x2 as fixed option.

The artifacts are because the path tracing cannot find a light source in the specified min samples. In our tests, we thought 512 or 1024 samples would be enough to eliminate the artifacts. Seems some scenes would need even higher samples and we always wanted to remove adaptive strength because it increases render time by the percentage of reduced adaptive strength But if needed we can have it back.

[justix]

Noticed the interface of the AS has now changed into only 2 Values BUT with AS ON I now have this kind of artifacts that I never had before .. Min Adaptive Samples 128 Noise Treshold 0.020

Increase the Min adaptive sample value to either 256 or 512.

Right so this value does change depending on the scene..quite confused really ( in a way ) because It thought high value will take longer to clean?

ALSO I forgot to ask BUT when I use the Region Lasso there is no way to get a full render back unless I close the standalone and re-open the project.. is that normal? with the Lasso region the Max Samples is not taken into consideration?..is just go without stopping?

[abstrax]

Increase the Min adaptive sample value to either 256 or 512.

Right so this value does change depending on the scene..quite confused really ( in a way ) because It thought high value will take longer to clean?

The problem is that the noise estimate is just an estimate. It depends on the values of previous samples and if it indicates that there is no noise, we stop sampling. The key is to gather enough initial samples to make the noise estimate reliable enough. As a rule of thumb: If you increase the noise threshold (i.e. stop sampling while there is still some higher amount of noise, like 0.05 .. 0.1), you should increase min. samples to make sure that your noise estimate is good enough (for example 500 .. 1000). If you reduce the noise threshold (i.e. wait for less noise before stopping sampling, like 0.01 .. 0.02) you don't have to set min. samples to a very high value (often 128 .. 256 is enough).

ALSO I forgot to ask BUT when I use the Region Lasso there is no way to get a full render back unless I close the standalone and re-open the project.. is that normal? with the Lasso region the Max Samples is not taken into consideration?..is just go without stopping?

To remove the film or interactive render regions, enable the corresponding render region tool and click into the viewport. The render region will then be removed / reset.

[calus]

We fixed a bug in calculating noise estimation, then after, in all our tests results, group pixel modes don't make any big difference. so we decided to have 2x2 as fixed option.

The artifacts are because the path tracing cannot find a light source in the specified min samples. In our tests, we thought 512 or 1024 samples would be enough to eliminate the artifacts. Seems some scenes would need even higher samples and we always wanted to remove adaptive strength because it increases render time by the percentage of reduced adaptive strength But if needed we can have it back.

I had more test and can't reproduce in 3.06 TEST 2 the obvious artifacts (visible squares) I had in previous version.
I guess my artifacts were also relative to the noise estimation bug... don't need removed options back

Adaptive feel "slower" or more sensitive to noise in this version, but I guess this is also relative to better noise estimation.

[aLeXXtoR]

Thank you for release! Still some issues:

• Saving Z-pass to Untonemapped EXR still gives blank white image (with alpha if enabled).
• I also have some problems with no camera motion blur, but too few tests for now... will provide more info later.

[calus]

Thank you for release! Still some issues:

• Saving Z-pass to Untonemapped EXR still gives blank white image (with alpha if enabled).

You should share an EXR example of this problem, so the issue could be constated as real, or you could get explanations to how to use/display a linear z-depth pass.

[aLeXXtoR]

You should share an EXR example of this problem, so the issue could be constated as real, or you could get explanations to how to use/display a linear z-depth pass.

Oh, now we've come to idea that I don't know how to work with EXR?
You can easily try it yourself: open Octane Standalone, open some scene, enable Z-Depth pass, render image, save it as EXR (tonemapeed and untonemapped for comparison).

Here are the example files:

• Tonemapped
• Untonemapped

It is the same rendered image saved with 2 different ways.

[funk]

You should share an EXR example of this problem, so the issue could be constated as real, or you could get explanations to how to use/display a linear z-depth pass.

Oh, now we've come to idea that I don't know how to work with EXR?
You can easily try it yourself: open Octane Standalone, open some scene, enable Z-Depth pass, render image, save it as EXR (tonemapeed and untonemapped for comparison).

Here are the example files:

• Tonemapped
• Untonemapped

It is the same rendered image saved with 2 different ways.

I just loaded these into photoshop and get the expected result. The untonemapped image doesnt tonemap the zdepth pass. You can lower the exposure to see all the data is there (I think it should be in meters eg. value of 1.0 is 1 meter)