Page 1 of 2

Weird PMC Kernel problem.

Posted: Tue Jul 02, 2013 12:24 am
by TheMightySpud
Hi all,

I wonder if anyone could tell me what's going on here.

The attached two images are the same scene.
PoolBalls_ScreenGrab@99%.jpg
Screen grab of the render window at 99%

And
PoolBalls_f9_Render.jpg
The final result of the 'F9' render....

Using the PMC Kernel the final render always gets majorly blown out, and I can't figure out why. I've detached the post processing node, so it can't be that. It's very confusing.

Anyone have any ideas?
Thanks
TheMightySpud

Re: Weird PMC Kernel problem.

Posted: Tue Jul 02, 2013 6:52 am
by lixai
it seems to bee a gamma problem, can you disable any gamma settings if there are any and try again ?

Re: Weird PMC Kernel problem.

Posted: Tue Jul 02, 2013 7:40 am
by juanjgon
Yes, I think that you have some kind of gamma correction enabled in Layout. Be sure that all LW gamma settings are set to Linear.

-Juanjo

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 11:12 am
by TheMightySpud
Thanks for the replies.

I've double checked all the settings and there's no LW gamma correction enabled, all the color space options are set to linear etc.

All very weird :-/

Thanks
TheMightySpud

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 11:18 am
by juanjgon
And this problem is there only with PMC kernel? ... with path tracing or direct light kernels all works fine?

-Juanjo

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 11:20 am
by juanjgon
Also if you can send to me the scene (or a scene with this problem) could be great to test it here.

You can use my email if you want: [email protected]

-Juanjo

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 3:32 pm
by TheMightySpud
Hi,

Thanks for the replies. I've tried with the pathtracing Kernel but the problem persists, so I'm guessing I've missed something somewhere in regards to the gamma settings.

I've emailed the scene etc. over.

Thanks
TheMightySpud

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 4:03 pm
by juanjgon
Ok ... I can see this issue here.

Problem is that you have a pure linear settings in the imager node (gamma = 1.0 and response = linear). With this settings plugin read the HDRI Octane frame buffer without clamp the image range and without tone mapping (that is, you get a full not clamped HDRI image, not a tone mapped one).

The preview functions which allow plugins get the preview image while rendering are always clamped and tone mapped and never get the full floating HDRI information.

This is why you can see the diference between preview and final rendered image.

You can easy fix this issue (if you don't need the full not clamped HDRI image) simple setting the gamma in the imager node to somthing like 1.0001 ... this force the plugin to get the final image using the clamped HDRI buffer, and preview and final image are going to be the same.

Perhaps I should add a switch to enable the full HDRI not clamped image only if user need this kind of output image to avoid this issue with preview. Usually all Octane users output the image with some kind of tone mapping (some film response or gamma correction), this is why it is not easy to have this problem.

-Juanjo

Re: Weird PMC Kernel problem.

Posted: Wed Jul 03, 2013 4:22 pm
by TheMightySpud
Aaaah, interesting.

Thanks for that, and the really quick reply, I shall take a look :)

Thanks
TheMightySpud

Re: Weird PMC Kernel problem.

Posted: Sun Jul 07, 2013 10:03 am
by juanjgon
To avoid problems, next plugin release is going to have a new switch to get the full no clipped and no tone mapped HDRI image: "Get full linear HDRI image".

-Juanjo