Page 1 of 1

Floating point (32 bit) image output

Posted: Wed Dec 28, 2016 1:26 pm
by Janmansilver
Hi, is it possible to have true floating point output from Octane? The highlights seem to be crushed although I choose, openEXR as output?

Re: Floating point (32 bit) image output

Posted: Wed Dec 28, 2016 1:36 pm
by juanjgon
The EXR untonemapped used with the EXR half precision option disables should write complete unclamped image files. The EXR tonemapped can have the highlights crushed if you are using, for example, a camera response or other tone mapping features.

Thanks,
-Juanjo

Re: Floating point (32 bit) image output

Posted: Thu Dec 29, 2016 11:14 am
by Janmansilver
Hi Juan

Thanks for the quick reply. :-)

Is this also the case when saving an image from mplay? if I use the exposure preview controls in mplay, the whites are still crushed? Are there other settings in the rendertarget node?

Re: Floating point (32 bit) image output

Posted: Thu Dec 29, 2016 11:21 am
by juanjgon
The gamma and exposure controls in MPlay should not affect the floating nature of the image. The limitation is in the Octane tone mapping functions related to the camera response. The camera response is done using LUT curves that are only defined for the display range, so all other floating information above or under this range is clipped.

Thanks,
-Juanjo

Re: Floating point (32 bit) image output

Posted: Fri Dec 30, 2016 8:55 am
by Janmansilver
Hi Juan,

Thanks, would something like vignette of the camera response also limit the dynamic range? :-)

Re: Floating point (32 bit) image output

Posted: Fri Dec 30, 2016 9:06 am
by juanjgon
I am not sure, but I think that if you set the camera response to linear, you should get a unclamped HDR image.

-Juanjo

Re: Floating point (32 bit) image output

Posted: Mon Jan 02, 2017 12:52 pm
by Janmansilver
Hi Juan

Thanks the images are unclamped now, however, I have issues with the alpha channel - is this stil 8bit? When I for instance really push the image in after affects it looks fine, however, the "alpha edges" become hard/harsh. Is this a limitation with alpha? This is my first project where alpha channel is required. I guess i could probably use a separate alpha or maybe it's simply an after effects issue/setting with the exposure filter.

The reason for this workflow, is that I can't predict the lighting intensity all the way through the animation, therefore being able to "capture raw" and then simply adjust the exposure in ae works well - especially if I can get the alpha working properly. :-)

Re: Floating point (32 bit) image output

Posted: Mon Jan 02, 2017 4:30 pm
by juanjgon
The alpha channel should be also a floating value, but it a value in the 0..1 range. I suppose that you can't push the alpha channel and the RGB values at the same time. You should extract the alpha channel before process the RGB data.

Thanks,
-Juanjo