Help with Auto exposure workflow.

[stevejjd]

Hi guys. Im having a hard time understanding the workflow to set up automatic exposure. Is anyone able to outline the workflow to get the camera autocorrections applied to the camera? Thanks

[bdancer]

So you need to enable the corrections, then after the render end you could press "Transfer ROP Auto Corrections" button at the physical camera properties. Will this work for you?

[chakarov]

In order to calculate auto exposure/white balance, V-Ray uses LightCache pass, so you need to start the render at least once. To enable the feature go to V-Ray Renderer ROP > Renderer > Camera > AutoCorrections. The exposure/white balance will be applied into VFB but not applied to the camera, to do this camera should be a V-Ray physical one and you should press "Transform ROP Auto Corrections" in it's V-Ray > Physical Camera > Main tab, then disable the "Auto Exposure" and "Auto White Balance" in V-Ray Renderer ROP.

[stevejjd]

Thank you guys......I have some questions if that’s okay.


QUESTION 1 - Intensity VS Automatic Exposure

In the look development process, we have to make sure that the shader works under different lighting conditions, so it's recommended to use many different HDR images from sunny to in-door.

Is it better to lock the Intensity of the VRayDomeLight to a value of 1 and use the Autoexposure for each HDRI image? (so that no matter what HDRi I am using, the exposure will be set perfectly in the camera).


QUESTION 2 - Camera exposure VS Light exposure

Renderman Lights offer an intuitive way of controlling the brightness of light using an 'Exposure' value. It appeals to photographers because brightness is controlled in stops. V-Ray offers an intensity value for lights and we adjust the exposure through the camera.

What is the difference between adjusting the exposure of a light VS adjusting the exposure of the camera - does each method result in the same image?



The reason I am asking is because I want to know if I should leave my HDRi Dome at a intensity value of 1 and to run auto exposure for each different HDRI image. I am trying to control the many variables and maintain as much consistency as possible. I’m trying to 'normalise' the lookdev process.
Thank you

[chakarov]

Hello stevejjd , as V-Ray works physically correct I can offer you to use Dome light with "Intensity" of 1 with all HDRIs, and to tweak camera settings. Like you will do with real camera in a real environment, you can also use auto-exposure/white balance.
As to the Camera Exposure vs Light Exposure, we don't have such a parameter in V-Ray Lights and I am not sure if this should produce similar or same results. If you are common with that approach, you can easily "implement" it with our lights. Just add "*pow(2, exposure_value)" after intensity value directly in intensity parameter, where exposure_value is this "Light Exposure".

[^Lele^]

Thank you guys......I have some questions if that’s okay.


QUESTION 1 - Intensity VS Automatic Exposure

In the look development process, we have to make sure that the shader works under different lighting conditions, so it's recommended to use many different HDR images from sunny to in-door.

Is it better to lock the Intensity of the VRayDomeLight to a value of 1 and use the Autoexposure for each HDRI image? (so that no matter what HDRi I am using, the exposure will be set perfectly in the camera).

If it's what you need, sure.
However, if those HDRIs were meant to be brighter (say, a sunny outdoor), or darker (a mossy cave), then you'd be missing the material behaviour under those conditions, a behaviour which may or may not be symmetrical with a "normalised" exposure (f.e. SSS, Hair and such may change their look a lot based on how much light has hit them.).

QUESTION 2 - Camera exposure VS Light exposure

Renderman Lights offer an intuitive way of controlling the brightness of light using an 'Exposure' value. It appeals to photographers because brightness is controlled in stops. V-Ray offers an intensity value for lights and we adjust the exposure through the camera.

Brightness can be controlled also with ISO (by *far* the best way with a virtual camera.) and with exposure time (provided one isn't trying to achieve exact moBlur. as that would be skewed.).
Changing stops also changes the depth of field, which is another potentially big issue.
I'd say that method to define brightness is ageing, and in sore need of updates, rather than a law we should all conform to, because someone else used to do it this way.

I had to work through two movies with sups calling a quarter of a stop up on lights, and having to do the multiplication math backwards to verify the output intensity was matching, as VFBs do not have any info about stops, they only have float values linearly multiplied by whatever light or camera changes one made.
When the Sup wasn't convinced I did what was requested (do not ask...), having to show them the math was even more painful.
As they wanted half a stop less, having to write down negative powers made most go haywire.
I wouldn't want this on my worst enemy.

Alas, in Houdini this is a piece of cake, as chakarov well explained above.

What is the difference between adjusting the exposure of a light VS adjusting the exposure of the camera - does each method result in the same image?
[/quote]
Provided you're not at the edges of precision, with insanely small values for lights and surfaces, yes, it ought to match (with the in-render exposure, crucially NOT with the VFB/Post one!).

The reason I am asking is because I want to know if I should leave my HDRi Dome at a intensity value of 1 and to run auto exposure for each different HDRI image. I am trying to control the many variables and maintain as much consistency as possible. I’m trying to 'normalise' the lookdev process.
Thank you

If you want to lookdev a shader for a specific scene, your safest bet is always going to be to see it in its proper context.
While personally very fond of lookdeving once, properly, to then reuse many times (or across facilities, and ever render engines, like we did at Pixomondo for the Oblivion bubbleship.), i have hardly ever seen it happen in the real world: most of the times tweaks will have to be brought to bear, based on the specifics of a given scenario.

Regardless, you will need controlled lighting (i.e. properly captured domes, normalised to match a Macbeth chart, in the HDR and in the lit scene.), and very strong adherence to the physical properties for the shaders (i.e.: no adding artistic "magic" here or there.).
If either is arbitrary, you'll be chasing your tail endlessly, and will likely have to chase it again as the lighting or shader change ever so slightly.

[stevejjd]

Wow i really have to thank you guys so much for taking the time to explain. This is very good information !

Thank you @chakarov
Thank you lele

This is pure gold.
Yes I think what I am after may be out of my ability to control and therefore I will be endlessly chasing my tail to make the shader work under any HDRI lighting condition. Thanks again for the help, its very timely for me.

lele, congradulations on working at Pixomondo. I am trying to put a demo reel together for the position of Texture artist at ILM in Sydney.

All the best
Steve