Mismatch between VRay Physical Cam and Max Physical Cam

[maxwood]

Rendering with VRay GPU 6 Update 1.2, Max 2020.

Attached screenshots of the settings.

I might I have missed something but from what I can see, these settings are identical, but rendering with the VRay Physical Cam the render appears darker (possibly a stop darker?)



Cheers

[^Lele^]

The exposure Control in the Max camera may be responsible for it, it's not linear afair.
It can be easily overridden (first checkbox) but then one has to match unit scaling (*).
Our cameras ignore the control entirely, the max ones seem to pick up something else, so the scaling isn't linear between the two.

See for yourself:
In an empty scene, set the Bg to a pure white color.
Expose the V-Ray camera to produce a 1.0 white value (in EVs it's 8.5855).
Set the Max camera to the same EVs.
In my test, the Max camera produces a value of 1.370 to V-Ray's 1.000.
Set the Physical Scale in the exposure control to 1097.89 to match the V-Ray's camera output.
(*)However, the above breaks as soon as the exposure is changed, requiring readjustment, thanks to the non-linear nature of the max's exposure control.
Which is why we wrote our version of it.

[Joelaff]

Though I have not specifically tested it I thought I noticed that VRay’s camera took into account focus distance in relation to exposure compensation (bellows compensation as we large format photographers called it). Thus the closer you focus the darker the image gets, like in a real camera. (Yeah, it does, though it varies by lens design.)

I thought that disabling exposure control completely prevented this behavior.

But I would need to test this. Was I imagining things?

[Joelaff]

I just tested it, and YES, The VRayPhyscialCamera DOES alter the exposure based on focus distance as I describe above. The closer you focus the darker it will be.

While this is accurate to a real camera, I was hoping for a mode where we can still set the EV, but NOT have the exposure change with focus distance. There have been many expensive systems designed to compensate for this in the real world.

Note that disabling exposure completely removes this change in exposure based on focal distance, but both Physical Exposure and EV modes in the VRay camera demonstrate it. I assume it is by design, which is cool, though again would be nice to be able to turn it on only when you actually wanted it.

I cannot seem to get this behavior to show using the Max Physcial Camera, but I really hardly ever touch those (use the VRay Cam); so maybe I am missing something there.

Note: For anyone who goes out and tests this with their 50mm lens on their DSLR only to notice no exposure change between minimum focus and infinity this is most likely due to the fact that most DSLRs store a lens profile that internally compensates for this (and vignetting) before you even see the image.. The effect is real, and can be significant, especially with lens that focus close (and definitely in macro).

[^Lele^]

Nope, that's also a factor, in fact, in exposure.
Doubling default focus distance (in the sample above) from 200 to 400 cm results 1.014 (so up by 1.4%.).
This gets very prominent for very short focus distances, at 1cm focus distance the resulting exposure is down to 0.036, so down to 3.6%.

This is however mirrored in the max physical camera, so assuming one matches all parameters on the camera controls, the only difference would be in the exposure controls.

[Joelaff]

Oh, I could not get the max camera to exhibit exposure change based on focal distance. But I was probably doing something different.

Were you using one of the exposure controls in the environment window, or is there done other trick?

[Joelaff]

And, yes, just like a real lens (inverse square) the exposure change at closer focus is much greater at short focus distances because the lens has to move a lot more away from the film plane to effect the change in focal length.

A rule of thumb (and of law given ideal components) is a 1:1 reproduction of your subject on the film plane will require two stops exposure compensation, and the distance from the optical center of the lens to the film plane will be double the focal length.

[^Lele^]

Working exclusively on camera, no exposure control needed for the vraycam.
Set a target distance, set an EV, change target distance.

[Joelaff]

Working exclusively on camera, no exposure control needed for the vraycam.
Set a target distance, set an EV, change target distance.

Yes, this is what I see. I mean, it is disabled if you set exposure control in the VRayCamera to No Exposure, but otherwise it is there.

Were you seeing it on the Max camera? That is where I was not able to affect an exposure change by changing the focal distance.

[^Lele^]

Were you seeing it on the Max camera? That is where I was not able to affect an exposure change by changing the focal distance.

Absolutely not, you're right!
I assumed it was the case as the code was given to them, but i was clearly wrong (it may have been a later addition, they may have had other needs, etc.).
So it's another thing to keep into consideration.

The best way to ensure you're matching exposure is to render the cameras against a white BG, and do proportions with the rendered color value.
It will need some fiddling, i guess.

[Joelaff]

Basically using the Max camera vs the VRay camera is like using two different physcial camera/lenses. There will be subtle differences in exposure between. (Though in the case of the close focus exposure compensation we have been discussing the differences could actually be huge (stops) if you are in a macro setup.)

I just tested it, and the VRayCamera seems to accurately make the image 2 stops darker when you are focused at twice the focal length (which would be that 1:1 reproduction on the film plane, and have the film plane also twice the focal length from the lens, which is indeed two stops falloff (all given perfect components, of course)). Pretty cool.

Though I would still like a way to use EV exposure and yet turn off this close focus exposure (anti)compensation. Though one can always animate the EV value.

NOTE: I just tested it, and the Target Distance is used to compute the internal falloff if you have not defined a Focus Distance. And if the camera is not targeted the target distance is still used for this. (Not sure I like that choice, but if you know about it...)