Confusing results when attempting to create a balanced/calibrated scene

[TomasEsperanza]

I'm perplexed be the exposure settings, in relation to the scene, when attempting to create a balanced/calibrated scene (eg: grey material + light + exposure + render = grey is grey). Although I can get the desired grey, it's just that the the exposure (or something) is just not behaving as I would expect. I don't believe this is a gamma issue, and I'm reasonably confident with V-Ray, but this is weird....

 

The scene has:

 

VRaylight Dome, Multiplier 30, color is 255,255,255

 

VRay Plane: Simple Vray Material (default), 128,128,128

 

VRayLightMeter: 1m x 1m, Total Illuminance, Lux = 7970 (EV 11.5)

 

Background: color 128,128,128

 

VRay Exposure Control: Photographic mode, Shutter speed 125, f-no 9.8, ISO 100, White balance = Neutral

 

 

VFB, Pixel Information Color (8bit): Reads 128,128,128 (so that is fine but...)

 

 

My query is: To maintain a grey of 128,128,128 why do I have to set the f-number to 9.8 ?!

 

I assume that if these elements are to behave logically together; in a physically accurate approximation, should they not require an f-stop of 12.5 (or at least something less arbitrary than 9.8) ?

 

 

note: Additionally, the Render settings follow a LWF, and the V-Ray Environment skylight override is checked and set to black, so the background doesn't contribute light.

 

 

So, can any shed some light on this?

 

(oh no - unintended poor pun!)

 

If your wondering why it matters to me, this is a test scene I'm doing to consolidate my understanding of these components together. So I can confidently light and expose my scenes with real world parameters throughout the workflow. An f-stop of 9.8 in this scenario, just doesn't fit my understanding at the moment. I'd hate to find out the VRay exposure, lights, or lightmeter are unreliable, so hopefully I have just miscalculated.

[Nic H]

how is a fstop of 12.5 less arbitrary that 9.5?

 

a good work around to get 'perfect' colours is to put the values (128,128,128) into a light material then use that as the material for the wall / ceiling / element/ that way when you colour pick in the VFB or Photoshop there wont be any deviation from the perfect 128, 128,128.

light materials are also useful because they don't catch annoying shadows or light (which can change the pixel values)

[TomasEsperanza]

I'm looking at the EV of 11.5 (deduced from the lightmeter reading of 7970) and expecting the Photographic Exposure settings to require a realistic f-stop (given the other settings). For an EV of 11.5, at 1/125 and ISO 100 I would expect to use f-stop 12.5, as with these settings i would expect the f-number to be numerically 1 above the EV (As in the Sunny Sixteen Rule).

 

Perhaps that's where I'm going wrong (as the EV scale is logarithmic).

 

Incidentally, there are no shadows in this scene as there are no objects. The groundplane and the background are both grey, for checking with the colour picker.

[TomasEsperanza]

Version 2

 

 

OK, here's a much more straightforward scenario to illustrate the same issue. In this scene there is:

 

 

No lighting at all.

 

VRayLightMeter: 1m x 1m, Total Illuminance, Lux = 0 (EV n/a)

 

VRay Plane: VRayLightMtl, 128,128,128

 

Background: color 128,128,128

 

("Affect background" checkbox is off, while "Process Background and Environment Maps" appears to make no difference, either on or off.)

 

VRay Exposure Control: Photographic mode, Shutter speed 125, f-no 1.26, ISO 100, White balance = Neutral

 

 

VFB, Pixel Information Color (8bit):

Reads grey ground plane as 128,128,128

Reads grey background as 128,128,128

 

 

Query: So, I can expose the ground plane to match the Background (using the f-number) but why specifically is it f1.26 that exposes the ground's grey at 128,128,128 ?

 

Also: Is the VRayLightMtl emmiting a small amount of light ? and does the VRayLightMeter not detect VRayLlightMtl ?

Edited October 5, 2014 by TomasEsperanza

[Ismael]

 

No lighting at all.

 

 

In that scenario, does changing the background color to pure white from mid 128, does it change the final RGB of the VrayPlane? I think that it would.

[TomasEsperanza]

Going back to scene one....

 

The scene has:

 

 

 

VRaylight Dome, Multiplier 30, color is 255,255,255

 

VRay Plane: Simple Vray Material (default), 128,128,128

 

VRayLightMeter: 1m x 1m, Total Illuminance, Lux = 7970 (EV 11.5)

 

Background: color 128,128,128

 

VRay Exposure Control: Photographic mode, Shutter speed 125, f-no 9.8, ISO 100, White balance = Neutral

 

 

 

VFB, Pixel Information Color (8bit): Reads 128,128,128 (so that is fine but...)

 

 

 

My query is: To obtain a grey ground of 128,128,128, what makes it f-number 9.8 when the LightMeter indicates the scene requires an EV of 11.5 ?

 

 

Additionally, turning "Affect background" on allows the exposure control to affect the background, and at f9.8 this produces a very dark grey of 4,4,4. Of course, the exposure control is still effecting the ground too. This means obtaining the desired results (grey background, grey ground, with the same realistic EV) are still eluding me.

[TomasEsperanza]

Hi Ismael,

 

Nope. The background is totally independent of the VRayPlane. The background is not illuminating anything. The ground is a "self-illuminating" grey, seen as 128,128,128, with f1.62 in exposure control. The background in this case is not affected by the exposure control. So if I want to set the EV to 11.5 (as the LightMeter would suggest), I have to adjust either the background or the ground plane, away from their respectively designated 128,128,128. This means to match the two I have to "fake" the result rather than achieving it in a physical model.

[Ismael]

Have you asked in the V-Ray ChaosGroup forum?

[Nic H]

can i ask what the goal of this is?

are you doing detailed lighting analysis or some kind of product thing that requires 'perfect colour?

[TomasEsperanza]

can i ask what the goal of this is?

are you doing detailed lighting analysis or some kind of product thing that requires 'perfect colour?

 

It's for my personal workflow development, to be utilized in any pertinent cg scenario, although I am currently working in Arch-Viz; these are quick tests I'm doing in my spare time to consolidate my understanding of these components together, so I can more confidently light and expose my scenes with real world parameters throughout the workflow. I want to be able to put skies and objects (lit with a bright white light of real-world light measurements) and use real-world EV for both and get a consistent (thus reliable) exposure (and colour) with skies, backgrounds and objects in the scene. I want the real-world analogy to be consistent. I know I can "cheat" and adjust either the sky or the objects to get the same visual result, but that breaks the theory, changes the paradigm, and leaves more scope for tweaking arbitrary parameters. A simple method, will facilitate a "go to" procedure that I can employ at the lighting stage of many projects.

 

I use HDRI's too, but nevertheless to have a calibrated exposure for a scene would fit in with my growing understanding of photography. As I said, an f-stop of 9.8 in this scenario, just doesn't fit my understanding at the moment. I'm not saying it's wrong, but I'd hate to find out the V-Ray exposure, lights, or lightmeter provide unrealistic results because they don't to work together as the would in reality, so hopefully I have just miscalculated.

 

I'll ask over at Chaos.

[RyderSK]

I am bit confused by all this too.

 

But White color at Domelight at multiplier 30 (not 1) is already just another potato value so...you're already starting with something that doesn't resemble anything at all in real world.

[TomasEsperanza]

Yes, but It's not really the light that is the problem, its the disparity between 128' in the background slot and 128' as a material. They ought to be the same under the same exposure irrespective of the lighting. So it's an average day, say bright with some clouds, you set the camera to f-11 or thereabouts, and you see your grey object and your grey background, and they are the same. Perhaps a little darker or lighter than 128 because of the lighting and exposure, but they are the same. Unfortunately this doesn't happen.

[Ismael]

:pThe problem is the light and the disparity of an environment/background/sky vs a scene object irregardless of their texture or color.

They are different elements. that's my two cents.

 

I am still waiting for this 'problem' to be brought out at the Chaos Forum.

[Nic H]

im eagerly awaiting the outcome of this possible fundamental v-ray flaw

i cannot continue until it is solved!

[TomasEsperanza]

I knew you had an attitude problem already nicnic, which is why I was dissapointed that you chose to respond to my post. Let's avoid each other shall we. I didn't post here to receive juvenile sarcastic comments. Why comment if you don't know an answer to a query?

[Nicolai Bongard]

Im not sure if i understand your comparison to the real world. In your scene: The background color will not be affect by the light. The object in your scene will. If you use a default vray material, it will have black/none reflectivity. Still, depending on the angle it you view it and the direction the light hits your surface, the color of it will come out different. In the real world, your surface that is painted or whatever to be the 128 color would still reflect light and have some sort of glossyness would it not? so try to mimic your 128 material to what a painted surface would be in the real world. As it would likely have some sort of low glossyness and reflect quite abit more of the white light than none, the surface in your scene would become "brighter" as it catches more of the light, thus you could get away with a darker exposure, possibly something closer to the 12,5 you would expect? Im sorry if this is way off, just quickly browsed through the thread and i might have missed something.