Page 1 of 1
[req] Calibrating the scene
Posted: Fri Nov 30, 2012 5:44 am
by pixie
In photography one can calibrate any given photography by aiming a
Gray Card. It would be nice to have some sort of equivalence, namely when in some caes is known the original values, so one didn't need to aim at a specific value, but at a given known value from the scene.
Re: [req] Calibrating the scene
Posted: Fri Nov 30, 2012 8:28 am
by tremo
I like to pick the white point on an RGB gray material. Do you mean this?
Re: [req] Calibrating the scene
Posted: Fri Nov 30, 2012 8:53 am
by pixie
tremo wrote:I like to pick the white point on an RGB gray material. Do you mean this?
It's somehow akin to it but regarding exposure. See bellow.
The top is over exposed, bellow it's the correct exposition.
The concept is akin to
http://vimeo.com/49677472, but instead of doing it manually, indigo might do it internally, searching for a correct exposure. I do it by tweaking the exposure until the image reaches the rgb of the material.
Re: [req] Calibrating the scene
Posted: Fri Nov 30, 2012 1:03 pm
by tremo
I look at the histogram. In the first case, the right wall in the visible region (220-245 RGB). By the second case, the wall is overexposed. That depends on what you want to show. The lights or the shadows.
Re: [req] Calibrating the scene
Posted: Sat Dec 01, 2012 1:07 am
by galinette
pixie wrote:In photography one can calibrate any given photography by aiming a
Gray Card. It would be nice to have some sort of equivalence, namely when in some caes is known the original values, so one didn't need to aim at a specific value, but at a given known value from the scene.
So you would like to click to an arbitrary object in the scene, and that the resulting exposure does not depend directly on the albedo of this object, so that you do not need to insert a standard gray (albedo=0.18) in your scene?
In that case, the process would be:
- The user clicks somewhere in the image
- Indigo computes the luminance of this point, either from image values, or by sending a bunch of rays
- Indigo gets the corresponding geometry point and its albedo from material properties
- Indigo finds the proper tonemapper gain parameter so that the resulting pixel value brightness is equal to the material albedo. This way a 18% gray card will look 18% bright, and a 64% gray material will look 64% bright.
NB : I'm talking here of power-linear brightness values. For converting to sRGB values, you need to raise them to a power of 0.45)
Re: [req] Calibrating the scene
Posted: Sat Dec 01, 2012 1:56 am
by pixie
galinette wrote:
So you would like to click to an arbitrary object in the scene, and that the resulting exposure does not depend directly on the albedo of this object, so that you do not need to insert a standard gray (albedo=0.18) in your scene?
You pretty much nailed it!

Re: [req] Calibrating the scene
Posted: Sat Dec 01, 2012 9:51 am
by FakeShamus
I would take this a little further and suggest something like a photoshop "levels" capability to the histogram, adding white, mid and black point pickers, which would allow you to set custom values for each and then clicking within your image would recalibrate the exposure at those points.
Re: [req] Calibrating the scene
Posted: Sat Dec 01, 2012 10:13 am
by pixie
FakeShamus wrote:I would take this a little further and suggest something like a photoshop "levels" capability to the histogram, adding white, mid and black point pickers, which would allow you to set custom values for each and then clicking within your image would recalibrate the exposure at those points.
I saw what you did there, taking advantage of the full 32 gammut and beyond!

Re: [req] Calibrating the scene
Posted: Sat Dec 01, 2012 5:41 pm
by CTZn
pixie wrote:FakeShamus wrote:I would take this a little further and suggest something like a photoshop "levels" capability to the histogram, adding white, mid and black point pickers, which would allow you to set custom values for each and then clicking within your image would recalibrate the exposure at those points.
I saw what you did there, taking advantage of the full 32 gammut and beyond!

I'm quite uneducated regarding those imaging thingies, hence my question: are those propositions assimilable to new tonemapping techniques or am I mixing things up here ?