Getting illuminance in Lux from linear values given by camera's IDT/Input Transform


(Leonid Jakobson) #1

Hello,

I am trying to learn the ACES workflow, and need some help with this topic.

It is my understanding that the Input Transforms are made for each camera (or camera-setting combination) so that the resulting linear values all represent the same physical quantity. So if I shoot the same scene with two different sensors, the results should be virtually identical, all other factors considered. What I would like to know, is how do these values correlate with real illumimance as experienced by the sensor?
Of course, to get useful information about the actual scene, one would have to adjust this data according to the exposure time and lens’ transmission and aperture. But the sensor’s response to the light it receives should be accurately described by the manufacturer’s Input Transform. How do I convert from one to another? Even better if anyone could point out where this is documented.

Many thanks


(Jim Houston) #2

The output of the Input Transform is ‘scene relative exposure’, which is a photographic measure dependent on the lighting conditions, T-stop, ISO, frame-rate, shutter, etc… ACES sets an exposed 18% reflectance grey card to be equal to a 0.18 in linear OpenEXR. Similar settings on other cameras should ideally produce a similar result.

The easiest way to have a correlation to the exact illuminance used is to have the measurement from a DP’s illuminance meter at the test card location and a record of the exposure settings on the camera.

If you only have the camera information, you can estimate the illuminance by getting additional information from each setting – ISO 800 is often the native gains on sensors (you camera might vary), the amount of light that comes through the lens can be estimated from the T-stop, a correction for lens and filters in the camera can be applied from specs from the manufacturer – or measured with a reference light of how much fall-off occurs going through the lens at a certain T-stop, and then an overall adjustment can also be made based on shutter-angle duty cycle (180 degrees being 50%). There are more details, but this gives you the general idea, and what you get is an estimate of how much light was in the scene.


(Leonid Jakobson) #3

Thanks Jim,

As I mentioned in my original post, I am aware of the adjustments needed to be made to find the absolute scene luminance, more or less, what I cannot wrap my head around yet is how to accurately measure the illuminance of the sensor (I hope I am using the terms correctly). That is, the amount of light registerred by the sensor.
What I don’t understand about the 18% and 90% grey card explanation (which correspond linearly to 0.18 and 0.9 as you say) is that as far as I understand those values describe reflective properties of materials, i.e. the fraction of reflected light. What you are suggesting is recording the illuminance at the location of the grey card, then comparing that to the camera’s settings.
Let’s simplify the situation and assume I am shooting a diffused light source in a black room at a known distance. I have a good lens with measured T-stops, shoot at my base 800 ISO, and all other exposure factors are accounted for. My sensor registers the RGB pixel value of 14.0 at light source in ACES space. Before I calculate the source’s intensity, I need to know how many Lux seconds is received by the sensor. I feel like this should be so simple, yet I couldn’t find a direct explanation for ACES. There are plenty of general formulas, which always include a Camera Constant, which I feel should be known for ACES.

Many thanks, and I really appreciate the info you’ve already shared.


(Jim Houston) #4

unfortunatly, the camera constant can vary and the Academy doesn’t track those. Roughly ISO x for 18% is = 10 / I where I is in lux seconds. The is only a rule of thumb that can be off because camera marketing rounds off the actual ISO of the sensors. For example, maker A has an ISO of 654 but they round it to 800 and make B has 925 which also gets rounded to 800. So ISO as used is a photographer tool for mental doubling of light and or using as a fix for some EV. Getting actual light is tricker and looking at the standard can give several ways to get it. There is an academy doc on IDTs but it doesn’t really answer your question. There is more detail out there but it would require some digging.