4

I’m trying to get a uniformed value that indicates a light level from the Android back (main) camera.

I’m using several Android devices and the object captured is absolutely static (same object in the same light condition, in the same angle and the same position) and I get deferent value results. My calculation is based on Aperture, ISO, exposure time and exposure compensation. I tried to do:

EV = AV + TV = log(Aperture^2,2) + log(1/exposure_time,2).

I also tried some other methods from the following article: http://dougkerr.net/pumpkin/articles/APEX.pdf

Anyhow… the results I get differ from device to device…

Does anyone know how to solve it? Maybe there are more values that I’m ignoring?

Thanks.

joao2fast4u
  • 6,868
  • 5
  • 28
  • 42
Ofer Orgal
  • 192
  • 9
  • In photography, you would use a light calibration card. The calibration card normalizes the light measurement against a known standard. [link](http://www.lastoliteschoolofphotography.com/stills/Calibration-Greycards-VT1.jpg) – Fred F Feb 19 '13 at 17:14
  • Thanks Fred, what is the known standard? – Ofer Orgal Feb 20 '13 at 08:25
  • The calibration card is the known standard. When a camera looks at a room, it is measuring reflected light. The amount of light viewed will vary based on the surface the light reflects off of. The calibration card will reflect (without glare) a specific percentage of light, the reflection will be perfectly gray and smooth. To calibrate a camera, the card can be viewed under fixed lighting situation such as direct sunlight, then have other light situations measured against the known constant. – Fred F Feb 20 '13 at 15:06
  • Another method of light metering that light meters have frosteddomes over the sensor. The frosted dome causes the light to smooth and scatter to allow direct measurement of light. Since you are using a camera, my first thought was calibrating a camera. I'm not sure for a camera what would make a good choice to act as the frosted dome. Again, the baseline measurement would need a known light source, such as sunlight. http://img.wonderhowto.com/img/38/14/63475358911345/0/use-external-hand-held-light-meter.300x140.jpg – Fred F Feb 20 '13 at 15:28
  • I need a simpler method... a more mathematical method. – Ofer Orgal Feb 28 '13 at 12:07
  • Have you ever heard the expression, "how long is a piece of string?" Without more information, the question is unanswerable, much like yours. If you were in a room, with the walls painted perfectly black, no matter how much light is in the room, when the camera looks at the wall it would say that the room had no light. Conversely if the room was covered in mirrors, and the camera looked at the reflection of a light bulb, it would say the room was very bright, even if the light was just one small LED. – Fred F Feb 28 '13 at 19:17
  • Given one room with walls painted gray and a single 60 watt bulb in the center of the room and taking a picture of the wall standing directly below the bulb. Given 2 different camera set with the same exposure settings, the resulting images will very based on the lens and the cmos/ccd chip used by the different cameras. Somehow you need to neutralize these parameters. – Fred F Feb 28 '13 at 19:26
  • thanks for the replay Fred. to be more specific, im shooting a LED light in a completely Dark box. i want to do that with two different camera phones and get the same EXIF parameters. – Ofer Orgal Mar 02 '13 at 17:29

1 Answers1

0

I'm not sure how much you proceeded with this but I thought some answers might help you come to some conclusion.

With many Android devices, the Image Signal Processing chain can vary from one model to another, depending of course on who the makes the ISP and even the tunings that are done for the device. Some models may have tuned their devices successfully so you get some certain level of expected results, but others may not.

Other factors include unknown tone curves and other post-processing blocks which are hard to find out exactly unless you've been working in developing the device.

jimbo
  • 416
  • 1
  • 4
  • 14