2
$\begingroup$

Samsung Galaxy S21 Plus has a CMOS camera.

I recorded an ultra dark raw (DNG format) image using its Pro mode in the following conditions:

  1. ISO 50
  2. Shutter speed 1/12000
  3. Cover the aperture with multiple sheets of black printed paper in a dark room.

Each pixel is a triplet of values from 0 to 255 denoting Red, Green, and Blue.

I counted the frequency of occurrence of each value from 0 to 255 for each of Red, Green, and Blue colors, over a total of 12192768 pixels. The data for values 0 to 50 was as follows:

enter image description here

Imagine my intrigue seeing that among 12192768 pixels, there is no pixel with values 1, 3, 4, 6, 8, 9, etc.!!

What in the world is the reason for this?

Please allow me to express that I am very grateful for your time.

P. S. There seems to be absolutely no malfunction in the camera.

Improve this question
$\endgroup$
19
  • 3
    $\begingroup$ How sure are you that the signal is from the sensor and not from a pre-processing unit? I do not know of any non-scientific camera that has no pre-processing even for the uncompressed raw. Also, are you sure the file is uncompressed raw? $\endgroup$
    – José Andrade
    May 2, 2022 at 18:36
  • 1
    $\begingroup$ Also a third comment that probably 50 ISO is an extended ISO and not native, ie, it undergoes digital post-processing. $\endgroup$
    – José Andrade
    May 2, 2022 at 18:39
  • 1
    $\begingroup$ This may help. It still can be the effect of in-camera processing. Additionally ISO50 mode is designed to cut-off histogram with low noise values, i.e. makes image dimmer. This may be the cause of highly repeating 0 patterns. Of course CCD/CMOS sensors also has spectral response patterns of sensitivity to a different wavelengths, but I doubt that it is the current issue. $\endgroup$
    – Agnius Vasiliauskas
    May 2, 2022 at 19:05
  • 1
    $\begingroup$ Typical base ISO value is 100 $\endgroup$
    – Ruslan
    May 2, 2022 at 20:14
  • 1
    $\begingroup$ Just to clarify once again: from my camera knowledge (which might be wrong) I do not know of any consumer camera (phones, DSLRs, etc) that will give you the raw values of the sensor. Camera sensors first undergo a pre-processing treatment that you have no control over and this is the lowest level of data you can store. And that only as uncompressed raw data. I highly doubt that the .dng files you are getting from your phone are uncompressed raw, so you will also have some loss of data from the compression on top. $\endgroup$
    – José Andrade
    May 7, 2022 at 19:01

1 Answer 1

Reset to default
5
$\begingroup$

This was happening due to non-standard ISO values as is clear from the results of my experiments below. The strange values vanish on increasing the ISO.

A very peculiar implementation of low ISO indeed!

Thanks a ton to José Andrade, Agnius Vasiliauskas, and Ruslan for their guidance, without which I would never have found the reason.

Experiment Results

Improve this answer
$\endgroup$
8
  • 1
    $\begingroup$ I was the one who upvoted your question, so might as well upvote your good answer as well! And now I will never go below ISO 400 unless I have no other option! $\endgroup$
    – Ed V
    May 2, 2022 at 22:29
  • 1
    $\begingroup$ Albeit, you have found the reason,- there's nothing wrong in ISO-50. Each ISO standard is created with different goals in mind. Low iso values, especially ISO50 is designed to filter-out low power noises, so it's useful when filming in very-bright light intensities, near light sources. As high-iso values are designed for shooting picture in dark areas, so there's nothing wrong in any ISO mode. Use-it when necessarily. $\endgroup$
    – Agnius Vasiliauskas
    May 3, 2022 at 7:23
  • 1
    $\begingroup$ @EdV you would handicap yourself if you take 400 as the minimum for all the cameras. I've checked the histogram of a photo taken at ISO 100 by Canon 80D, and it has no zero dips except a dozen per ~2600 values. Taking ISO to 800 retains this status. So do note that this is very dependent on camera model. $\endgroup$
    – Ruslan
    May 3, 2022 at 9:27
  • 1
    $\begingroup$ @Ruslan I was just joking, actually! I use a Canon T7 to photograph echellograms in my homemade echelle spectrograph, shown in my answers here and the link therein. I have gone as low as ISO 200 if the light source is sufficiently intense, e.g., the sun. But it is interesting to see what the OP found and you commenters helped him resolve. $\endgroup$
    – Ed V
    May 3, 2022 at 11:53
  • 1
    $\begingroup$ @RiteshSingh It seems like some part of the camera is measuring the numbers 0-25, and then the software in the camera has to adjust them to 0-50 for some reason. I don't know enough about cameras to tell you why it works that way. $\endgroup$
    – user253751
    May 21, 2022 at 17:03

Not the answer you're looking for? Browse other questions tagged or ask your own question.