Basic CCD Parameters v.0.3.1 Readout Depth

[freddy1]

Hello Every Body

I have a question
How is excactly the meaning of Readout Depth value? and How is the difference between this value and A/D bit ?

I have clear that A/D bit is the ADC or Analogical to Digital Converter value, which is specific for each camera, provided at camera specifications; for example Canon 450D EOS is 14bit; ZWO AISI 1600 MC is 12 or 10 bit; etc.

BUT How is Readout Depth?;  What value I must to put here?

I appreciate comments!!
Thanks
Federico
 
 

[Niall Saunders]

Hi Federico,

Readout Depth is something that you can think of in the same way as the ScreenTransfer Function - it acts as a finalinterface between the 'actual' value that PixInsight is using and what you, the user, sees on the screen.

Remember, your imager may have a built-in A/D converter that is, perhaps, running at 16-bits - and so it will store 16-bit integers in the image file during capture. However, you might only have a 14-bit A/D converter - but will still probably generate 16-bit integer values in your captured data.

Then, you come to read these images into your processing program (here, naturally, you have chosen to use PixInsight). The software may choose to store the pixel values (ADU values in reality) in any suitable format whilst the image is being held in RAM - typically, PixInsight defaults to storing the ADU values as 32-bit FloatingPoint values. Further, the original range of ADU integer values might have been something like 00000 to 65535 for a 16-bit A/D digitiser. However, PI will now hold that range as a 'Real' number between 0.000000 and 1.000000 (which allows for far easier internal processing).

Then you come along, as the user, and 'you' want to know what a particular value is. But, the question is, "What range would you like to use to represent the ADU range?". You are not confined to what the camera originally provided (via its A/D circuit), nor are you even confined to what PixInsight is using internally. PixInsight will present the data to you in a format that *you* want to visualise it in.

So, you get to choose - if you choose a REAL format, PixInsight even lets you set how many decimal places you want to see. If you choose a (BINARY) INTEGER format, then you get to choose the bit-depth (8-bit, 10-bit, 12-bit, 14-bit, 16-bit, etc.) - all that PI does is to perform a quick internal calculation from the ACTUAL value to the range that you have defined. You can also define a DECIMAL INTEGER format - where the range will be from 0 to some 'power of ten' - in otherwords, the brightest pixel would be assigned a displayed value of '1', followed by the 'number of zeroes' that the user specifies.

As always, flexibility is the name of the game. PixInsight puts you in control 

I hope this helps, but please ask for clarification where needed.

[freddy1]

Hello Mr Niall; Thanks very much for your reply

So them (if I understood correctly): The camera ADC (Analogical to Digital Converter (in my Canon 450D 14 bit) give a 0 - 16383 differents values for all the possible electrons catched by one pixel, them this ADC range will be, by the sofware, divided  in (for example using 16 bit) at 0 to 65535 new values ranges. Right ?... so then the ratio between this two scales are : 4

I made an exaple to take the gain value camera, using b]BasicCCDPapameters[/b] 

Gain using  16 bit Readout Depth  & 14 bit ADC   = 0,155 e- / ADU
Gain using  14 bit Readout Depth  & 14 bit ADC   = 0,621 e- / ADU

 The rate between gain :  0,621/0,155 = 4

I hope I understood the subject

Thanks very much
Federico


 

[Niall Saunders]

Hi Federico,

As I understand the option for the  Basic CCD Parameters Script :

You need to tick the CFA selection box
Readout Depth should be set to 16
A/D Bits should be set to 14

Then, you need to know what value is reported as the Maximum ADU value when you super-saturate an exposure (e.g. a long exposure taken during the day). The value that you want to use would be the one obtained when you open that exposure in PixInsight, and examine the image using the Readout Tool, for example.

Further, you would need to have set the readout tool, or Statistics Process, to give you the ADU value as a 16-bit Integer value (0 to 65535).

That should be all that you need (other than the images required to actually calculate the results.

In your case it is meaningless to calculate the gain for 14-bit readout depth as you are confined to 16-bit depth for your particular equipment configuration.

I hope this helps - and I am happy to stand corrected if anybody else wishes to comment.

[freddy1]

Hi Niall Thanks very much again

That´s raight as you say.
I made some a example and work very well
federico

[freddy1]

I return to the ring with this post, because I´ve another doubts related to Redaut Depth.

I have a Canon EOS 450D with 14 Bit ADC. 
I take a saturation FLAT frame (30 seconds ISO 800) and I see that the mean ADU value measured with Statistic tool setting at 16 Bit Readout Depht, was 15704 ADU instead of 65536 Which I expected to obtain by setting the 16 bit RD....

Could someone explain what happens?

Thank very much  !!

Federico

[Niall Saunders]

Hi Federico,

OK - you start with a 14-bit physical imager (in your DSLR). By the time that image gets into PixInsight, it will have, effectively, been transformed into a floating-point number (16-bit, 32-bit or 64-bit - it doesn't matter) in the range 0.000 -> 1.000. In this format. 50% exposure returns a value of 0.5, and 100% returns 1.000, etc.

But - then you inspect the image with a readout tool. That tool looks at this 'real, floating-point' number and scales it back into the range that you have selected in the Readout Options.

So, you might now see 50% full range as '32768' if you hhave selected a 16-bit unsigned range of 00000 - 65535. If you had selected an 8-bit unsigned ramge, this would have appeared as '128'.

As I said - nothing has actually changed - other than how you have asked PixInsight to 'present' the Readout Data.

Does that make any more sense?

[bulrichl]

Hi Frederico,

I have a Canon EOS 450D with 14 Bit ADC. 
I take a saturation FLAT frame (30 seconds ISO 800) and I see that the mean ADU value measured with Statistic tool setting at 16 Bit Readout Depht, was 15704 ADU instead of 65536 Which I expected to obtain by setting the 16 bit RD....

This is the expected result. The sensor of the camera outputs analog intensity values for each pixel. These analog values are converted to digital values by the ADC (analog-to-digital converter) of the camera. In your case, the ADC has a resolution of 14 bits, accordingly it outputs values between 0 and 2^14 - 1 (= 16383).

In order to get the raw data, you adjust 'Pure Raw' in Pixinisght, RAW format Preferences (Format Explorer: DSLR_RAW: Edit Preferences). When you load a CR2-file, 'dcraw' (the decoding software for raw images used by Pixinsight) decodes the data. Then the intensity values for 1 pixel are stored as 16-bit unsigned integers. This data format can represent values from 0 to 2^16 - 1 (= 65535). However, only the range from 0 to 16383 is used, because the ADC of your camera has a resolution of 14 bits. If the ADC of your camera had 12 bits, you would see values in the range 0 to 2^12 - 1 (= 4095).

Bernd

[freddy1]

Thanks Guys !!!

That's ok; But I'm still not understand how that's really work, I give an example:

3 differnts FLAT (saturated) by 3 differnets cameras. Measured using Statistics process at 16 bit

Canon EOS 450D 0,157 [e-/ADU] Exposure time: 32 sec
ADC :14 Bit (sensorgen sources data) (16384 steps)
Mean : 15704 ADU
Max: 16383 ADU

Nikon D3100  Gain 0,526 [e-/ADU]; exposure time: 180 sec
ADC :14 Bit (sensorgen sources data) (16384 steps)
Mean : 4072 ADU
Max: 4095 ADU

ZWO ASI 1600 MC Color Gain [e-/ADU] exposume time: 1 sec
ADC :12 Bit (4095 steps)
Mean : 65504 ADU
Max: 65504 ADU

Really extrange


I appreciate more comments

Thanks
Federico

[freddy1]

Attached

[Niall Saunders]

Hi again Freddy,

Strange? Perhaps - but not 'really strange" !!

With those three images - each being examined with the Statistics Process without (apparently) changing the configuration of the Process them, yes, you are seeing significantly 'different' results.

But - there isn't (I believe) any fault or problem with your data.

However, PixInsight may well be applying different algorithms when it opens each file. More importantly, we do not necessarily know what algorithms are being used when the capturing software actually creates the file in the first place.

For example, given an imager with 14-bit resolution, you would (correctly) expect to gather data in the range of [0 to ((2^{^14})-1)], or [0 to 16386] (a total of 16384 steps). However, there is nothing that mandates that this is how the camera should save that data - it could simply multiply every ADU value by '4' - and now your image file would be in the range of [o to 65535] - apparently now 65535 steps. But, you haven't really 'acquired' an extra 49152 steps at all - because the 'gap between steps' has changed from '1' (as it was originally) to '4' (now).

The data within the file, and its validity, remain the same during this sort of process - it is just that it becomes more difficult to interpret.

Looking then at your Statistics data for the three images:

The Canon has a 14-bit sensor, and would seem to 'correctly' hold these 14 bits within the confines of a 16-bit 'word' such that the maximum possible ADU value remains as ((2^{^14})-1).

Your Nikon is doing something 'strange'. It also has a 14-bit sensor, and so it should create images looking like those of your Canon. However, what it actually seems to have done is to drop the 'resolution' from 14 bits to 12 bits. This could be a setup configuration within the Nikon, or it could just be the way that Nikon does things. Either way, it doesn;t 'really' matter - remember that this is a DSLR, and as such is not 'perfectly' suited for long-exposure low-light astronomy work. 12-bit resolution should still give you enough data in any images that you choose to process.

Your ZWO (designed as an astro-imager) has 12-bit resolution, and then applies a x16 multiplier before it stores that data in the image file. As I described above, you haven't actually 'gained' anything, it just means that the 'brightest' pixel, with the highest ADU, will be represented as 65535 when PixInsight analyses the image data. Soon enough - once you start calibrating or post-processing this RAW frame, the initial 'gaps' in the ADU counts will be 'filled up' with the results of any applied processed.

SO, it might then appear that your 'highest resolution' is available form your Canon - but those 14 bits are only really useful if the least-significant bits represent data with a useable Signal-to-Noise ratio. If they do not (quite possible that this might be the case for a DSLR) then they might as well simply not be there. Calibration and post-procesing would identify these poor SNR candidates and would eliminate them soon enough. But, it is as well to have them present, just in case they do contain statistically viable data.

I know you are struggling to make sense of this - don't worry, a light-bulb might just suddenly turn on for you if someone can explain it to you in a way that helps. I am happy to keep on trying 

[freddy1]

Strange? Perhaps - but not 'really strange" !!

However, PixInsight may well be applying different algorithms when it opens each file. More importantly, we do not necessarily know what algorithms are being used when the capturing software actually creates the file in the first place.

For example, given an imager with 14-bit resolution, you would (correctly) expect to gather data in the range of [0 to ((2^{^14})-1)], or [0 to 16386] (a total of 16384 steps). However, there is nothing that mandates that this is how the camera should save that data - it could simply multiply every ADU value by '4' - and now your image file would be in the range of [o to 65535] - apparently now 65535 steps. But, you haven't really 'acquired' an extra 49152 steps at all - because the 'gap between steps' has changed from '1' (as it was originally) to '4' (now). ....

HI Nialll !!! Thanks very much !! yes yes !! Literally I slept with this  And I wake up this morning with the answer, the point (as you say) is the capture sofware range (in bit)

             ADC       Capture Sofware                 Px
Canon     14             14                                  16   --> will show max ADU 16384 Even though setting is16 bit
Nikon       14         Apparently 12 ??!!!           16   --> will show max ADU   4095 
ZWO        14             16 (SGC)                      16   --> will show max ADU 65536


As you remark:

[".....Your Nikon is doing something 'strange'. It also has a 14-bit sensor, and so it should create images looking like those of your Canon. However, what it actually seems to have done is to drop the 'resolution' from 14 bits to 12 bits....."]

yes is very extrange. I don't use a capture Sofware for this camera, I simply catch RAW images from the memory card of the camera, so I believe that the internal capture sofware into the camera have 12 bit output... this is very very extrage....  I don't Know why....The only reason That I supposeis is in order to have a quickly way to show an imagen at the camera...Nikon have an extrenal (for PC) capture Software but I don't teasting (120$)...

Thank very much Naill !!
Federico
 

[Niall Saunders]

Hi Federico.

If I have helped in any way, then I am glad to have been of assistance. But, please get back in touch if you are still having problems.