fredmiranda.com Login
This topic is in the 2015 archive. You can read it and like its posts here. It is read only.

Archive 2015High DR for 5d4

  
 
alundeb
Offline
Details
p.6 #1High DR for 5d4
snapsy wrote:
alundeb wrote
snapsy wrote
alundeb wrote
snapsy wrote
alundeb wroteNow I'm not following you.

But yes, the ADC is a linear device, and operates with tiers of fixed size 1 ADU if you want to put it that way. The ADC is tuned to the sensor's FWC, represented by the tier between 16382 and 16383.
Maybe we're looking at tiers from opposite end. The top tier of this ADC near FWC would capture e- values from 40,804 through 81,607. In your chart you list the average e- value of the top tier as 81608, leaving the entire MSB of the ADC's encoding dedicated to it.
No, the MSB (binary 10000000000000) represents 40804.
81608 is represented by binary 11111111111111, or the sum of the values represented by all bits.
There are two ranges - the input analog range of e- and the output range of digital ADUs. The MSB relates to the digital output of the ADC (ADUs), not the input (e-).
But there is a linear relationship between the input range and the output range, as shown in my first table.

I really don't understand why you call the range between MSB and FS a tier.
Yes, there is a linear relationship between the but the ADC is hashing a larger universe of analog e- values into a smaller universe of digital ADU values. Each doubling of e- values within the ADC's programmed input range correlates to the next MSB used in its output range (ie, a tier)...but the actual encoding of the bits correspond to ADU values (0..2^14-1) and not directly to e- values.


Yes, I can agree on that, and will try to follow your use of "tier".
What does the tier between 2^13 and (2^14) -1 have to do with DR measurements? It is only (2^14)-1 that is interesting for DR measurements.
Apr 14, 2015 at 10:34 AM
snapsy
Offline
Details
p.6 #2High DR for 5d4
alundeb wrote:
snapsy wrote
alundeb wrote
snapsy wrote
alundeb wrote
snapsy wroteMaybe we're looking at tiers from opposite end. The top tier of this ADC near FWC would capture e- values from 40,804 through 81,607. In your chart you list the average e- value of the top tier as 81608, leaving the entire MSB of the ADC's encoding dedicated to it.
No, the MSB (binary 10000000000000) represents 40804.
81608 is represented by binary 11111111111111, or the sum of the values represented by all bits.
There are two ranges - the input analog range of e- and the output range of digital ADUs. The MSB relates to the digital output of the ADC (ADUs), not the input (e-).
But there is a linear relationship between the input range and the output range, as shown in my first table.

I really don't understand why you call the range between MSB and FS a tier.
Yes, there is a linear relationship between the but the ADC is hashing a larger universe of analog e- values into a smaller universe of digital ADU values. Each doubling of e- values within the ADC's programmed input range correlates to the next MSB used in its output range (ie, a tier)...but the actual encoding of the bits correspond to ADU values (0..2^14-1) and not directly to e- values.
Yes, I can agree on that, and will try to follow your use of "tier".
What does the tier between 2^13 and (2^14) -1 have to do with DR measurements? It is only (2^14)-1 that is interesting for DR measurements.


To bring this back around, the point of me focusing on ADC bit tiers is because that is the constraint which limits the representable DR for a linear conversion. If you increase the range of input values (sensor DR) but keep the bit resolution of the ADC the same (14-bits), the tonal resolution for each EV is stretched/reduced and the encodeable DR remains the same. For me to understand how a 14-bit ADC can encode 15.8EV of DR I would like to see how each linear range of input values specifically gets encoded into output values. The e- -> ADU chart you presented is at 2x the scale of mine because you listed the MSB tier as FWC only rather than FWC through FWC/2. If you could re-present your chart using e- ranges like me instead of averages that might get us on the same page.
Apr 14, 2015 at 10:42 AM
alundeb
Offline
Details
p.6 #3High DR for 5d4
Ok:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
Apr 14, 2015 at 10:51 AM
technic
Offline
Details
p.6 #4High DR for 5d4
dhphoto wrote:
alundeb wroteI would have preferred High MP, low FPS and extended DR, but doubt it will come anytime soon.
I expect a majority here would, which is why I said Canon was shooting themselves in the foot. They make a big fuss about the oversized 7D2 sensor in the 5Ds/r, I think they are unlikely to put significantly superior image quality in a cheaper model.


They should be working on a 6D2 with such a sensor (with extended DR), as they still don't have a typical landscape/architecture camera (high DR, high MP, low FPS; relatively low weight and tilt screen also required for me). They better hurry, because by the time that 6D2 might come to market (2016?) Sony/Nikon will probably be another step ahead in FF sensor technology.
Apr 14, 2015 at 11:20 AM
alundeb
Offline
Details
p.6 #5High DR for 5d4
snapsy wrote:
alundeb wroteOk:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
You're depicting a resolution of 4 e- for each ADU but the actual e- resolution is less; it's 4.98 e- for each ADU based on a FWC of 81608. Your depiction is giving you 2x the linear representable e- range than is actually available. Again, a chart with one line for each e- to ADU tier range will be much simpler to follow. As I depicted earlier:

http://horshack.smugmug.com/photos/i-RNZFTLR/0/O/i-RNZFTLR.jpg


No, I am depicting 5 e- for each ADU
-2 -1 0 1 2
3 4 5 6 7
8 9 10 11 12

Sorry, but it is plain wrong to use bit tiers to represent the upper constraint on DR.
Apr 14, 2015 at 11:21 AM
snapsy
Offline
Details
p.6 #6High DR for 5d4
alundeb wrote:
snapsy wrote
alundeb wroteOk:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
You're depicting a resolution of 4 e- for each ADU but the actual e- resolution is less; it's 4.98 e- for each ADU based on a FWC of 81608. Your depiction is giving you 2x the linear representable e- range than is actually available. Again, a chart with one line for each e- to ADU tier range will be much simpler to follow. As I depicted earlier:

http://horshack.smugmug.com/photos/i-RNZFTLR/0/O/i-RNZFTLR.jpg
No, I am depicting 5 e- for each ADU
-2 -1 0 1 2
3 4 5 6 7
8 9 10 11 12

Sorry, but it is plain wrong to use bit tiers to represent the upper constraint on DR.


Yes but if I extrapolate out your 5 e- representation from FWC down through zero I get an e- range of -312 to -308 for ADU of zero. A tier corresponds to a doubling or halving of light, which is the linear ADC-represented relationship for each EV of DR. Why do you consider that wrong?

If you're dead-set against the brevity of a tier representation can you instead share a 16,384 row spreadsheet that shows each and every e- range -> ADU relationship?
Apr 14, 2015 at 11:26 AM
alundeb
Offline
Details
p.6 #7High DR for 5d4
snapsy wrote:
alundeb wrote
snapsy wrote
alundeb wroteOk:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
You're depicting a resolution of 4 e- for each ADU but the actual e- resolution is less; it's 4.98 e- for each ADU based on a FWC of 81608. Your depiction is giving you 2x the linear representable e- range than is actually available. Again, a chart with one line for each e- to ADU tier range will be much simpler to follow. As I depicted earlier:

http://horshack.smugmug.com/photos/i-RNZFTLR/0/O/i-RNZFTLR.jpg
No, I am depicting 5 e- for each ADU
-2 -1 0 1 2
3 4 5 6 7
8 9 10 11 12

Sorry, but it is plain wrong to use bit tiers to represent the upper constraint on DR.
Yes but if I extrapolate out your 5 e- representation from FWC down through zero I get an e- range of -312 to -308 for ADU of zero. A tier corresponds to a doubling or halving of light, which is the linear ADC-represented relationship for each EV of DR. Why do you consider that wrong?

If you're dead-set against the brevity of a tier representation can you instead share a 16,384 row spreadsheet that shows each and every e- range -> ADU relationship?


1). Please. I have rounded 4.98 to 5 to get whole numbers. Sometimes in between the rounding will give an occasional 4 e interval.

2) It is wrong because when we measure DR, we don't account for a possible doubling of the light intensity above saturation.

3) Please.

Apr 14, 2015 at 11:33 AM
snapsy
Offline
Details
p.6 #8High DR for 5d4
alundeb wrote:
snapsy wrote
alundeb wrote
snapsy wrote
alundeb wroteOk:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
You're depicting a resolution of 4 e- for each ADU but the actual e- resolution is less; it's 4.98 e- for each ADU based on a FWC of 81608. Your depiction is giving you 2x the linear representable e- range than is actually available. Again, a chart with one line for each e- to ADU tier range will be much simpler to follow. As I depicted earlier:

http://horshack.smugmug.com/photos/i-RNZFTLR/0/O/i-RNZFTLR.jpg
No, I am depicting 5 e- for each ADU
-2 -1 0 1 2
3 4 5 6 7
8 9 10 11 12

Sorry, but it is plain wrong to use bit tiers to represent the upper constraint on DR.
Yes but if I extrapolate out your 5 e- representation from FWC down through zero I get an e- range of -312 to -308 for ADU of zero. A tier corresponds to a doubling or halving of light, which is the linear ADC-represented relationship for each EV of DR. Why do you consider that wrong?

If you're dead-set against the brevity of a tier representation can you instead share a 16,384 row spreadsheet that shows each and every e- range -> ADU relationship?
1). Please. I have rounded 4.98 to 5 to get whole numbers. Sometimes in between the rounding will give an occasional 4 e interval.

2) It is wrong because when we measure DR, we don't account for a possible doubling of the light intensity above saturation.

3) Please.

I was just still trying to retrofit my linear bitwise conception of e- representation to your distribution. In other words, I wasn't being difficult, just obtuse Ok, I created a full spreadsheet using your depiction; I rounded the FWC to 81920 to make it a clean 5 e- per ADU for easier presentation. And I think it's finally starting to make sense to me. Thanks for your effort and patience.
Apr 14, 2015 at 11:55 AM
alundeb
Offline
Details
p.6 #9High DR for 5d4
snapsy wrote:
alundeb wrote
snapsy wrote
alundeb wrote
snapsy wrote
alundeb wroteOk:

http://i306.photobucket.com/albums/nn242/Overflate/ADU14_zpsubgmuesp.png

MSB is in the middle of the linear range and is not a constraint on DR. The FS value is a constraint on DR.
You're depicting a resolution of 4 e- for each ADU but the actual e- resolution is less; it's 4.98 e- for each ADU based on a FWC of 81608. Your depiction is giving you 2x the linear representable e- range than is actually available. Again, a chart with one line for each e- to ADU tier range will be much simpler to follow. As I depicted earlier:

http://horshack.smugmug.com/photos/i-RNZFTLR/0/O/i-RNZFTLR.jpg
No, I am depicting 5 e- for each ADU
-2 -1 0 1 2
3 4 5 6 7
8 9 10 11 12

Sorry, but it is plain wrong to use bit tiers to represent the upper constraint on DR.
Yes but if I extrapolate out your 5 e- representation from FWC down through zero I get an e- range of -312 to -308 for ADU of zero. A tier corresponds to a doubling or halving of light, which is the linear ADC-represented relationship for each EV of DR. Why do you consider that wrong?

If you're dead-set against the brevity of a tier representation can you instead share a 16,384 row spreadsheet that shows each and every e- range -> ADU relationship?
1). Please. I have rounded 4.98 to 5 to get whole numbers. Sometimes in between the rounding will give an occasional 4 e interval.

2) It is wrong because when we measure DR, we don't account for a possible doubling of the light intensity above saturation.

3) Please.
I was just still trying to retrofit my linear bitwise conception of e- representation to your distribution. In other words, I wasn't being difficult, just obtuse Ok, I created a full spreadsheet using your depiction; I rounded the FWC to 81920 to make it a clean 5 e- per ADU for easier presentation. And I think it's finally starting to make sense to me. Thanks for your effort and patience.


I took it as a sign of convergence that you were being obtuse. Thank you for not giving up, I am not good at explaining, but I have trust in your ability.
Apr 14, 2015 at 12:22 PM
Pixel Perfect
Offline
Details
p.6 #10High DR for 5d4
High DR from Canon sounds like an oxymoron, they already think they have high DR.
Apr 14, 2015 at 04:55 PM
EB-1
Offline
Details
p.6 #11High DR for 5d4
Does any of this help the high ISO noise that Canon is infamous for having?

EBH
Apr 14, 2015 at 05:01 PM
mttran
Offline
Details
p.6 #12High DR for 5d4
Pixel Perfect wrote:
High DR from Canon sounds like an oxymoron, they already think they have high DR.


Apr 14, 2015 at 08:17 PM
juststeve
Offline
Details
p.6 #13High DR for 5d4
The discussion about how many bits can dance on the head of a pixel is fascinating, but the one key feature of the rumored 5Div is it will have the dual pixel tech of the C300ii, the high dynamic range C300ii.

It is my guess Canon is not only using the dual pixel tech for focusing on the C300ii, but is also doing a dual readout at varying gain levels, similar to Magic Lantern, to get the extended dynamic range while maintaining a high level of resolution.

Another hypothetical would be regarding the 5DS and whether it possesses a scaled up 7Dii sensor, which is dual pixel. It seems the forums world has generally assumed that the 5DS sensor is the 7Dii sensor times 2. If it is actually, then perhaps a greater dynamic range result could be gotten from the 5DS sensor at some later date by firmware upgrades. It might even be possible for the 7Dii.

It seems to me the dual pixel technology is very significant. It may be limited currently by processing power, both for autofocusing and for exposure/dynamic range. Remember, the 7Dii is in reality a 40 mega-rectangular-pixel camera, the output of which is binned to 20 MP. There is much potential there. But more computational horsepower may be needed.
Apr 15, 2015 at 09:39 AM
FM ForumsCanon ForumJoin Upload & Sell
  
 

You are not logged inLog inRegister