EOS-1D X Mark II Claims of 15 Stops of DR [CR3]

[neuroanatomist]

When DxO test cameras and say > 14 stops of DR for Nikon, everyone here says "bullsh*t, DxO are stupid/wrong."

....

I suspect if Canon said "The Sun will rise in the west tomorrow" lots of people here would go "Cool! Where can I go and see it?"

Your understanding of facts is as astute as ever. The point was that DxO reported >14 stops of DR as a calculation based on downsampling an image to 8 MP, and the cameras reported with >14 stops of DR would be incapable of recording a scene with >14 stops of DR without clipping highlights, shadows or both.

I wonder if the sun rises in the west in dilbertland? :

On the other hand statement that 15 stops DR is more than enough is more than odd /strange.

I don't know that anyone here has argued that more DR wouldn't be better. It's just not a priority for everyone, even though for some it's obviously the sine qua non, and many of those people seem to think their opinion/desire is universally shared, despite substantial evidence to the contrary.

Ancient CR proverb: There are none so blind as those who must lift shadows by six stops to see.

 

[memoriaphoto]

...
Another example of this is Canon's HTP that gives you an additional stop of highlight recovery, but the lowest available ISO is 200. Or, as in my 1Dc: ISO400 (required for log) would give me 10+2=12 stops.

HTP is just for JPEG. It does not impact raw files at all.

Actutally, it does. I believe when activated, ISO is dropped one stop and then a low-mid curve is applied by the imageprocessor when the RAW file is produced by the camera.

The ALO however, is another story and only affects JPEG (and/or RAW conversions if done in DPP)

 

[docsmith]

When you really need lots of DR, 15 stops is not enough. DR is, however, basically a measure of noise, so lower noise means higher DR and better high ISO.

I do find this statement a bit odd, though I agree with some. When my camera clips highlights, and I try to expose for the highlights, making me raise the shadows, with noise and banding, the 11 stops of DR my sensor gives me is not enough. That does not make a 15 stop DR - sensor not necassary. Perhaps 13 stop would do in my case, which is still 2 stop more than I get from my current setup. Hence, a 15 stop sensor would be great, and more than enough.

Mt. Spokane's description fits my experience perfectly. I shoot with the 5DIII and I run into 2 types of "DR" issues. The first is bright sun and shadows. This far exceeds 10-15 stops DR. In other words, expose for the shadows and blow out the part of the frame in full light. Expose for the bright light and the shadows are completely dark. An additional stop or two of DR will not help me in this circumstance.

But, the other time I run into DR is noise in the blacks/shadows. This can be in astro or a few other situations, like with a background behind a waterfall. I do not run into it often (you know, proper exposure), but occasionally I see some shadow noise. Here, simply cleaning up noise in the blacks/shadows will help my images and expand DR by 1-2 stops. From what I've gathered over the years on this forum and others is that the primary cause of this issue is noise that is gained as the analog signal moves from the chip to off the chip before being converted to a digital signal, thus, the hope is that on chip A/D converters would minimize this noise and the issue with black/shadow noise.

Quickly on the question someone had about 15 bits needed for 15 stops of DR, my understanding that this is needed is because each "bit" is really a digit in a binary sequence. So a 4 "bit" sequence is 0000, 0101, 1111, etc. 5 bit is 00000, 11111, etc. For simplicity in understanding why 15 bits are needed for 15 stops of DR, I imagine that each bit measures the light filling up the pixel well behind a bayer sensor. So, in my 4 bit system, 0000 would be black, and 1111 would be completely bright. So, to define 15 stops of brightness, you need 15 digits in a binary sequence to quantify that light.

Simplistic, but that is my level of understanding. If others know better, please expand.

 

[AndreeOnline]

Simplistic, but that is my level of understanding. If others know better, please expand.

2bit would be 22 levels of luminance = 4
8bit would be 28 levels of luminance = 256
10bit would be 210 levels of luminance = 1024
14bit would be 214 levels of luminance = 16384

For 15 stops, as I've already stated above, you need 32768 or 215 (15bit).

This, of course, assumes a single exposure or only one image processing pipeline. I can imagine an 14bit solution with parallell processing (sort of like internal HDR bracketing) that generates 15 stop files.

15 stops can be compressed into 14, or even 10bit files for that matter. But you can't take a single 14bit originated file and expand it to 15 stops.

 

[jeffa4444]

Here are your 15 stops broken down to discreet levels. Each stop must per definition be a doubling of the previous value in a linear representation:

0. Stop: 0-1
1. Stop: 1-2
2. Stop: 2-4
3. Stop: 4-8
4. Stop: 8-16
5. Stop: 16-32
6. Stop: 32-64
7. Stop: 64-128
8. Stop: 128-256
9. Stop: 256-512
10. Stop: 512-1024
11. Stop: 1024-2048
12. Stop: 2048-4096
13. Stop: 4096-8192
14. Stop: 8192-16384
15. Stop: 16384-32768

Since it would be very impractical to assign 1 value to the first stop and 16384 values to the last stop, we use a logarithmic function to distribute the values in a non-linear fashion. The log curve will assign roughly the same amount of discreet values per stop.

Now, you might think that you can take these values and, with the help of a log curve, break them out to even more "stops" by using a flatter profile. But the only thing that means is that you start to define fractions of a real, linear light stop. So that isn't possible.

In practical terms, on a signal level, Dynamic Range (DR) in dB is calculated:

DR (db) = 20 log (Peak signal at Full Well Saturation/r.m.s. Noise)

For an image sensor, in essence, each stop will correspond to a 6dB change.

15 stops would in theory need a 90dB sensor. The C300 mkII achieves 67dB (which Canon is pretty proud of). 67dB is equivalent to 11.17 stops. So, there is a discrepancy here.

My own little theory is that cameras recording log images do some significant pulling of the signal (under exposing) and then use the ISO to boost the levels back up (and recover the shadows). Kind of like a dual ISO thing.

Meaning if you shoot at ISO800 you get 3 stops highlight protection (based off of ISO100). 9 real stops + 3 ISO stops = 12 stops. A Sony that requires ISO3200 for log would give you a whopping 5 stops extra. 9 real stops + 5 ISO stops = 14 stops.

This technique would be especially advantageous if the sensor has low readout noise—something the Sony sensors have been good at. And now the new generation of Canon sensors too, I think.

Another example of this is Canon's HTP that gives you an additional stop of highlight recovery, but the lowest available ISO is 200. Or, as in my 1Dc: ISO400 (required for log) would give me 10+2=12 stops.

I completely agree with your evauation in terms of dB per stop. Arri use a Esser light sphere with a Esser Plates to accurately measure their cameras and indeed developed one with Esser to read the Alexa. We have the same device which can be used for stills cameras as well as video cameras and we use it to make sure manufacturers claims are accurate. Both DXOMark and Cinema5D methods are open to a degree of interpritation whereas the Esser method we have found gives very accurate results and shows the variation even of cameras of the same type (very important in visual effects).
We have not tested the Canon C300 MKII yet but have tested cameras like the Sony F55 & F5 which give more DR than the FS7 but not Alexa levels.
The Alexa 65 uses the same sensor (three stitched sideways) as the regular Alexa XT and gives the same 14 stop DR.
Cinematographers will always use more DR, however we fixate on it yet dont about the limitations created by CFAs which "lock in" to a certain degree the colorimagery thats why the big advances going forwards will come from this area.

 

[neuroanatomist]

...
Another example of this is Canon's HTP that gives you an additional stop of highlight recovery, but the lowest available ISO is 200. Or, as in my 1Dc: ISO400 (required for log) would give me 10+2=12 stops.

HTP is just for JPEG. It does not impact raw files at all.

Actutally, it does. I believe when activated, ISO is dropped one stop and then a low-mid curve is applied by the imageprocessor when the RAW file is produced by the camera.

Neither of you are correct, although dilbert's statement is closer to the reality. HTP does not directly affect the RAW image data, the tone curve is applied only to the jpg image. However, although the RAW image data aren't directly affected, the RAW file is affected because the metadata are recorded incorrectly. What HTP does is deliberately underexpose by one stop, and 'misrecord' the ISO in the metadata - that's why ISO 100 isn't available when you turn on HTP, i.e. you set ISO 200, it shoots at ISO 100 but records 200, or you set ISO 800, it shoots at 400 and records 800. If shooting JPG, it processes the underexposed image to brighten everything except the highlights (meaning it applies a tone curve). If shooting RAW, it sets a metadata flag so DPP can apply that tone curve.

If you open that RAW file in a 3rd party converter, results vary. Some ignore the flag and you just get an underexposed image. Others compensate by just boosting the total exposure by one stop - and that just re-blows your highlights and adds shadow noise.

 

[Pitbullo]

I do find this statement a bit odd, though I agree with some. When my camera clips highlights, and I try to expose for the highlights, making me raise the shadows, with noise and banding, the 11 stops of DR my sensor gives me is not enough. That does not make a 15 stop DR - sensor not necassary. Perhaps 13 stop would do in my case, which is still 2 stop more than I get from my current setup. Hence, a 15 stop sensor would be great, and more than enough.

Mt. Spokane's description fits my experience perfectly. I shoot with the 5DIII and I run into 2 types of "DR" issues. The first is bright sun and shadows. This far exceeds 10-15 stops DR. In other words, expose for the shadows and blow out the part of the frame in full light. Expose for the bright light and the shadows are completely dark. An additional stop or two of DR will not help me in this circumstance.

But, the other time I run into DR is noise in the blacks/shadows. This can be in astro or a few other situations, like with a background behind a waterfall. I do not run into it often (you know, proper exposure), but occasionally I see some shadow noise. Here, simply cleaning up noise in the blacks/shadows will help my images and expand DR by 1-2 stops. From what I've gathered over the years on this forum and others is that the primary cause of this issue is noise that is gained as the analog signal moves from the chip to off the chip before being converted to a digital signal, thus, the hope is that on chip A/D converters would minimize this noise and the issue with black/shadow noise.

My go-to lens is the Sigma 35mm ART.


Simplistic, but that is my level of understanding. If others know better, please expand.

Usually my problems start when I take pictures of the kids outside in the afternoon. Low sun, not too bright outside, and quite fast moving subjects. Perhaps ISO 800 to get the shutter speed right. DR is a problem then. Either blown highlights or misserable shadows. A few more stops of DR (esp. at ISO above base 400-3200) would be great.

 

[davidmurray]

Quickly on the question someone had about 15 bits needed for 15 stops of DR, my understanding that this is needed is because each "bit" is really a digit in a binary sequence. So a 4 "bit" sequence is 0000, 0101, 1111, etc. 5 bit is 00000, 11111, etc. For simplicity in understanding why 15 bits are needed for 15 stops of DR, I imagine that each bit measures the light filling up the pixel well behind a bayer sensor. So, in my 4 bit system, 0000 would be black, and 1111 would be completely bright. So, to define 15 stops of brightness, you need 15 digits in a binary sequence to quantify that light.

Simplistic, but that is my level of understanding. If others know better, please expand.

The signal that has the dynamic range is analogue. We want the same degrees of difference between each stop, so it in reality doesn't matter if we take a 8/10/12/14/15/16 stop analog dynamic range and map the analog signal onto arbitrary digital values.

What does matter is whether or not there are enough digital values and the gaps in the analog values they represent are sufficiently small so that digital quantization noise is at or below the analog noise floor.

So if the mapping of the analog signal onto digital values is done carefully the analog DR shouldn't be a problem.

Or, to put it another way, adding an additional 6dB to the analog signal doesn't necessarily mean you need another digital bit to record the values. You MIGHT, but not necessarily. It all depends on how big the gaps in the analog voltage becomes between each digital value representing the analog voltage.

 

[rfdesigner]

Quickly on the question someone had about 15 bits needed for 15 stops of DR, my understanding that this is needed is because each "bit" is really a digit in a binary sequence. So a 4 "bit" sequence is 0000, 0101, 1111, etc. 5 bit is 00000, 11111, etc. For simplicity in understanding why 15 bits are needed for 15 stops of DR, I imagine that each bit measures the light filling up the pixel well behind a bayer sensor. So, in my 4 bit system, 0000 would be black, and 1111 would be completely bright. So, to define 15 stops of brightness, you need 15 digits in a binary sequence to quantify that light.

Simplistic, but that is my level of understanding. If others know better, please expand.

The signal that has the dynamic range is analogue. We want the same degrees of difference between each stop, so it in reality doesn't matter if we take a 8/10/12/14/15/16 stop analog dynamic range and map the analog signal onto arbitrary digital values.

What does matter is whether or not there are enough digital values and the gaps in the analog values they represent are sufficiently small so that digital quantization noise is at or below the analog noise floor.

So if the mapping of the analog signal onto digital values is done carefully the analog DR shouldn't be a problem.

it's more complex than that.

The analogue signal has intrinsic noise in it, equal to the square root of the signal (because the original signal is a number of electrons).

The brightest pixels do not need 14 bits recorded for them, in reality 9 would be more than adequate, just so long as we know it's the brightest 9 we're talking about.

A non-linear curve between the ADC and recorded file is fine, just so long as it is completely defined and reversible without error (no reason why this is not possible)

 

[davidmurray]

Quickly on the question someone had about 15 bits needed for 15 stops of DR, my understanding that this is needed is because each "bit" is really a digit in a binary sequence. So a 4 "bit" sequence is 0000, 0101, 1111, etc. 5 bit is 00000, 11111, etc. For simplicity in understanding why 15 bits are needed for 15 stops of DR, I imagine that each bit measures the light filling up the pixel well behind a bayer sensor. So, in my 4 bit system, 0000 would be black, and 1111 would be completely bright. So, to define 15 stops of brightness, you need 15 digits in a binary sequence to quantify that light.

Simplistic, but that is my level of understanding. If others know better, please expand.

The signal that has the dynamic range is analogue. We want the same degrees of difference between each stop, so it in reality doesn't matter if we take a 8/10/12/14/15/16 stop analog dynamic range and map the analog signal onto arbitrary digital values.

What does matter is whether or not there are enough digital values and the gaps in the analog values they represent are sufficiently small so that digital quantization noise is at or below the analog noise floor.

So if the mapping of the analog signal onto digital values is done carefully the analog DR shouldn't be a problem.

it's more complex than that.

The analogue signal has intrinsic noise in it, equal to the square root of the signal (because the original signal is a number of electrons).

The brightest pixels do not need 14 bits recorded for them, in reality 9 would be more than adequate, just so long as we know it's the brightest 9 we're talking about.

A non-linear curve between the ADC and recorded file is fine, just so long as it is completely defined and reversible without error (no reason why this is not possible)

Yes - agreed.

The key is the ability to take the digital value and get back the original electrical voltage without introducing more noise than what was in the original analog signal.

 

[kaihp]

Simplistic, but that is my level of understanding. If others know better, please expand.

2bit would be 22 levels of luminance = 4
8bit would be 28 levels of luminance = 256
10bit would be 210 levels of luminance = 1024
14bit would be 214 levels of luminance = 16384

For 15 stops, as I've already stated above, you need 32768 or 215 (15bit).

This, of course, assumes a single exposure or only one image processing pipeline. I can imagine an 14bit solution with parallell processing (sort of like internal HDR bracketing) that generates 15 stop files.

15 stops can be compressed into 14, or even 10bit files for that matter. But you can't take a single 14bit originated file and expand it to 15 stops.

All this assumes integer calculations - which isn't a bad assumption at all. But it might be that Canon chooses to store the luminance levels as a floating-point number. It's more computational work, but it could definitely be done to reduce the data to a (say) 14 or even 12 bit number.

 

[rrcphoto]

who's to say it's 14 bit RAW data?

canon was the first to go 14 bit, they could conceivably be the first to go to 16 bit (in the non medium format space that is).

Historically the first camera they upped the bit depth on was the 1 series (the 1D Mark III was the first with 14 bit RAW data).

there's alot of talk / whining / hand wringing and none of you explored the possibility that canon just simply went to 15/16 bit RAW files

*IF* canon implemented the ADC patents which are dual slope ADC - they are not ADC in the traditional sense, but a time to digital value. quantifying that could easily occur with a variety of bit depths.

 

[3kramd5]

When Canon does a press release and says "15 stops of DR", everyone goes "wow, cool."...
And not only that, people are accepting Canon's "15 stops of DR" statements over testing that actually shows less.

Did you and I read the same thread?

15 stops at what resolution?

If on the single pixel level, that would truly be phenomenal!

If on a scaled down image of 1 Megapixel, that wouldn't be very impressive.

When you really need lots of DR, 15 stops is not enough. DR is, however, basically a measure of noise, so lower noise means higher DR and better high ISO.

One more claim I'm not sure I'd believe completely. Hoping for best of course, but...

etc

 

[PureClassA]

At 16 bit, how much larger of a file are we talking here though? It's pretty substantial. I think Canon would give serious consideration to that. As previously mentioned, the full "15 stops of DR" Canon is boasting is likely something firmware driven that can/will only be recoverable within proprietary software like DPP. Once the RAW file gets imported and interpreted, it can be exported in a loss-less TIFF for LR or PS manipulation if desired. I'm not saying they won't go 16bit RAW files, just saying I'd be surprised if they did. The DR on current Canon bodies is more than satisfactory for most pros, however I understand there are SOME shots where more would help and that there are those who specialize in certain styles where more is commonly demanded. Canon isn't a panacea for everyone, but do serve the majority of pro togs very well. That being said, if you have a shot where you really need to squeeze every drop of highlight and shadow manipulation out of the widest image range possible on the new Canon models, you'll need to use DPP for those particular shots first. (I would suspect)

who's to say it's 14 bit RAW data?

canon was the first to go 14 bit, they could conceivably be the first to go to 16 bit (in the non medium format space that is).

Historically the first camera they upped the bit depth on was the 1 series (the 1D Mark III was the first with 14 bit RAW data).

there's alot of talk / whining / hand wringing and none of you explored the possibility that canon just simply went to 15/16 bit RAW files

*IF* canon implemented the ADC patents which are dual slope ADC - they are not ADC in the traditional sense, but a time to digital value. quantifying that could easily occur with a variety of bit depths.

 

[rfdesigner]

At 16 bit, how much larger of a file are we talking here though? It's pretty substantial. I think Canon would give serious consideration to that. As previously mentioned, the full "15 stops of DR" Canon is boasting is likely something firmware driven that can/will only be recoverable within proprietary software like DPP. Once the RAW file gets imported and interpreted, it can be exported in a loss-less TIFF for LR or PS manipulation if desired. I'm not saying they won't go 16bit RAW files, just saying I'd be surprised if they did. The DR on current Canon bodies is more than satisfactory for most pros, however I understand there are SOME shots where more would help and that there are those who specialize in certain styles where more is commonly demanded. Canon isn't a panacea for everyone, but do serve the majority of pro togs very well. That being said, if you have a shot where you really need to squeeze every drop of highlight and shadow manipulation out of the widest image range possible on the new Canon models, you'll need to use DPP for those particular shots first. (I would suspect)

who's to say it's 14 bit RAW data?

canon was the first to go 14 bit, they could conceivably be the first to go to 16 bit (in the non medium format space that is).

Historically the first camera they upped the bit depth on was the 1 series (the 1D Mark III was the first with 14 bit RAW data).

there's alot of talk / whining / hand wringing and none of you explored the possibility that canon just simply went to 15/16 bit RAW files

*IF* canon implemented the ADC patents which are dual slope ADC - they are not ADC in the traditional sense, but a time to digital value. quantifying that could easily occur with a variety of bit depths.

it's only 2/14ths larger.

So a 22MPixel 16 bit file will look much like a 19MPixel 14 bit file, assuming the same number of bits are thrashing at the bottom.

 

[Maiaibing]

We’re told that Canon will claim 15+ stops of dynamic range for the new Canon EOS-1D X Mark II. This claim was also made for the Cinema EOS C300 Mark II, which may be true at the hardware level, but in practice it may not actually perform to that specification. </p>
<span id="pty_trigger"></span>

Wow. If Canon wants to attract bad press and flak at the release of the 1D X II making a baseless/deceptive claim on one of Canon's perceived shortcomings towards its direct competition must be the sure-fire way of getting it.

I hope for Canon they are smarter than this. So Canon either does not make the claim or delivers were it counts...

 

[FEBS]

We’re told that Canon will claim 15+ stops of dynamic range for the new Canon EOS-1D X Mark II. This claim was also made for the Cinema EOS C300 Mark II, which may be true at the hardware level, but in practice it may not actually perform to that specification. Let’s hope it’s actually the case with the new 22mp sensor.</p>
<p>Specifications for this camera have been extremely slow to come in from known sources. There is definitely a much tighter ship being run at Canon, but we do expect more to leak out as we approach an announcement in the next 4-8 weeks.</p>
<span id="pty_trigger"></span>

If the RAW files are still 14 bit, then say no to 15 stops of dynamic range..... If the RAW files are 16 bit, then say hello!

+1

I hope to say hello, but a change of 14 to 16 bits sampling is a major upgrade!

 

[kubelik]

When Canon does a press release and says "15 stops of DR", everyone goes "wow, cool."...
And not only that, people are accepting Canon's "15 stops of DR" statements over testing that actually shows less.

Did you and I read the same thread?

15 stops at what resolution?

If on the single pixel level, that would truly be phenomenal!

If on a scaled down image of 1 Megapixel, that wouldn't be very impressive.

When you really need lots of DR, 15 stops is not enough. DR is, however, basically a measure of noise, so lower noise means higher DR and better high ISO.

One more claim I'm not sure I'd believe completely. Hoping for best of course, but...

etc

he's just trolling. everyone else is here having a pretty enlightening discussion on how it's unlikely that Canon has actually packed in 15 stops of DR, how it could be claimed through marketing nonsense/obfuscation, or how it could actually happen if Canon used the right technology and gave us 16-bit files.

 

[neuroanatomist]

I hope to say hello, but a change of 14 to 16 bits sampling is a major upgrade!

Unless you're Sony, who 'innovatively' claim 16-bit image processing.

 

[PureClassA]

it's only 2/14ths larger.

So a 22MPixel 16 bit file will look much like a 19MPixel 14 bit file, assuming the same number of bits are thrashing at the bottom.

Forgive my ignorance. Ya lost me. You're saying a 16bit RAW file from 22MP sensor will have a similar file size to a 14bit RAW file from a 19MP sensor?? I'm missing something. How could that be smaller/same size?