Pages: [1] 2 3 4   Go Down

Author Topic: Canon 5Ds Dynamic range--any accurate information not just speculation??  (Read 36502 times)

gdh

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 539
    • https://www.facebook.com/Gallery833-Fine-Art-in-Redding-160225810674500/

Canon 5Ds Dynamic range--any accurate information not just speculation??

I've researched the specs but none seem to deal with dynamic range--anyone have any specifics? I've read the speculations and the guesses based on extrapolations but is there anything official from Canon or from someone who has actually tested(if that's possible at this point)?

Thanks for any information


Dennis

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

Canon 5Ds Dynamic range--any accurate information not just speculation??

I've researched the specs but none seem to deal with dynamic range--anyone have any specifics? I've read the speculations and the guesses based on extrapolations but is there anything official from Canon or from someone who has actually tested(if that's possible at this point)?

Hi Dennis,

Not necessarily 100% representative (because from a pre-production model), but according to RawDigger I get the following suggested values based on what's assumed to be masked pixels that only exhibit read-noise:

READ-NOISE (StdDev) @ ISO 100, before Black-Point subtraction:
R  = 5.82 , G  = 5.85 , B  = 5.77 , G2= 5.83 .
Saturation (14-bit) is presumably at 14733, although one color plane has an outlier of 15441, and an un-clipped black-point of on average 2047.

That would give an engineering dynamic range of Log(14733/5.82)/Log(2) = 11.31 stops (screen mode in DxOMark terminology).

All as expected from the info that was given that it would be comparable to 5D3 levels (which has significantly large sensels).

Cheers,
Bart

P.S. If one were to subtract the Black-point from all readings, that would bring the DR to 11.8 stops.
« Last Edit: May 02, 2015, 03:27:03 pm by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog

Not necessarily 100% representative (because from a pre-production model), but according to RawDigger I get the following suggested values based on what's assumed to be masked pixels that only exhibit read-noise:

READ-NOISE (StdDev) @ ISO 100, before Black-Point subtraction:
R  = 5.82 , G  = 5.85 , B  = 5.77 , G2= 5.83 .
Saturation (14-bit) is presumably at 14733, although one color plane has an outlier of 15441, and an un-clipped black-point of on average 2047.

That would give an engineering dynamic range of Log(14733/5.82)/Log(2) = 11.31 stops (screen mode in DxOMark terminology).

Hi Bart,

All in line with what Canon said, it's basically the 7DII pixel in a FF area then?

P.S. If one were to subtract the Black-point from all readings, that would bring the DR to 11.8 stops.

Have you tried a PTC?

Jack
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

All in line with what Canon said, it's basically the 7DII pixel in a FF area then?

Hi Jack,

Yes, it seems to look similar, although the sensels have a slightly different pitch. So it is probably somewhat different, but maybe pretty similar in technology.

Quote
Have you tried a PTC?

No, I just have 2 Raws, and I have difficulty seeing at first glance what's different in the settings that were used. The camera was running at some 33 degrees Celsius, according to the EXIF, so I don't know if that is typical or not. Maybe it was running a bit hot due to a long shooting sequence.

I prefer to use sequences of 1/3rd stop real exposure frames, because we do not know if something specific is happening with noise reduction. I also prefer to subtract image pairs, to eliminate pattern noise and PRNU and improve overall robustness of the data.

Canon usually doesn't fiddle with the Raw data, no noise reduction, no lossy compression or tone curves, doesn't pre-scale for White Balance, and leaves the black-points intact. However, I'd prefer to have that confirmed in practice, rather than rely solely on masked pixels. We also have seen in the past that e.g. apertures wider than f/2.8 have an influence on gain, just as DxOMark suggested, so actual practice may show other things as well that a single frame might not reveal. I also do not know if the nominal ISO 50 is any different from ISO 100, or if there are other jumps or a just a linear increase in gain as ISO goes up.

Cheers,
Bart
« Last Edit: May 02, 2015, 05:17:18 pm by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

gazwas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 539

Having now tried the (natural looking) HDR feature in Lightroom CC the DR of the new Canon 5DS is of much less concern now than the quality of the pixels.
Logged
trying to think of something meaningful........ Err?

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog

Yes, sensors as we know them have been eclipsed by the introduction of this disruptive technology.  From now on all we will need is a single high quality pixel, shifted/stacked/stitched ;)
Logged

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375

Having now tried the (natural looking) HDR feature in Lightroom CC the DR of the new Canon 5DS is of much less concern now than the quality of the pixels.

I've used HDR in both Photoshop and Photomatix Pro.  While you can do some really great things with either, neither is a substitute for available DR at time of capture.  They are work arounds.

Most current sensors can capture the DR of the vast majority of imaging opportunities.  However, expanded DR sensors allow some forgiveness and are obviously better when extended DR opportunities present themselves.  All else equal, it is better to have the higher DR sensor at your command.
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

I've used HDR in both Photoshop and Photomatix Pro.  While you can do some really great things with either, neither is a substitute for available DR at time of capture.  They are work arounds.

On the contrary. Properly executed HDR imaging will trounce the 14 stop DR that a single sensor shot can hope to achieve. The best solution for DR is to record more photons and thus boost the Signal to Noise ratio. Obviously not all subjects are equally suited for bracketed exposures, but then not all of them need to be.

Quote
Most current sensors can capture the DR of the vast majority of imaging opportunities.  However, expanded DR sensors allow some forgiveness and are obviously better when extended DR opportunities present themselves.  All else equal, it is better to have the higher DR sensor at your command.

It always helps, but there is more than one road that leads to technical quality.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Petrus

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 952

Looking at the dpreview samples the 11 something stop figure for 5Ds is believable. Shadow detail is slightly less than with D810.

But like I have said before, internal lens reflections start to mess with maximum shadow detail even before that 14 stop figure the Nikon sensor alone can achieve. It the the sensor-lens combination that matters in the end.
« Last Edit: May 08, 2015, 02:40:49 pm by Petrus »
Logged

gazwas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 539

All else equal, it is better to have the higher DR sensor at your command.

All else being equal it would be nice if the 5Ds had 18 stops of DR but unfortunately it doesn't.

What has been demonstrated from the limited RAW files available and with first generation file support in Lightroom is a very sharp, detailed high resolution file with great colour (better than the Sony/Nikon?) and with the help of the new HDR engine in Lightroom CC an excellent solution to my 18+ stop request.

I'm still on the fence between the 5Ds and the rumoured Sony competitor but the DR difference is no longer in contest to which I choose.
« Last Edit: May 08, 2015, 11:26:05 am by gazwas »
Logged
trying to think of something meaningful........ Err?

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375

On the contrary. Properly executed HDR imaging will trounce the 14 stop DR that a single sensor shot can hope to achieve. The best solution for DR is to record more photons and thus boost the Signal to Noise ratio. Obviously not all subjects are equally suited for bracketed exposures, but then not all of them need to be.

It always helps, but there is more than one road that leads to technical quality.

Cheers,
Bart

On the contrary, in extreme DR situations, the higher DR sensor, all else equal, can execute the HDR multi shot scenario just as well AND provide a single capture of better quality than the lower DR sensor.  Hence, the higher DR camera will allow one to execute a larger universe of shots than the lower DR camera.  The 1st question to ask is how much DR is required for the shots intended to be executed with this camera?  If the answer to that question is more than the DR provided, then only the incompetent would prefer to have less DR at their disposal than more.

And you are correct that there is usually more than 1 way to skin the cat, but the higher DR sensor is the one that provides the most options and flexibility.

Logged

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog

On the contrary. Properly executed HDR imaging will trounce the 14 stop DR that a single sensor shot can hope to achieve.

Hi Bart,

I would agree in theory and only in images where there are effectively two separate, different exposures (say inside the room and outside the window).  Otherwise in continuous tone images in practice I've found HDR blending by the usual best practice suspects to almost always provide less than satisfactory (to me) transitions so far (say 14 stops by a single capture vs 3x 11 stop captures 1 or so stop apart). YMMV.

Jack
« Last Edit: May 08, 2015, 01:34:04 pm by Jack Hogan »
Logged

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2005
    • http://www.guillermoluijk.com

I would agree in theory and only in images where there are effectively two separate, different exposures (say inside the room and outside the window).  Otherwise in continuous tone images in practice I've found HDR blending by the usual best practice suspects to almost always provide less than satisfactory (to me) transitions so far (say 14 stops by a single capture vs 3x 11 stop captures 1 or so stop apart). YMMV.

If you found the results of HDR unsatisfactory it was because of the software you used and/or the way you used it, not because of the real potential of a bracketed series of shots.

Information contained in 3 shots 1EV apart from an 11-stops DR sensor CAN be linearly fused to achieve the equivalent to a single shot from a 13-stops DR sensor. The general formula would be:

DR_equivalent = DR_sensor + (N_shots - 1) * EV_interval
13 = 11 + (3 - 1) * 1
15 = 11 + (3 - 1) * 2
...

This DNG file contains the best information that could be gathered from 2 shots 4EV apart from a 8-stops DR sensor (Canon 350D). It is indistinguisable in terms of DR from a 12-stops DR sensor RAW file (e.g. a Nikon D800 RAW file):

12 = 8 + (2 - 1) * 4

HDR composite


« Last Edit: May 08, 2015, 08:00:25 pm by Guillermo Luijk »
Logged

LKaven

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1060

See Bill Claff's preliminary numbers charted here:

http://www.dpreview.com/forums/post/55774910

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog

If you found the results of HDR unsatisfactory it was because of the software you used and/or the way you used it, not because of the real potential of a bracketed series of shots.

Thanks Guillermo.  Yes, that was my complaint.

Jack
Logged

dwswager

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1375

If you found the results of HDR unsatisfactory it was because of the software you used and/or the way you used it, not because of the real potential of a bracketed series of shots.

Information contained in 3 shots 1EV apart from an 11-stops DR sensor CAN be linearly fused to achieve the equivalent to a single shot from a 13-stops DR sensor. The general formula would be:

DR_equivalent = DR_sensor + (N_shots - 1) * EV_interval
13 = 11 + (3 - 1) * 1
15 = 11 + (3 - 1) * 2

This is inarguable true, but not the point.  Yes, if your camera is DR limited below the DR of the shot you are trying to execute AND you have the ability to shoot an exposure bracket set AND you have the tools and talent to post process the shots, you can execute it.  Or you can execute the shot in a single capture with much simpler file handling and processing with a sensor that provides 13 or more stops of DR.  In addition, you can also execute shots that don't allow for a bracket set and exposure stacking. 
Logged

gazwas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 539

This is inarguable true, but not the point.  Yes, if your camera is DR limited below the DR of the shot you are trying to execute AND you have the ability to shoot an exposure bracket set AND you have the tools and talent to post process the shots, you can execute it.  Or you can execute the shot in a single capture with much simpler file handling and processing with a sensor that provides 13 or more stops of DR.  In addition, you can also execute shots that don't allow for a bracket set and exposure stacking.  

In theory yes but often the need to composite or HDR an image is because it falls outside the DR capabilities of any photographic sensor currently available. Sure we all want more of everything but wanting a Sony sensor solely for its DR just (for me) became obsolete if all else remains unchanged (36Mpix and 13 stops DR) on the new A7rII as the 5DsR files I just downloaded look amazing.
Logged
trying to think of something meaningful........ Err?

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog

... as the 5DsR files I just downloaded look amazing.

Looking amazing indeed.  But the amazingness range is not what is being discussed here - especially when evaluated on typically 8-stop DR monitors or prints.
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913

In theory yes but often the need to composite or HDR an image is because it falls outside the DR capabilities of any photographic sensor currently available.

I agree. And even a modest +2EV exposure bracket will improve the signal(!) level of the deepest shadows by 4x over the same signal level that gets its detail from the higher DR of a Sony sensor without bracketing. Granted, the read noise remains at a relatively higher level, but it's hard to beat real signal which becomes 30% less noisy  because it has more photons. That's significant, and helps shadow color accuracy. So even a recent Sony sensor would benefit from exposure bracketing, less out of necessity, but shadow quality still benefits.

Quote
Sure we all want more of everything but wanting a Sony sensor solely for its DR just (for me) became obsolete if all else remains unchanged (36Mpix and 13 stops DR) on the new A7rII as the 5DsR files I just downloaded look amazing.

Indeed. I also wouldn't object to a sensor with higher native DR, but the image quality already does look quite nice (even from an ACR conversion). I'm looking forward to how Capture One does. And given that the noise of the 5DS looks much better than we're used to, a mild noise reduction will not hurt detail but benefit the 'pushability' of shadows. Canon Raw files may also benefit from the fact that Read noise is not clipped, and therefore a good Rawconverter can be very effective in addressing noise in the Raw stage of conversion.

Color rendition is mainly a matter of profiles, but the images that I've seen so far do seem to have a slightly richer color. Maybe the profile does it, but apparently there is little wrong with the data it can work with.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

gazwas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 539

Looking amazing indeed.  But the amazingness range is not what is being discussed here - especially when evaluated on typically 8-stop DR monitors or prints.

Indeed but the main criticism I've read why people think the 5Ds is a fail is all down to DR even before they have seen (until now) RAW files that demonstrate what the camera can do. Sure, some people don't like the effort of using polarisers or grad filters to tame contrast and only recognise a good camera as one which can be pushed and pulled in post but I see no problem now with the 11.8 stop DR of the 5Ds. Maybe the Sony will be 56Mpix, even more DR and better colour and have a greater pixel level detail rendering this topic irrelevant but until that day I'm still very interested in the Canon.
Logged
trying to think of something meaningful........ Err?
Pages: [1] 2 3 4   Go Up