Dynamic Range: Canon 5D Mark IV vs Sony A7R III vs Nikon D900

[Guillermo Luijk]

Following the previous experiment in How would a Canon 5D Mark IV with 80D's photosites perform?, and with recent data in DxOMark for the new Sony A6300, I have extrapolated the performance of the three APS kings: Nikon D7200 (I didn't chose the D500 since it's more high ISO oriented), Canon 80D and Sony A6300, to what their equivalent FF counterparts would be with similar technology. This gives an idea of how the replacements of the Canon 5D Mark III, Nikon D810 and Sony A7R II could perform.

NOTES:

• I improved APS DxOMark's DR curves by log2(1,6)=0,68EV for Canon and log2(1,5)=0,58EV for Nikon and Sony (Data: Canon EOS 80D vs Sony a 6300 vs Nikon D7200).
  
• I didn't consider the possibility of ISO64 in the Nikon, just used the Nikon D7200 performance, which at photosite level and equal ISO beats the Nikon D810.
• For the Nikon I corrected the tail of DxOMark's curve which indicated more than one entire stop of DR reduction in the last two ISO's. This doesn't make too much sense and probably it's due to measure variance.

Quick conclusions:

• The future Nikon D900 would again be the winner, and the first camera on the market with more than 15 stops of DR according to DxOMark criteria (13 stops of photographic DR, HDR territory gets again stretched), and assuming ISO100 as the lowest.
• Paradoxically, Sony's sensor has the less isoless behaviour of the three, what helps it be the high ISO winner.
  
• At last Canon is getting closer to competition (as with the 80D), but with the new FF iterations from Nikon and Sony would still be the weakest in the group.

Regards

[BernardLanguillier]

Hi Guillermo,

Those are screen results, right?

What would this mean in print assuming that the resolution of the Nikon and Sony remain 50% higher than the Canon?

Thanks.

Cheers,
Bernard

[Guillermo Luijk]

Nope, it's all PRINT results.

Regarding the pixel count, I didn't make any correction since it's hard to model. This is exactly what a FF camera with the same photosites of the APS ones would perform (65Mpx for the Canon, and 55Mpx for the Sony and Nikon). If the pixel count is less than these (bigger photosites), the PRINT performance could be even higher, or at least never worse than what these figures indicate.

Regards

[BernardLanguillier]

Thanks.

Cheers,
Bernard

[shadowblade]

Interesting, but I'm not certain the next generation of high-resolution bodies (which will probably be the 5Ds2 rather than the 5D4, since the latter will be a general-purpose, more action-oriented camera) will be a simple scaling up of APS-C designs.

In particular, the non-linearity of the A6300 and the camera's focus on action photography leads me to believe the A6300 sensor sacrificed some low-ISO performance for better high-ISO performance - something that a detail-oriented A7r3/A9r is unlikely to do.

Besides, all the rumours point to the next high-resolution Sony sensor being in the 70-80MP range - in other words, denser than the A6300. BSI technology will mitigate the effects of denser circuitry on light collecting area, though, so SNR may not necessarily suffer as a result.

[dwswager]

Following the previous experiment in How would a Canon 5D Mark IV with 80D's photosites perform?, and with recent data in DxOMark for the new Sony A6300, I have extrapolated the performance of the three APS kings: Nikon D7200 (I didn't chose the D500 since it's more high ISO oriented), Canon 80D and Sony A6300, to what their equivalent FF counterparts would be with similar technology. This gives an idea of how the replacements of the Canon 5D Mark III, Nikon D810 and Sony A7R II could perform.

Pretty sure the D7200 uses an updated Toshiba sensor from the D7100 while the D900 will almost definitely utilize a Sony sensor.

[BernardLanguillier]

Pretty sure the D7200 uses an updated Toshiba sensor from the D7100 while the D900 will almost definitely utilize a Sony sensor.

Not what I heard. I believe that the D7200 uses a Sony part.

Cheers,
Bernard

[dwswager]

Not what I heard. I believe that the D7200 uses a Sony part.

Cheers,
Bernard

My source was NikonRumors.com List of all Nkon DSLRs and their Sensor Manufacturers

D40: 6 MP CCD Sony
D40x: 10 MP CCD Sony
D50: 6 MP CCD Sony
D60: 10 MP CCD Sony
D70: 6 MP CCD Sony
D80: 10 MP CCD Sony
D90: 12 MP CMOS Sony
D7000: 16 MP CMOS Sony
D7100: 24 MP CMOS Toshiba
D7200: 24 MP CMOS Toshiba (source)

D3000: 10 MP CCD Sony
D3100: 14 MP CMOS Nikon
D3200: 24 MP CMOS Nikon
D3300: 24 MP CMOS Sony

D5000: 12 MP CMOS Sony
D5100: 16 MP CMOS Sony
D5200: 24 MP CMOS Toshiba
D5300: 24 MP CMOS Sony (?)
D5500: 24 MP CMOS Sony

D100: 6 MP CCD Sony
D200: 10 MP CCD Sony
D300: 12 MP CMOS Sony

D600: 24 MP CMOS Sony
D610: 24 MP CMOS Sony
D750: 24 MP CMOS Sony

D700: 12 MP CMOS Nikon
Df: 16 MP CMOS Nikon

D800/D800E: 36 MP CMOS Sony
D810/D810A: 36 MP CMOS Sony

D1: 2.7 MP Sony
D1h: 2.7 MP Sony
D1x: 5.47 MP Sony
D2h: 4 MP LBCAST Nikon
D2x: 12 MP CMOS Sony (Nikon designed)
D3: 12 MP CMOS Nikon
D3s: 12 MP CMOS Nikon
D3x: 24 MP CMOS Sony
D4: 16 MP CMOS Nikon
D4s: 16 MP CMOS Nikon

[dwswager]

Just in case...

It is a fact that NikonRumors has a page that list the sensor manufacturer for the D7200 as Toshiba.  It is only my opinion that the D7200 has a Toshiba sensor. 

[BernardLanguillier]

Just in case...

It is a fact that NikonRumors has a page that list the sensor manufacturer for the D7200 as Toshiba.  It is only my opinion that the D7200 has a Toshiba sensor. 

Thanks, I don't remember clearly what my source is, but if I am not mistaken it is Thom Hogan.

Anyway, this is irrelevant now since the sensor division of toshiba was bought by... Sony.

https://www.engadget.com/2015/12/04/sony-toshiba-image-sensors/

Cheers,
Bernard

[Hulyss]

D300 & D300s: 12 MP CMOS RENESAS
D700: 12 MP CMOS RENESAS
Df: 16 MP CMOS RENESAS
D3: 12 MP CMOS RENESAS
D3s: 12 MP CMOS RENESAS
D3x: 24 MP CMOS Sony
D4: 16 MP CMOS RENESAS
D4s: 16 MP CMOS RENESAS
D500: 20 MP CMOS probably RENESAS (similar signal output as D300).
D5: 20 MP CMOS probably RENESAS (similar signal output as D3).

[Guillermo Luijk]

Interesting, but I'm not certain the next generation of high-resolution bodies (which will probably be the 5Ds2 rather than the 5D4, since the latter will be a general-purpose, more action-oriented camera) will be a simple scaling up of APS-C designs.

In particular, the non-linearity of the A6300 and the camera's focus on action photography leads me to believe the A6300 sensor sacrificed some low-ISO performance for better high-ISO performance - something that a detail-oriented A7r3/A9r is unlikely to do.

Besides, all the rumours point to the next high-resolution Sony sensor being in the 70-80MP range - in other words, denser than the A6300. BSI technology will mitigate the effects of denser circuitry on light collecting area, though, so SNR may not necessarily suffer as a result.

Canon

Yes, the 5D4 for sure won't be 65Mpx, so it's unlikely to be just a scaled version of the 80D (as the 5DS is a scaled version of the 7D II). But it would be a shame that Canon doesn't introduce its new sensor technology (80D) on it with larger sensels, so I have great expectations about the 5D4 performance. Coming soon...


Sony

I don't agree the 6300 sacrificed any significant low ISO DR like the D5 did. Its response, not being the usual found on Sonys, is still quite isoless. Also the 70-80 Mpx jump seems a too big gap from 42Mpx in my opinion, specially when the 50Mpx 5DS has become far from a bestseller. According to rumors the pro A9 should be on the market a year ago. Anyway I'm guessing about the A7R II replacement.


Nikon

The Toshiba 7200 had indeed an outstanding performance. Now with Sony owning the Toshiba sensor division there is a lot of uncertainty about how the future Nikon D900 can perform.


Regards




www.guillermoluijk.com

[dwswager]

Thanks, I don't remember clearly what my source is, but if I am not mistaken it is Thom Hogan.

Anyway, this is irrelevant now since the sensor division of toshiba was bought by... Sony.

https://www.engadget.com/2015/12/04/sony-toshiba-image-sensors/

Cheers,
Bernard

I don't believe it irrelevant as nobody knows what Sony will do with the Toshiba tech and any newer Sony sensor may or may not be derivative of the Toshiba work with the D7200.

While this speculation is fun, I usually like to wait for some concrete data.  Would I have predicted the D5 would perform as it does?  No.  While evidence suggest the D5 is a spectacular body, I'm still not sure what was gained outweighs what was spent to get it.  Maybe for some (Nikon's target users for the D5).  But I would have hoped for better at the low end from a 21MP next gen sensor than I already get from a current 36MP sensor.

[Guillermo Luijk]

While this speculation is fun, I usually like to wait for some concrete data.  Would I have predicted the D5 would perform as it does?  No.

It depends on how related the APS and FF sensors are. Do not underestimate predictions (blue was my 5DS prediction):



www.guillermoluijk.com

[dwswager]

It depends on how related the APS and FF sensors are. Do not underestimate predictions (blue was my 5DS prediction):



www.guillermoluijk.com

I'm not saying prediction is useless, only that it depends on knowing significant parameters going into it.  What would a predication of the D5 DR have looked like using the latest D7200 and D810 sensors?  While it is a relatively simple matter extrapolating the technical aspects of like items, it doesn't speak to business decisions or innovations that change the rules.

[Guillermo Luijk]

We happen to know those parameters: the 5D4, and future A7R III/D900 are not action cameras, that means low ISO response will prevail over high ISO. So extrapolating these APS sensors is a better starting point than the random choice.

Regards

www.guillermoluijk.com

[NancyP]

Forgive me for asking, but why do this sort of prediction/comparison?

[Guillermo Luijk]

Is that a phylosophical question? then the answer is why not?

[John Koerner]

But I would have hoped for better at the low end from a 21MP next gen sensor than I already get from a current 36MP sensor.

The D500 has better low-end Dynamic Range than either the new D5 or the new Canon 1Dx II ... and the D5 takes over the high-end DR right about where the D500 trails-off.

[John Koerner]

The Toshiba 7200 had indeed an outstanding performance. Now with Sony owning the Toshiba sensor division there is a lot of uncertainty about how the future Nikon D900 can perform.

I am not sure it makes a difference at all.

Nikon buys its D810 sensor from Sony, its D7200 sensor from Toshiba, so now (with the acquisition) Nikon will buy both from Sony.

What would be really interesting is if Sony purchased Nikon

Then Sony would have full control over the best bodies, best primes, and best sensors ...

If Nikon's problem isn't quality, but management, this would really put Canon behind the 8-ball trying to compete with Sony who would thereby have a total lock on every aspect of camera quality.