Nikon D5 Sensor Score from DXOMark

[Mikehit]

But aren't there architectural changes which could alleviate some of the load on the digic unit, for example digitizing the signal before it even reaches the processor?

I thought the processor did the digitising so how can it be digitised before the processor?

 

[bwud]

But aren't there architectural changes which could alleviate some of the load on the digic unit, for example digitizing the signal before it even reaches the processor?

I thought the processor did the digitising so how can it be digitised before the processor?

Current canon architecture has that function in the DIGIC processor, but there's no reason they couldn't move them earlier in the path (like on the sensor unit).

 

[Don Haines]

But aren't there architectural changes which could alleviate some of the load on the digic unit, for example digitizing the signal before it even reaches the processor?

I thought the processor did the digitising so how can it be digitised before the processor?

Current canon architecture has that function in the DIGIC processor, but there's no reason they couldn't move them earlier in the path (like on the sensor unit).

Digitising is the function of the A/D converter. Depending on how the system is designed, this can be done as a stand-alone chip(s), on the processor die, or on the sensor die.... Each has advantages and disadvantages.

As far as processing power goes, the need for speed also depends on how the system is designed. You can process the data as it is generated or you can dump it straight from the A/D into memory, to be processed later... and once again, each has advantages and disadvantages.

 

[bwud]

But aren't there architectural changes which could alleviate some of the load on the digic unit, for example digitizing the signal before it even reaches the processor?

I thought the processor did the digitising so how can it be digitised before the processor?

Current canon architecture has that function in the DIGIC processor, but there's no reason they couldn't move them earlier in the path (like on the sensor unit).

Digitising is the function of the A/D converter. Depending on how the system is designed, this can be done as a stand-alone chip(s), on the processor die, or on the sensor die.... Each has advantages and disadvantages.

As far as processing power goes, the need for speed also depends on how the system is designed. You can process the data as it is generated or you can dump it straight from the A/D into memory, to be processed later... and once again, each has advantages and disadvantages.

Sure, I was just giving one example of a function which could be offloaded from the processor. ADCs need not be particularly "smart," and parallelizing the function (for example one per column of pixels or, at the extreme, one per pixel) could free up space on the processor dye, remove a bottleneck, etc.

The initial post I replied to talked in bits/second, but part of that figure includes creating bits. Throughput could theoretically increase if the processor is fed digital data from the onset.

 

[Don Haines]

But aren't there architectural changes which could alleviate some of the load on the digic unit, for example digitizing the signal before it even reaches the processor?

I thought the processor did the digitising so how can it be digitised before the processor?

Current canon architecture has that function in the DIGIC processor, but there's no reason they couldn't move them earlier in the path (like on the sensor unit).

Digitising is the function of the A/D converter. Depending on how the system is designed, this can be done as a stand-alone chip(s), on the processor die, or on the sensor die.... Each has advantages and disadvantages.

As far as processing power goes, the need for speed also depends on how the system is designed. You can process the data as it is generated or you can dump it straight from the A/D into memory, to be processed later... and once again, each has advantages and disadvantages.

Sure, I was just giving one example of a function which could be offloaded from the processor. ADCs need not be particularly "smart," and parallelizing the function (for example one per column of pixels or, at the extreme, one per pixel) could free up space on the processor dye, remove a bottleneck, etc.

The initial post I replied to talked in bits/second, but part of that figure includes creating bits. Throughput could theoretically increase if the processor is fed digital data from the onset.

Yeah, there are so many ways to do things.... and it all depends on your final goal....

 

[Sporgon]

How many threads have there been on CR about how DR doesn't matter?

None.

Of course DR matters. The asinine part is arguing that the advantage of 13 over 11.5 offers much improved 'IQ' in the majority of photography, and the sensors are much more advanced, so Canon are behind etc etc. I remember watching an interview with the Canon CEO where the two-bit journalist asked him what Canon were going to do about their 'poor IQ'. The poor guy nearly fell off his chair in surprise !

The Nikon D5 sensor is one of the funniest things that happened in the photographic gear world for some time - watching the web journalists who have hung their hats on 'extra DR is king' and ' off sensor ADC is so medieval' wriggle and squirm their way around this most modern sensor and its 'old' characteristics.

 

[bgateb]

#dxo #fairandbalanced

- A

i just had to LOL at this.

 

[Loswr]

The "mind set" of CR is that more DR is not important, therefore it doesn't matter if the camera has 11, 12, 13 or 14 stops of DR.

Nice to see that your reading comprehension skills are as astute as ever. :

 

[Woody]

The "mind set" of CR is that more DR is not important, therefore it doesn't matter if the camera has 11, 12, 13 or 14 stops of DR.

Your sarcastic line of attack cuts both ways.

If 11/12/13/14 stops of dynamic range does not matter, then it also means that all the DXOMark scores are rubbish and all the arguments about the superiority of Sony Exmor sensor hold no water.

You dare to stick your neck out for this?

 

[Orangutan]

How many threads have there been on CR about how DR doesn't matter?

Sorry, I have to jump on this too. The reason you (and others) receive harsh replies for dwelling on DR is not that it's unimportant, but that you seem to assert that it's the only characteristic that's important. It's as if you claimed that fuel efficiency were the only characteristic that's important in a vehicle. Everyone agrees that fuel efficiency is important, but if you need to carry a lot of cargo you certainly wouldn't choose a fuel-efficient scooter.

Of course, you could now ask whether we would be interested in a vehicle that carries a lot of cargo and also has the fuel efficiency of a scooter. Of course! But that doesn't exist, so we're left to choose from among the vehicles that do exist.

Low-ISO DR is one important characteristic, but not the only one.

 

[Keith_Reeder]

But DPR also sings a very pretty song about the AF in the D5/D500 so maybe that makes up for it when money is on the line...

Ah - so when it's a Nikon, you're OK with the idea that "superior" sensor performance isn't necessarily the whole story, and that for some photographers, other things might matter more.

Interesting...

 

[Keith_Reeder]

Aye, except clearly the D5 has better DR than the 80D at ISO 50 :

Which is, of course, more, and more, and more marginally important to the majority...

 

[Keith_Reeder]

How many threads have there been on CR about how DR doesn't matter?

And now, in your rush to defend the D5, you suddenly accept the premise...

 

[bdunbar79]

But DPR also sings a very pretty song about the AF in the D5/D500 so maybe that makes up for it when money is on the line...

Ah - so when it's a Nikon, you're OK with the idea that "superior" sensor performance isn't necessarily the whole story, and that for some photographers, other things might matter more.

Interesting...

Well yeah when the sensor sucks according to your one metric, you gotta come up with something else.

 

[Keith_Reeder]

But when we've talked about this sort of thing in the context of Canons, we've been making excuses to defend Canon's incompetence.

Now that a Noink has taken a step back, low ISO DR-wise, it's suddenly OK for other parameters to be more important.

The word "hypocrite" doesn't even come close to describing Dilbert, does it?

 

[bdunbar79]

But when we've talked about this sort of thing in the context of Canons, we've been making excuses to defend Canon's incompetence.

Now that a Noink has taken a step back, low ISO DR-wise, it's suddenly OK for other parameters to be more important.

The word "hypocrite" doesn't even come close to describing Dilbert, does it?

That's only one of many words I can think of.

 

[3kramd5]

Well if you happily own a Canon camera then you wouldn't care about that extra half stop of DR, would you?

??

Care to share your reasoning?

You must be reading a different canonrumors website than I am.

Is that a no, you won't share it?
...

How many threads have there been on CR about how DR doesn't matter?

The "mind set" of CR is that more DR is not important, therefore it doesn't matter if the camera has 11, 12, 13 or 14 stops of DR.

Therefore if you were a Canon person, the drop in 1 stop of DR from the 1DXII to (say) the 1DXIII would be unimportant and not worthy of consideration as to whether the camera was a worthy successor.

If you don't believe me, try start a thread on CR about how DR is vital and important and how the difference between 12 and 13 stops of DR is like life and death. See how many people shout you down as being "incompetent" or "using the wrong technique", etc. Less DR is perfectly acceptable to a Canon shooter, just ask just about anyone on CR.

What most people aren't interested in is comparing between brands, which is the inexorable context of almost all DR threads. And you set up a self-selecting sample by invoking people who are happy with their canon cameras. They don't care about datasheet parameters of the primary I/O device in a Nikon camera any more than a happy Nikon owner cares about the MTBF of a canon shutter. Why should they if they're happy with what they have? People who aren't happy with what they have are far more likely to pay attention to alternatives.

If you were to start a thread: would you like your canon camera to be better in any category, including DR, the answer would be a resounding yes, even if a given improvement would make no practical difference to every participant.

And of course if I set up a thread suggesting that "the difference between 12 and 13 stops of DR is like life and death," I would be rightly ridiculed and labeled a troll for using such hyperbole.

 

[zim]

Aye, except clearly the D5 has better DR than the 80D at ISO 50 :

Which is, of course, more, and more, and more marginally important to the majority...

Well yes for the majority of sonik measurebaitors that seem to frequent CR.

DXO will have to find some way of defining that D5 sensor as superior my guess is that they will take the total ISO range as the advantage even when it's clearly not better than the 810 at 64 (which is actually looking like a rather special brew in fairness) and less than the 80D at ISO100

 

[IglooEater]

A worthy successor...with less low ISO DR than the APS-C sensor in Canon's 80D.

By whose measurement? Yours? Canon's? Where is the 80D review in DxO? Oh yeah, they don't have one.

Bill Claff has measured so. Check out the chart generated from his website.

Perhaps- but I'd rather the D5's performance from these charts. I don't need more dr at ISO100 (sure it could be nice, but I don't need it). I want more DR at ISO 6400-that's where I'm limited.

Eh? Not from the low ISO DR brigade, are you?

You need to understand what you are comparing - the Nikon flagship FF D5 vs. the lowly Canon APS-C 80D. Apples Vs. Oranges. The two cameras are hardly competition to each other. A more meaningful analysis would be between D5 and 1DX II.

Nope, never have been on that brigade- check my past posts. I've agreed it would be nice, but I have higher priorities.

I agree it's a rather silly comparison- I don't think anyone would take the 80D over the D5 were money not an object.

 

[Jack Douglas]

3kramd5, you've said it as clearly as anyone could. Never the less there are some pretty dense brains that exhibit in CR threads. Canon has me hook, line and sinker and the other brands are for other folk. Still, it's fun to look at the others brands and know what's going on.

Jack