If 5d3 really is 22Mp, what will low light (ISO) performance likely be?

[birdman]

I hate to ask, as it has probably been beaten to death all over the world, but I need to know. My 5dII does fairly well, and based on my needs seldom pushes for TOO much of an increase. However, if they are using a sensor very close in MP size, it seems they are just following the approach by Nikon in the D3s revision.

I would expect images at 25,600 to rival 6400 on current 5d. Two stops is asking a lot, but I really think after 4 years of technology development this would be possible--even at 22MP.

If there is merely a clear one-stop advantage, I don't see a worthwhile reason to upgrade if AF is unimportant. Maybe low ISO/Shadow noise improvements could warrant a reason. As they always say, the proof is in the pudding. The more I look at the D800, which on paper appears to be wonderful at its price, the more I believe it will be noisier than a bunch of bad kids on Saturday morning. And that is a major reason NOT to jump ship!! At least for me, it is.

 

[Grum]

I seem to remember reading somewhere that the D800 high ISO performance would be broadly similar to the D700 - which to me would be disappointing given how old the D700 tech is now (though it's high ISO performance is still pretty decent). I personally much favour high ISO over MP.

What I'm not quite clear on though is if a high ISO D800 image resampled to 12MP would be significantly better than a D700 12MP one - I kind of assume the resampled one would be better but I get confused about these things easily.

 

[BRNexus6]

LOL, your expectations are way too high. I don't see more than 1 stop of ISO low-noise improvement. If the 5D Mark lll had the same 18 MP sensor as the 1DX, then you could have expected a big improvement over the Canon 5D Mark ll. I want Canon to release a full-frame 12 megapixel low-light beast HDSLR made for video.

 

[chito]

I think the best we can hope for is close to 1 stop, probably D3s performance... Looking at the quantum efficiency ratings:
http://www.sensorgen.info/

D3s - 57% (large pixels, less wiring, etc, etc).. note that the optimal QE would be 100% which is less than a 1 stop improvement over the D3s right now...

5D2 - 33%... hopefully they feel the pressure, the NEX-7 is 47% ! with super small pixels and APS-C size! I assume the 5D3 will surpass this, and the S100 - 52% also.

 

[Positron]

Looking at generaional gaps between other cameras my guess would be roughly a stop of improvement in terms of real world use, assuming, as you stated, that the resolution stays similar to the 5D2's. Just speculation though. Hopefully it won't take too long from the announcement for high ISO samples to be available for perusal.

 

[FredBGG]

If the 1D X is anything to go by and the 5D III has similar improvements it should be interesting.

IF you look at the blacks in the sample files from the Canon 1D X they are much cleaner with more very dark and dark distinction. This is an area where 35mm DSLR cameras have been lagging way behind MF CCD sensors.

I heard about a year ago form someone at Canon that this generation of camera will have a significant increase in dynamic range.

I also found that the 1d X files also have remarkably natural skin tones and hold up better to modification in post that the samples of the Nikon d800.

It will be interesting to see the two different paths taken.

Canon IQ through tonal range/dynamic range and Nikon through MP count.

 

[mrmarks]

It will be interesting to see the two different paths taken.

Canon IQ through tonal range/dynamic range and Nikon through MP count.

Yes indeed.

 

[jrista]

Just to point out, no one knows the dynamic range of any of the new Canon cameras yet. So far, Nikon, via Sony Exmor sensors, is outperforming Canon sensors by around two full stops on the DR front, and offers FAR lower noise at low ISO levels than any Canon sensor (shadow recoverability with a Sony Exmor is unbelievable, where as Canon sensors exhibit pronounced and detail-obliterating fixed-pattern or banding noise.)

I think the assumption has so far been that because Canon increased their maximum ISO, that DR would also increase. I don't think such an assumption is safe. It could very well be that Canon's new sensors still lose around 2 stops of DR at low ISO (where DR is highest) because of read noise introduced by their electronics (particularly ADC.) If thats the case, Nikon will still be winning the DR war thanks to Sony Exmor technology.

 

[Radiating]

I hate to ask, as it has probably been beaten to death all over the world, but I need to know. My 5dII does fairly well, and based on my needs seldom pushes for TOO much of an increase. However, if they are using a sensor very close in MP size, it seems they are just following the approach by Nikon in the D3s revision.

I would expect images at 25,600 to rival 6400 on current 5d. Two stops is asking a lot, but I really think after 4 years of technology development this would be possible--even at 22MP.

If there is merely a clear one-stop advantage, I don't see a worthwhile reason to upgrade if AF is unimportant. Maybe low ISO/Shadow noise improvements could warrant a reason. As they always say, the proof is in the pudding. The more I look at the D800, which on paper appears to be wonderful at its price, the more I believe it will be noisier than a bunch of bad kids on Saturday morning. And that is a major reason NOT to jump ship!! At least for me, it is.

I think your technical understanding of photography is off the mark, to say the least.

Noise improvements happen from two areas, noise reduction and sensor technology. Seeing as you can just apply the latest noise reduction to any raw file the advantage doesn't go to any camera in this regard.

To give you the simple answer no the camera will not deliver 2 more stops of noise.

In fact in 9 years since the earliest days of DSLR's Canon has only increased sensor sensitivity by a QUARTER stop. To give you an idea of how off the charts what you're asking is, not only would it be 16 times the improvement Canon is used to delivering, it is actually against the laws of physics to improve the current 5D II 2 stops at the sensor level. The current 5D gathers 33% of the light that hits it. 2 stops would mean that the camera gathers more photons than actually exist. The camera would have to create more energy than it takes in which is along the realm of perpetual motion machines. Perhaps in the 5D Mark 1,003 but not the 5D mark 3.

It's worth mentioning though that Canon has let their sensors drift behind the field in terms of noise. So the 5D III is poised to have a massive 0.25 stop to 1 stop increase in noise handling, most likely a 0.33 stops or 0.5 stops to catch up to current standards. This sort of leap is considered the equivalent of going from air travel to space travel so it's a big deal and will be the biggest improvement that Canon has ever done in noise from one generation to another largely thanks to pressure from Nikon which has made extreme leaps in noise handling. Canon has already caught up the rest of their line to Nikon which has been done in smaller jumps thanks to a shorter life cycle so a leap of this magnitude is very likely.

So yes while the 5D II > 5D III noise won't meet your standards for being worth it, it will likely be the biggest improvement in a single generation that ever has or will happen.

As far as JPG's go you can expect 2 stops though as Canon implements revolutionary new noise reduction algorithms which have been available in only photoshop plug-ins so far.

 

[Legio]

I think your technical understanding of photography is off the mark, to say the least.

Noise improvements happen from two areas, noise reduction and sensor technology. Seeing as you can just apply the latest noise reduction to any raw file the advantage doesn't go to any camera in this regard.

To give you the simple answer no the camera will not deliver 2 more stops of noise.

In fact in 9 years since the earliest days of DSLR's Canon has only increased sensor sensitivity by a QUARTER stop. To give you an idea of how off the charts what you're asking is, not only would it be 16 times the improvement Canon is used to delivering, it is actually against the laws of physics to improve the current 5D II 2 stops at the sensor level. The current 5D gathers 33% of the light that hits it. 2 stops would mean that the camera gathers more photons than actually exist. The camera would have to create more energy than it takes in which is along the realm of perpetual motion machines. Perhaps in the 5D Mark 1,003 but not the 5D mark 3.

It's worth mentioning though that Canon has let their sensors drift behind the field in terms of noise. So the 5D III is poised to have a massive 0.25 stop to 1 stop increase in noise handling, most likely a 0.33 stops or 0.5 stops to catch up to current standards. This sort of leap is considered the equivalent of going from air travel to space travel so it's a big deal and will be the biggest improvement that Canon has ever done in noise from one generation to another largely thanks to pressure from Nikon which has made extreme leaps in noise handling. Canon has already caught up the rest of their line to Nikon which has been done in smaller jumps thanks to a shorter life cycle so a leap of this magnitude is very likely.

So yes while the 5D II > 5D III noise won't meet your standards for being worth it, it will likely be the biggest improvement in a single generation that ever has or will happen.

As far as JPG's go you can expect 2 stops though as Canon implements revolutionary new noise reduction algorithms which have been available in only photoshop plug-ins so far.

Ill agree with you, however since my 5DM2 takes to noisy pictures at 12800 and above I would consider it as I got more iso in the new one if the quality would be the same as one lower iso in the old one, even if that could be tributed to better noise handling in the sensor reading (RAW) (SNR handling in sensor and noise reduction in post should be diffrent teqniques as I see it).

So next one probobly would have some aditional steps vs the "old" mk2 but as we all know the upper steps are not so good looking. But I wouldn't be suprised if we saw a H2 of 102400 if it would be usefull is another topic.

To be able to lift the shadows more than the mk2 would be good, but even if the mk2 gets critisized for this its relly is a wonderful camera. More DR is not a bad thing so hopefully we will see just that in the new mk3