Noise Pattern Evaluation: D800/E, D600, A99

Started Oct 5, 2012 | Discussions
JohnBee
Veteran MemberPosts: 3,966
Like?
Noise Pattern Evaluation: D800/E, D600, A99
Oct 5, 2012

I managed to snag this little demonstration showing a noise pattern comparison between the A99, Nikon D800 and 800E as well as the D600. Samples were all taken from ISO6400 NR0 in RAW.

Make of it what you will, but I think the A99 holds its own quite well against the competition.

Nikon D600 Nikon D800 Sony SLT-A99
If you believe there are incorrect tags, please send us this post using our feedback form.
danny006
Contributing MemberPosts: 817Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 5, 2012

This shows my initial thoughts. Better noise performance than D800 and about the same as the D600.

Thx for sharing.

 danny006's gear list:danny006's gear list
Sony RX100 Sony SLT-A77 Tamron SP AF 200-500mm F/5-6.3 Di LD (IF) Sigma 105mm F2.8 EX DG Macro
Reply   Reply with quote   Complain
Nordstjernen
Veteran MemberPosts: 6,194Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 5, 2012

If this is what to expect, I should preorder the A99 today!

There is a lot to like with this new camera. Better & more reliable AF than the A77 (excellent when configurated for the actual situation), better vignetting control than the A900, more detail due to the lower sensor/new AA filter design, lightweight and not that bulky. Even if the noise level isn't significantly better than the A900 I am sold!

A note on the sample: Sony 50 mm f:1.4? Hot spot? Not even lit background?

 Nordstjernen's gear list:Nordstjernen's gear list
Sony SLT-A99 Sony Alpha 7
Reply   Reply with quote   Complain
mick232
Contributing MemberPosts: 753Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 5, 2012

The A99 seems to have lowest noise of these 4 cameras. I think it beats the D600 by a very small but noticable margin.

 mick232's gear list:mick232's gear list
Sony Alpha NEX-5 Sony SLT-A99 Tamron SP AF 70-200mm F/2.8 Di LD (IF) MACRO Sony 75-300mm F4.5-5.6 Sony 85mm F2.8 SAM +21 more
Reply   Reply with quote   Complain
Emopunk
Regular MemberPosts: 145Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to mick232, Oct 5, 2012

I don't know how scientific your test might be, but for what I see, A99 clearly has the best output of this bunch. I look forward to owning one very soon..!!

 Emopunk's gear list:Emopunk's gear list
Sony Alpha DSLR-A200 Sony SLT-A99 Sigma 70-200mm F2.8 EX DG Macro HSM II Sony 20mm F2.8 Tamron AF 28-75mm F/2.8 XR Di LD Aspherical (IF) +1 more
Reply   Reply with quote   Complain
sandy b
Veteran MemberPosts: 7,201Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 5, 2012

Whats with the non even lighting across the A99 shot? Looks a little weird

 sandy b's gear list:sandy b's gear list
Nikon Coolpix P330 Nikon D7000 Nikon D600 Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR Nikon AF-S Teleconverter TC-14E II +3 more
Reply   Reply with quote   Complain
lim yau tong
Senior MemberPosts: 2,859Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 5, 2012

Can you also include the Nikon D4 and D3s.

 lim yau tong's gear list:lim yau tong's gear list
Sony Alpha DSLR-A350 Sony Alpha DSLR-A700 Sony Alpha DSLR-A550 Sony SLT-A55 Sony SLT-A77 +25 more
Reply   Reply with quote   Complain
Michel J
Veteran MemberPosts: 4,009Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 6, 2012

johnbee wrote:

[...]  Samples were all taken from ISO6400 NR0 in RAW. [...]

Make of it what you will, but I think the A99 holds its own quite well against the competition.

Thanks for sharing this, John.
Since we know that the lighting conditions affect noise level a lot, this test proves nothing, until we know more about the test protocol.
If now the 4 cameras had exactly the same protocol (what we guess...) it would be very encouraging!

Kind Regards,

-- hide signature --

Michel J

 Michel J's gear list:Michel J's gear list
Sony SLT-A77 Sony SLT-A37 Konica Minolta Maxxum 7D Sony DT 18-55mm F3.5-5.6 SAM +5 more
Reply   Reply with quote   Complain
mick232
Contributing MemberPosts: 753Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to Emopunk, Oct 6, 2012

Emopunk wrote:

I don't know how scientific your test might be, but for what I see, A99 clearly has the best output of this bunch. I look forward to owning one very soon..!!

Judging the visual appearance, I would agree. However, I actually just wrote a little program that calculates the standard deviation for the R,G,B channels. If the ideal image here would be uniform grey, calculating for each pixel how much it is off from grey and summing that up (squares actually) should allow us to estimate the noise, right?

The result: the A99 came out worst.

But I don't think it is due to the noise. If you look closely, the A99 image seems to have some kind of vignetting effect. Slightly brighter in the middle than on the borders. Unfortunately, this probably renders the test useless at least for the A99.

In case you are interested, here's the result (lower = better):

d600.png: 6.7744590090235 6.010670902528859 8.070111508983086
d800.png: 8.7797877073513 7.287356209770788 10.947690877804124
d800e.png: 8.398218004457668 7.149128388812794 10.186271161400114
a99.png: 10.859560376781483 9.218291222220033 10.923016082630252

 mick232's gear list:mick232's gear list
Sony Alpha NEX-5 Sony SLT-A99 Tamron SP AF 70-200mm F/2.8 Di LD (IF) MACRO Sony 75-300mm F4.5-5.6 Sony 85mm F2.8 SAM +21 more
Reply   Reply with quote   Complain
mick232
Contributing MemberPosts: 753Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to mick232, Oct 6, 2012

If I only take the center part of the A99 image, the result is:

a99_2.png: 7.872076207344561 6.568750458448936 9.543053084661068

Worse than D600, but better than D800(e).

 mick232's gear list:mick232's gear list
Sony Alpha NEX-5 Sony SLT-A99 Tamron SP AF 70-200mm F/2.8 Di LD (IF) MACRO Sony 75-300mm F4.5-5.6 Sony 85mm F2.8 SAM +21 more
Reply   Reply with quote   Complain
Michel J
Veteran MemberPosts: 4,009Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to mick232, Oct 6, 2012

mick232 wrote:

Emopunk wrote:

I don't know how scientific your test might be, but for what I see, A99 clearly has the best output of this bunch. I look forward to owning one very soon..!!

Judging the visual appearance, I would agree. However, I actually just wrote a little program that calculates the standard deviation for the R,G,B channels. If the ideal image here would be uniform grey, calculating for each pixel how much it is off from grey and summing that up (squares actually) should allow us to estimate the noise, right? [...] If I only take the center part of the A99 image, the result is [...] Worse than D600, but better than D800(e)

Of course!

Since we know that cheating to reduce noise does not help to get the best color rendition, you demonstrated the opposite!

Compared to Alpha DSLR-T's, Nikon has never been as good in these colour features:
It's probaly due to the fact of the Sony heritage:
http://sensing.konicaminolta.us/products/cl-500-illuminance-spectrophotometer/

Kind Regards,

-- hide signature --

Michel J

 Michel J's gear list:Michel J's gear list
Sony SLT-A77 Sony SLT-A37 Konica Minolta Maxxum 7D Sony DT 18-55mm F3.5-5.6 SAM +5 more
Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,721Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to mick232, Oct 6, 2012

mick232 wrote: I actually just wrote a little program that calculates the standard deviation for the R,G,B channels. If the ideal image here would be uniform grey, calculating for each pixel how much it is off from grey and summing that up (squares actually) should allow us to estimate the noise, right?

I assume your program works on the raw data (as does RawDigger), otherwise all bets are off.  In that case the mean is supposedly the 'uniform grey' level and you are calculating the patches' standard deviation.  Make sure 'uniform grey' values are the same for all cameras you wish to compare: if not it's best to compare SNRs.

Cheers,

Jack

Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,721Gear list
Like?
Re: Noise Pattern Evaluation: D800/E downsampled to 24MP?
In reply to JohnBee, Oct 6, 2012

Were the D800/E downsampled to 24MP and the data captured by the same outfit (which one?) with the same parameters and methodology?  If not we are really looking at oranges and smugly think that their skin isn't as smooth as our apple's

Reply   Reply with quote   Complain
Michel J
Veteran MemberPosts: 4,009Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to Jack Hogan, Oct 6, 2012

Jack Hogan wrote:

mick232 wrote: I actually just wrote a little program that calculates the standard deviation for the R,G,B channels. If the ideal image here would be uniform grey, calculating for each pixel how much it is off from grey and summing that up (squares actually) should allow us to estimate the noise, right?

I assume your program works on the raw data (as does RawDigger), otherwise all bets are off.  In that case the mean is supposedly the 'uniform grey' level and you are calculating the patches' standard deviation.  Make sure 'uniform grey' values are the same for all cameras you wish to compare: if not it's best to compare SNRs.

Cheers,

Jack


Good idea. But what Jack suggest dosen't work, since we know that the three channels (RGB) are not aligned (in-camera)

Here before demosaicing:

Here what you see on the LCD:

It seems to be aligned but it's not...!

Btw, the level of noise is not the same for every channels. Here:

(Red VS Green (dcraw -h processing))

Every factory had his own soup (due to a compromise between daylight and thungsten or so...)

So, most of the time the luminance channel is blown up. So we can't deduct something right from that.

See the thread of Iliah Borg for more.

Kind Regards,

-- hide signature --

Michel J

 Michel J's gear list:Michel J's gear list
Sony SLT-A77 Sony SLT-A37 Konica Minolta Maxxum 7D Sony DT 18-55mm F3.5-5.6 SAM +5 more
Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,721Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to Michel J, Oct 6, 2012

Michel J wrote:  But what Jack suggest dosen't work, since we know that the three channels (RGB) are not aligned (in-camera)...

And?

So, most of the time the luminance channel is blown up. So we can't deduct something right from that.

Sure we can, perhaps not from these patches (see my other post), but in general mean and standard deviation measurements of the raw values is how one determines SNR and DR performance.  Raw values only came in R*, B* and G* flavors, and studio samples are designed not to blow them where it counts (the patches above are not blown). Assuming all other things are equal (including the light source - see my other post) measuring and comparing data from those is the first step in fairly comparing the noise performance of two cameras.

Cheers,
Jack

Reply   Reply with quote   Complain
Michel J
Veteran MemberPosts: 4,009Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to Jack Hogan, Oct 6, 2012

Jack Hogan wrote:

Michel J wrote:  But what Jack suggest dosen't work, since we know that the three channels (RGB) are not aligned (in-camera)...

And?

So, most of the time the luminance channel is blown up. So we can't deduct something right from that.

Sure we can, perhaps not from these patches (see my other post), but in general mean and standard deviation measurements of the raw values is how one determines SNR and DR performance.  Raw values only came in R*, B* and G* flavors, and studio samples are designed not to blow them where it counts (the patches above are not blown). Assuming all other things are equal (including the light source - see my other post) measuring and comparing data from those is the first step in fairly comparing the noise performance of two cameras.

Cheers,
Jack

Ok you can! If, and only if you:

  • know the real alignment of the three channels (RGB, before demosaicing, of course) for each camera you want to test;
  • use a magenta filter, such as the CC30M  (these suggested by Iliah Borg for daylight):
  • apply the required exposure compensation per channel;
  • correct the overall exposure according to (in the case of ETTR);
  • repeat the same with every camera of the test;

Then, you can get what you are looking for (a near to perfect grey pattern). I seriously doubt the O.P. nor mick232 have done that?

Just my two cents.

Kind Regards,

-- hide signature --

Michel J

 Michel J's gear list:Michel J's gear list
Sony SLT-A77 Sony SLT-A37 Konica Minolta Maxxum 7D Sony DT 18-55mm F3.5-5.6 SAM +5 more
Reply   Reply with quote   Complain
JohnBee
Veteran MemberPosts: 3,966
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 7, 2012

Here's another pixel peeping tidbit that I found interesting:

ISO6400, RAW, ACR, RGB

ISO6400, RAW, ACR, Blue Channel

In this detailed examination, it would appear as though the A99 sample is showing slightly more chroma noise in the blue channel than with the D600 sample toe to toe. Though the differences seem quite small to the naked eye.

That being said, I'd say the A99 will likely do very well for itself.

Reply   Reply with quote   Complain
Michel J
Veteran MemberPosts: 4,009Gear list
Like?
Re: Noise Pattern Evaluation: D800/E, D600, A99
In reply to JohnBee, Oct 7, 2012

johnbee wrote:

Here's another pixel peeping tidbit that I found interesting:

ISO6400, RAW, ACR, RGB

ISO6400, RAW, ACR, Blue Channel

In this detailed examination, it would appear as though the A99 sample is showing slightly more chroma noise in the blue channel than with the D600 sample toe to toe. Though the differences seem quite small to the naked eye.

That being said, I'd say the A99 will likely do very well for itself.

Thanks for sharing again John.

Well I don't the cause of this, but one of them is the fact that wavelength of the blue chanel have less relative power (here the D65)

Source: wikimedia.org/wikipedia/commons/d/d0/SPD_D65.png

So... As usual, by ETTR and filtering the luminance channel, that can increase the relative power of the Blue channel (and according to the filter, also the Red if so...)

It would be interesting to apply the Iliah Borg workflow at this point, because we have really no idea of what Nikon do with this blue channel and if the colour channels are already at the same level  in-camera to compensate something (or not) before demosaicing. imho...

Kind regards,

-- hide signature --

Michel J

 Michel J's gear list:Michel J's gear list
Sony SLT-A77 Sony SLT-A37 Konica Minolta Maxxum 7D Sony DT 18-55mm F3.5-5.6 SAM +5 more
Reply   Reply with quote   Complain
Jack Hogan
Senior MemberPosts: 3,721Gear list
Like?
Re: Noise Pattern Evaluation D600, A99
In reply to Michel J, Oct 7, 2012

Michel J wrote:  we have really no idea of what Nikon do with this blue channel and if the colour channels are already at the same level  in-camera to compensate something (or not) before demosaicing. imho...

Clearly the blue channel, having a lower absolute QE, would exacerbate the weaker camera's performance.  But without disturbing Mr. Borg, as long as the light source, testing methodology and photographic parameters are the same we really do get an idea of what the relative performance of the two cameras would be like - in this particular situation.  It would be useful to know what the situation is, though (John?).

Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads