[hilbert5]

It would have been properly exposed for ISO 6400, but the picture was taken at ISO 100, 6 stops underexposed. It was then over-exposed 8 stops in Lightroom to correct for the under-exposure (and then some). The 8 stops were achieved by combining a gradient filter (+4 stops) with the regular exposure slider (+4 stops also).

Maybe I'm asking a stupid question, but while I understand your 6 stops underexposure, I am not quite certain I can follow your calculation of 6 stops + 4 stops + 4 stops = 14 stops: Are we all sure that a slider at +4 in Lightroom actually is +4 stops? Has Adobe this ever published? Are we even sure the sliders are actually following the same scaling (namely that plus every stop means doubling the amount of light)? Maybe I'm missing something.

[Eruditass]

The dynamic range of a single pixel for 1:1 SNR is 13.61 EV according to them.

The quantization noise with a 14 Bit ADC is 1:1 (or better -- would have to think about) for the 14th bit (at -14.0 EV), leaving out color and sensor response calibration (a DxO mystery). So, DxO measured DR is below the 14 Bit "threshold".

However, both 1:1 SNR (0 dB) and spatial Nyquist frequency are arbitrary points of reference. DxO uses the 0 dB point indeed but the spatial frequency of an 8 MP image. DR increases when you downscale an image as does the image's effective bit depth. You get signal in the 15th and 16th bit by downscaling. Every reduction to 50% adds one bit to the true signal. So, Photoshop with its 15 Bit TIFF engine starts to destroy information if you downscale an ISO 80 K-5 image to 2.3 MPixel or below within Photoshop.

Thanks, that makes sense. Can you clarify DxO's reason for 12 bit files having over 12 stops per pixel? Instead of dividing by the lowest value, they divide by the noise standard deviation? But how do they measure less than 1 bit?

Question about Dxomark MEASUREMENTS of NIKON D90

[Olivier]

Maybe I'm asking a stupid question, but while I understand your 6 stops underexposure, I am not quite certain I can follow your calculation of 6 stops + 4 stops + 4 stops = 14 stops: Are we all sure that a slider at +4 in Lightroom actually is +4 stops? Has Adobe this ever published? Are we even sure the sliders are actually following the same scaling (namely that plus every stop means doubling the amount of light)? Maybe I'm missing something.

The number 14 comes from the precision of the ADC (14 bits per pixel) and is not related to the exposure.

The exposure is -6 +4 +4 = +2 over-exposure. But the exact amount does not matter. All I tried to do was to take a very under-exposed image and peek into the deep shadows using Lightroom.

[falconeye]

Thanks, that makes sense. Can you clarify DxO's reason for 12 bit files having over 12 stops per pixel? Instead of dividing by the lowest value, they divide by the noise standard deviation? But how do they measure less than 1 bit?

Question about Dxomark MEASUREMENTS of NIKON D90

They measure less than 1 bit pixel noise with the D90 indeed. They write:

"The maximum theoretical dynamic range from a 12bit output camera is actually 13.79."

This is what I meant by "The quantization noise [...] is 1:1 (or better -- would have to think about)" in my earlier response. I would have to do some math to compute the correct number for quantization noise. I'm too lazy to do it and it probably can be looked up in the literature.

However, if you actually go beyond the bit depth, you start to get banding artefacts which is an avoidable thing.



EDIT:
I found this which I copy paste w/o citing because I haven't read the text!!

The dynamic range of an ideal A-D converter is set by the ratio of the maximum signal to the quantization noise standard deviation. For cameras with linear A-D converters the ideal quantization noise is uniformly distributed over the range of -.5 to +.5 of the step size q, so the quantization noise standard deviation is q / sqrt(12). The maximum pixel dynamic range for N bits is therefore N + log2(12) / 2 which for 12 bits is 13.79.

---

Last edited by falconeye; 11-05-2010 at 02:47 PM.

[falconeye]

Are we all sure that a slider at +4 in Lightroom actually is +4 stops? Has Adobe this ever published? Are we even sure the sliders are actually following the same scaling (namely that plus every stop means doubling the amount of light)? Maybe I'm missing something.

No, that's just a common assumption worth a validation. Go for it!

Best would be to compare with pushing the exposure in Photoshop.

One should switch PS into 32 Bit mode (which applies a color model w/o gamma correction -- sRGB values are NOT linear!!).

The exposure slider in PS goes up to +20EV and avoids the LR layer trick which isn't verified to work either...

One may control the PS slider that it does indeed double the rgb values in 32 Bit mode (it does so!) for a +1EV push.

Reverting back to 8 or 16 bit changes rgb back to sRGB (values change) but the image stays the same. Then it can be read into LR and compared side by side.

[Class A]

No, that's just a common assumption worth a validation.

Guillermo Luijk has reverse-engineered the effects of the "basic panel" sliders in LR. These graphs (albeit indirectly since they are using the sRGB space) show that the EV values in LR are accurate and that +50 on the brightness slider corresponds to +1EV.

[falconeye]

Guillermo Luijk has reverse-engineered the effects of the "basic panel" sliders in LR. These graphs (albeit indirectly since they are using the sRGB space) show that the EV values in LR are accurate and that +50 on the brightness slider corresponds to +1EV.

Thanks for the great link.
Its a pitty they couldn't convert sRGB to linear rgb prior to their analysis. Esp in the deep shadows, we still don't know what exactly is going on. Note that sRGB is not a simple gamma-2.2 color space...

[alohadave]

Are 14-bit Raw Images Really Any Better Than 12-bit Raw? - Photo Tips @ Earthbound Light

14-bit Raw Versus 12-bit Raw: Further Analysis and Comparison - Photo Tips @ Earthbound Light

[RonHendriks1966]

So the lowest iso setting is 80 giving that great 14,1 on dynamic range. Would it be possible to stretch it to iso-64 or even iso-50 and how much would DR improve? Would it need 16-bit RAW to capture or is 14-bit still enough?

[falconeye]

So the lowest iso setting is 80 giving that great 14,1 on dynamic range. Would it be possible to stretch it to iso-64 or even iso-50 and how much would DR improve? Would it need 16-bit RAW to capture or is 14-bit still enough?

ISO 80 was possible because appearently, Sony has managed to keep the same full well capacity of their 12MP sensor pixels at 16 MP. It has nothing to do with stretching ISO. A large full well capacity is expensive and soon we'll know if Pentax ordered a more expensive production variant of the sensor than Nikon did. Appearently, the A55 carries a cheaper variant.

An artificial stretch to lower iso (a hidden +EV compensation actually) doesn't alter the DR score the way DxO measures it. They aren't that easily fooled.

[falconeye]

Note

In my above post, I see a link to Sony (Sonystyle) which I did not write. I will investigate the case. Should I find out that Adam deliberately empowered the forum software to alter my intellectual property to contain commercial links, then this will have consequnces.

EDIT
It is not happening in this post. I did not type or copy paste the link. I don't know what happened. If I edit the post, I see no link. Do others see the link too? I'm currently on an ipad which sometimes marks text like phone numbers too.

[dosdan]

Falconeye, no, I see no such link in your ISO 80 post in the FF browser.


Dan

[Class A]

It is not happening in this post. I did not type or copy paste the link. I don't know what happened. If I edit the post, I see no link. Do others see the link too?

I see it in my Seamonkey browser.

I just tried to create a post using "Sony" to see whether it would be embellished with a link. It wasn't. I've never seen such an embellishment before. Very strange that you cannot see the link when you are editing the post.

[dosdan]

I'm running ADBlock Plus. I see that link when I disabled ABP. The tooltip says "Shopping link added by Skimlinks".

Dan

[falconeye]

I'm running ADBlock Plus. I see that link when I disabled ABP. The tooltip says "Shopping link added by Skimlinks".

Dan

Dan, thanks for the update. Adam did indeed activate Skimlinks on the forum a while back. But he didn't activate its spam mode. Now, he obviously did. Or I just found a first customer (Sony).

Can somebody please consider starting a petition against this nonsense? I'm too tired at this BS. I won't even PM Adam about it. I'll rather just post elsewhere.

And to make it crystal clear to Adam: I don't grant permission to you to sell my intellectual property or alter it to contain spam links. I grant permission to you to publish it and that's it. I may sell you a license if you need it but it won't come free.