A700 terrible red channel noise problem

[David Kilpatrick]

I'm using the D5000 and processed from raw the images are between the A700 and A900 in final quality for Alamy. Sharper and less noise than the A900. I have not experienced any turquoise sky problems - which I did with the D80 and D70s - and generally like the response of the camera. It's a compact entry level model but produces perfectly acceptable results.

David

[agorabasta]

I'm using the D5000 and processed from raw the images are between the A700 and A900 in final quality for Alamy. Sharper and less noise than the A900.

If you make your raws available, I'm quite prepared to process them to even greater quality output, no matter if whatever Nikon or Sony.
But you are still trying to defy reality - they really killed the well depth with D5000 using the very same sensor; the results are quite predictable.

If you want a quick check - try more contrasty scenes at ISO 100 and judge for yourself what really happens there.

[agorabasta]

You tried to show us how your processing techniques would eliminate Sony red sky noise but you failed miserably.

-Sonolta

You are lying. Deliberately? Or just miserably?

[David Kilpatrick]

The D5000 base ISO is 200 - 100 does not exist, only Lo -1.0 which means they acknowledge from the start that it is not a recommended setting. Why on earth would I want to shoot contrasty scenes at ISO 100? And why would I want to send you raw files? I don't need to send anyone raw files and I have press NFR installations of every current commercial raw processor, plus the usual set of shareware and open source options.

I don't really care what you say, since my job involves testing as many DSLRs as I can secure - I've missed out on Panasonic and Samsung so far, but been able to get review samples of all the others (apart from Sony, which I tend to buy instead of borrowing). You might as well know that I can not think of a single DSLR which offers a better recovery of extreme contrast than the D5000, and that includes all my Sony DSLRs. It has a raw file which allows recovery of colour values from extremes, without colour clipping. I did not want to get to drawn into this 'colourful' exchange of views, but the record needs setting straight.

Here are some typical test pictures - the kind of thing I routinely do with DSLRs - taken with the D5000 using the new Sigma 10-20mm lens, aiming into a VERY bright sky and making exposures of 1/400th at f/11 (correct for the landscape at ISO 200), 1/1600th at f/11 (correct for the sky) and 1/4000th at f/11 (correct for the brightest clouds, extremely dark for the foreground):



In the next two posts, I will show the default Adobe Camera Raw conversion for the 1/400th and 1/4000th files, and an extreme adjustment of both with histograms visible.

David

[David Kilpatrick]

Example: 1/400th at f/11 - totally burned out sky. Correction down to -4 in ACR was possible without colour clipping, and Recovery could be added up to 100% also without arttifical grey posterization effects. However, I have only used -3 and 50 Recovery. Try that on a similarly bright A700/A900 or almost any other file.



The default (normal ISO 200 process)



With corrections as shown.

1/4000th extra dark version follows.

[David Kilpatrick]

Exposed for the brightest areas of the sky:



The original default process



The processed version with a rather extreme +4 exposure and recovery setting. I do not recommend such extremes but they are here to show it can be done.

I only have these files in the gallery on pBase so far for D5000:
http://www.pbase.com/davidkilpatrick/nikon_d5000
but may add others later.

David

[agorabasta]

Oh well, David...

First and foremost - that ISO 100 thingie is very real and important here, even if not exactly called so. At that setting the well depth is closest to overflow, so it really reveals the real well depth as it's being pushed harder than normal. And the cam's base ISO has a very faint relation to exploiting the well depth to the full. Actually, they should've put the base ISO closer to 400 - this way they could afford a higher quality raw curve...

Your 'extreme' example doesn't cut the mustard. Below is your 'blown' piece adjusted a bit from JPEG using mere single curve. Sure you didn't need much raw hassle pushing that wherever you please as it's mere 1.5 stops max overcooked - peanuts really; and raising shadows is simpler yet....

[agorabasta]

Now check to see what Agor's procedure produced. Did it solve the "Sony red channel noise problem"?

http://www.photoclubalpha.com/forum/vie ... isy#p25162

-Sonolta

Sure it did.

And you are looking at a very early a700 firmware from Nov'07. I don't remember if it's v1 or v2?
So what did you expect from a badly crippled raw?

Just go and do the same to the v4 sky and stop insinuating - it works as prescribed.

[David Kilpatrick]

Well, you have made your own point about your views and knowledge with your comments, and adjustment of the screen shot JPEG. I won't say anything further, as anyone reading this thread can now judge the value of your opinions for themselves. Thanks!

David

[stevecim]

David , Don . do both of you agree that there is a lot of noise in the RED channel? looking at the sky shots I have (a350) it also seems to have a lot of red noise, green has noise to a less extent and blue is very clean. (I can try and post crop is anyone is interested) Is the red noise a fault or is that they way Sony wanted it? Does the RED channel noise help produce "bright" colours?

[David Kilpatrick]

Yes, there is a lot of noise in the red channel in sky tones. No matter what camera make you buy, blue skies will always show the highest levels of red noise - blue tones in general will. That's because blue gradations depend mainly on the red channel for their luminance values. In my experience Sony is the 'worst' at this while Canon is the 'best' but there's a huge difference in the accuracy and look of the sky tones. Canon get smooth sky blue partly by not having a realistic sky colour at all. Nikon do much the same, pitching the sky colour towards a turquoise hue not a more purple tint.

Here is an example a RAW process preset in the set of Actions sold by Marcus Bell, and the effect it can have on Sony Alpha 900 sky blue. This preset is a 'Polariser' fake and it does not always work well, depends on the pic. Part of the Action setup changes the camera calibration in ACR, reducing blue luminance but also changing Vibrance and Saturation values. You can use the same controls to create raw process presets which clean up sky noise dramatically, without such an extreme adjustment (a useful one, though!).

David

[agorabasta]

Red channel noise of a700 is hardly worse than that of D5000. Actually, below ISO400 it's the other way around.

If you compare my 300% crop from Sonolta's ISO400 a700v1 raw file developed without any NR but with my 'cyan trick' http://lh6.ggpht.com/__bYe591qcSo/Slhcc ... _boost.jpg to this ISO200 D5000 example from DPR - http://a.img-dpreview.com/gallery/nikon ... c_0045.jpg , you may notice that measurable noise in the red channel is actually lower with the Sony sample.

[Birma]

Birma, something very crude is going on with the Sony red channel.

It certainly looks odd in your comparison - I think you possibly meant "Steve"?