Noise acid test: California Sky

[Gabriele Sartori]

I spent a bit of time this morning shooting the sky. My only purpose whas to see if any noise was there since I’ve a photo with a noisy sky.

I took nine different photos in the very same position in few minutes in order to avoid differences. The sky was totally blue, not a single cloud was there. Shots were rigorously NEF.
Photos were taken at 3 different ISO speeds: 200,400,800
Each speed had 3 level of sharpening: Auto, Normal, High

Lens was the AF 24-120 (Nikon) and the focal length was 24. Aperture priority with F11.

Test was done importing all the picture in Photoshop, enlarge them to their natural size of 100%, create 9 small windows of equal dimension showing the same portion of sky (I took houses on the horizon as well in order to have reference points.

All nine have noise! I think that is due to different sensitivity that each pixel has. This is probably one of the main factors that make the D60 different. I checked, the D60 has noise as well, but much less. Noise is clearly something that can be changed quite a bit playing with ISO, focusing, sharpening etc. This is probably what is making this machine more difficult to “tune”.

Interesting is that ISO wasn’t really making a huge difference, although it was clearly increasing with higher numbers, in particular @ 800.

I checked carefully old and new photos done with my CP950 and there was noise as well but less! The CCD of the D100 is evidently noisy (in the sense that each pixel has a different response like in all the CCD but is more pronounced here)

This is it. During most of my life I designed computers, by education and profession I like the scientific method. I did a simple test here but there are no doubts, the D100 is noisier than other cameras. I’m sure that is not visible on paper.

I suspect that it is fixable. If Nikon has the manner to map a curve for each pixel (just question to see how much flash memory they have in there) they can realign and equalize all of them. May be the camera is already working like that and simply we got the first batch that was put in production in great hurry.

Opinions?

Thanks
Gabriele

 

[B Petterson]

Interesting.

Did you print any 8x10 or larger?

Birger

 

[Gabriele Sartori]

I just watched with A LOT OF attention on the screen. Must pay attention though, the computer screen itself may have some similar noise due to the ramdac etc (or pixels if you have a TFT screen). You can see the photo noise though because if you scroll up and down the screen noise doesn't move while the photo noise move.

I'm planning to print them later, however I think that even with the best ink-jet photo printer it will not be noticeable. It is noticeable on the screen though. Again, I don't think is realy noise, I think that is more pixels response. They should have done a better job equalizing them

Gabriele

Interesting.

Did you print any 8x10 or larger?

Birger

 

[Rick Stirling]

California eh? What was the ambient temperature and what was the temperature of the camera? Did the camera spend much time in the sun soaking up infra-red? ;-) Try this next time ... after you shoot the sky, put the lens cap on and shoot a dark frame at the same settings ... analyze the dark frame for any pixels that are non-black ... quantity and deviation. Compare that to the sky images. Put an IR blocking or high pass filter over the lens, see how much IR leakage there is that is biasing the visiable image.

Cool the camera down to 40 deg F and do the tests over. Compare with previous results.

Above all, stay curious and keep an open mind. ;-)

cheers,
Rick

p.s. there are tools available to remove +90% of any random noise from digital images. Here's and example ... It's a 100% crop from an image I took last week at Yellowstone Park ... Old Faithful. Look at the left and right sides of the plume. The left is untouched. The right has had noise reduction processing done. No other manipulation to the image was done.

I spent a bit of time this morning shooting the sky. My only
purpose whas to see if any noise was there since I’ve a photo with
a noisy sky.

 

[garfield]

You didn't mention "noise reduction".

Didn't I read somewhere that the D100 has noise reduction "off" by default, the D60 "on" by default?

Did you shoot with noise reduction?

I spent a bit of time this morning shooting the sky. My only
purpose whas to see if any noise was there since I’ve a photo with
a noisy sky.

I took nine different photos in the very same position in few
minutes in order to avoid differences. The sky was totally blue,
not a single cloud was there. Shots were rigorously NEF.
Photos were taken at 3 different ISO speeds: 200,400,800
Each speed had 3 level of sharpening: Auto, Normal, High

Lens was the AF 24-120 (Nikon) and the focal length was 24.
Aperture priority with F11.

Test was done importing all the picture in Photoshop, enlarge them
to their natural size of 100%, create 9 small windows of equal
dimension showing the same portion of sky (I took houses on the
horizon as well in order to have reference points.

All nine have noise! I think that is due to different sensitivity
that each pixel has. This is probably one of the main factors that
make the D60 different. I checked, the D60 has noise as well, but
much less. Noise is clearly something that can be changed quite a
bit playing with ISO, focusing, sharpening etc. This is probably
what is making this machine more difficult to “tune”.

Interesting is that ISO wasn’t really making a huge difference,
although it was clearly increasing with higher numbers, in
particular @ 800.

I checked carefully old and new photos done with my CP950 and there
was noise as well but less! The CCD of the D100 is evidently noisy
(in the sense that each pixel has a different response like in all
the CCD but is more pronounced here)
This is it. During most of my life I designed computers, by
education and profession I like the scientific method. I did a
simple test here but there are no doubts, the D100 is noisier than
other cameras. I’m sure that is not visible on paper.

I suspect that it is fixable. If Nikon has the manner to map a
curve for each pixel (just question to see how much flash memory
they have in there) they can realign and equalize all of them. May
be the camera is already working like that and simply we got the
first batch that was put in production in great hurry.

Opinions?

Thanks
Gabriele

 

[Gabriele Sartori]

You didn't mention "noise reduction".

Didn't I read somewhere that the D100 has noise reduction "off" by
default, the D60 "on" by default?

Did you shoot with noise reduction?

The noise reduction in the D100 is probably the best in the market, however (if I'm not wrong) it kicks in only for when you have exposure of 1/2 second or longer. I hope that I'm wrong but this is what I learned in the manual, so all my photo weren't with noise reduction. I would love to have that I hope that we can get it in a future FW upgrade.

Regards

 

[Gabriele Sartori]

Test

 

[Gabriele Sartori]

California eh? What was the ambient temperature and what was the
temperature of the camera? Did the camera spend much time in the
sun soaking up infra-red? ;-) Try this next time ... after you
shoot the sky, put the lens cap on and shoot a dark frame at the
same settings ... analyze the dark frame for any pixels that are
non-black ... quantity and deviation.

Rick, I understand that you are trying to be humorous, but I'm not a moron. I'm a 44 Y/O veteran of the high tech industry. I do CCD photo using a telescope and I know these issues.

The camera has been kept on purpose in a fresh, ventilated room in the shadow at about 72 degree. All the photos were taken in few minutes and the camera was on a tripod in the shadow when I took them. Temperature in Fremont where I live was 77 degree today but lower in my backyard since I'm on the hilly area called Mission district and it is a little windy.

I didn't think that was necessary to use dry ice on the camera since it was important to test the camera in a natural environment although I took all the precautions in order to avoid false reading due to the heat.

Bottom line the CP950 has less noise than the D100. Sorry but is the fact.

Am I going to lose my sleep? No

Am I'm going to give the camera back? No

Is this noise visible when you print the photos? 99% I think No.

Does the noise upset me? Yes, quite a bit, because it shouldn't be worse than what I see on a camera 4 Y/O coming from the same company and because this noise is having other implications and according my opinion is the fundamental problem behind the entire sharpness, blurriness story. The camera is sharp and not blurry but it may be if you see the noise and try to avoid it.

There are chances that not everybody has the same noise level, either because of luck (a more uniform CCD) or better tuning of their camera. Mine is tolerable, someone else may have a worse situation.

About the noise cancellation SW I know there are many. Yours seems to be quite good, can you let me know what is it?

Problem with SW is that you have to manage photo by photo unless you have a tool that allows batch processing.

Regards
Gabriele

 

[Rick Stirling]

Gabriele,

Sorry about the humour. ;-) The software is called Neat Image ( http://www.neatimage.com/ ) and I understand they're going to be adding batch processing to the next release. It's an interesting program. I've revisited some of my older images and given them a new lease on life. As you might expect a major use of this program is cleaning up cropped images for internet viewing but the program can also make a high iso image look like it was shot with 2 - 3 stops lower ISO. Pretty cool.

I think that the prime issue with the D100 for everybody is they are not used to looking at an image with such high magnification. I think a D100 image at 100% on a 1024 * 780 screen respresents a poster size photo that you have your nose in. My LCD monitor is 15 inches horizontally , that means a D100 image 3000 x 2000 would be a 'photo' 30 x 45 inches that I'm viewing only 18 inches away. Try stepping away from the monitor about 8 feet and see how much noise you can see. ;-)

I imagine that if you printed some of the D100 images posted here on a nice dye-sub printer the results would be phenominal. ( I guess I'll have to go knock off a Kodak Photo Kyosk to get that printer! ;-) )

I'm lusting after a D100 but will have to wait a while ... just spent $5000 on a Suburban. ;-(

cheers,
Rick

ps. I went to school ( UCLA ) in SoCal many many moons ago. Spent 20 years there altogether. Now I'm in the mountains of Idaho.

Rick, I understand that you are trying to be humorous, but I'm not
a moron. I'm a 44 Y/O veteran of the high tech industry. I do CCD
photo using a telescope and I know these issues.

 

[Gabriele Sartori]

Gabriele,

Sorry about the humour. ;-)

No problem Rich, being Italian I'm a funny guy myself, I'm probably taking this too seriously.

The software is called Neat Image

Thanks, I need a SW like that I look into it and I'll try to get the one with batch processing.

I think that the prime issue with the D100 for everybody is they
are not used to looking at an image with such high magnification. I
think a D100 image at 100% on a 1024 * 780 screen respresents a
poster size photo that you have your nose in. My LCD monitor is 15
inches horizontally , that means a D100 image 3000 x 2000 would be
a 'photo' 30 x 45 inches that I'm viewing only 18 inches away. Try
stepping away from the monitor about 8 feet and see how much noise
you can see. ;-)

I agree with that, indeed I came up with the very same explanation on a previous post that I did when someone was bashing the D100. Mine was a relative discussion, why a CP950 has less noise than a D100? May be I was just lucky with the CP950 that was built in magnesium and it is still today a real queen according to my opinion. It has only 2 Mpixel but when you print the photos it kicks the butt of many "modern" 4 Mpixel with a CCD small as a bug.

I imagine that if you printed some of the D100 images posted here
on a nice dye-sub printer the results would be phenominal. ( I
guess I'll have to go knock off a Kodak Photo Kyosk to get that
printer! ;-) )

I'm sure about that, indedeed I said in all my post. Also, this noise is particular strong with the kind of sky that we have in CA around 1PM in the summer (on a day w/o pollution). I took photos with a sky not so blue and there was no noise or whatsoever

I'm lusting after a D100 but will have to wait a while ... just
spent $5000 on a Suburban. ;-(

I hope I don't upset anyone, if it wasn't for the glasses that I have I would have bought the D60. The D100 is a fantastic camenra and has more than what I need but I still believe that Canon did a slightly better job.

ps. I went to school ( UCLA ) in SoCal many many moons ago. Spent
20 years there altogether. Now I'm in the mountains of Idaho.

eheh I was born on the Alps I know what you mean. Recently I went back and for the first time in my life I said "WoW , why I left all this??? "

Thing is that it is hard to appreciate the beauty of a place where you have been for ever until you miss it for a while.

Ciao!
Gabriele

 

[Rick Stirling]

Grazie per le osservazioni intelligenti. Molto goda tuo nuova macchina fotografica.

ciao,
Rick

ps. my italian is pretty gross. ;-)

 

[Thorsten49231]

Sorry to point out that the noise reduction really screws up the plume as well. I prefer the plume without the color banding and the noise in the sky!

Also, the pixels in the D100 are larger than the ones in the CP990, so it should have less noise at the source. I really hope it doesn't!

--

• Thorsten

 

[Rick Stirling]

Huh? What's wrong with your monitor? There is no color banding in that image. Are you using 32 bit color? I can see some banding if I use 16 bit color on my monitor. The noise reduction had virtually no impact on the image detail in the plume. If you blow it up to 300% you can see some faint residual jpeg artifacts, but none of that will show in a print. Remember you are looking at a 100% crop. I could have run a jpeg artifact filter on it too but didn't in this example.

Rick Stirling

Sorry to point out that the noise reduction really screws up the
plume as well. I prefer the plume without the color banding and the
noise in the sky!

Also, the pixels in the D100 are larger than the ones in the CP990,
so it should have less noise at the source. I really hope it
doesn't!

--

• Thorsten

 

[alfred_schiele]

Hi

... I did a
simple test here but there are no doubts, the D100 is noisier than
other cameras. I’m sure that is not visible on paper.

I agree to your opinion, there is noise in D100 images. I think the D100 images look more "puristic" without artificial noice smoothing/edge enhancing like D60 images look to me. I think there are postprocessing ways to get similar results in most (not all) aspects.

But in the end, if you think it is not visible in prints, why worry about it? It's exacly the way I start looking at it. If prints (up to A4) look fine, I get what I want.

I suspect that it is fixable. If Nikon has the manner to map a
curve for each pixel (just question to see how much flash memory
they have in there) they can realign and equalize all of them. May
be the camera is already working like that and simply we got the
first batch that was put in production in great hurry.

A nice idea. Do you know any manufacturer doing this? Of course, real noise couldn't kept down this way...

Opinions?

All expectations in d-cameras should be adjusted

Mine is that growing 6MP cameras are opening the door to make digital SLR's more interesting general tools. Indeed I think with 6MP cameras the start to "serious" 35mm equivalent cameras has happened. To be honest, I had been happy having got shots with the D100 ASA 400 noise levels from film 20 years ago... But we are just on the beginning of this process and much better sensors and cameras will follow.

I think it is pretty useless to dig for the latest 10% better while in some years all these cameras will be hoplessly outdated.

Regards, A. Schiele

 

[Thorsten49231]

Uhhh, crush, gasp... I did leave my monitor in 16-bit color. Image looks fine in 32bpp. Darn, really got to turn those brain cells on before posting!
Sorry for the confusion!

Huh? What's wrong with your monitor? There is no color banding in
that image. Are you using 32 bit color? I can see some banding if
I use 16 bit color on my monitor.

--

• Thorsten