Go Back   Steve's Digicams Forums > Digital SLR and Interchangeable Lens Cameras > Nikon dSLR

Reply
 
Thread Tools Search this Thread
Old Jun 22, 2011, 10:47 AM   #1
Member
 
anm8ed's Avatar
 
Join Date: Jun 2005
Posts: 48
Default D80 Sensor dying?

I have a nikon D80 that I bought in August of '08. Taken less then 10k pictures and used to have some stuck/dead pixels. No big deal and they were not always visible. Easy to clean up a couple if I really liked the picture. But last night I was taking pictures of some lightning and what I came up with was tons of stuck/dead pixels.

Red, Blue, white, green, and they are everywhere. Below should be two pictures. The first is the actual picture, which looks clean probably because it was resized by photobucket. And the second is a 100% crop of the middle of the picture. And that is indicitive of the entire thing.

I have taken such pictures before and never seen it this bad. I have taken pictures an air shows and seen some of the red and white ones before. But usually just a few.



anm8ed is offline   Reply With Quote
Sponsored Links
Old Jun 22, 2011, 11:08 AM   #2
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

Well, that was a 30 second exposure at ISO 360 (my guess is that Auto ISO was used for an odd value like that). Was long exposure Noise Reduction turned on?

If not, then you can expect to see some hot pixels on longer exposures, especially with higher ISO speeds.

But, even if long exposure NR was turned on, with exposures that long, it's not going to be perfect.

Hot pixels are when photosites record a higher value (brighter) than they should, usually with longer shutter speeds. When the problem occurs at faster shutter speeds, they're considered "stuck" (always bright). Virtually all sensors will have hot pixels in some conditions. The longer the shutter is open (and the darker the conditions since not enough photons are hitting the sensor) the more likely a photosite won't be linear in it's output when responding to light, and it will record a higher value than it should.

Temperature also impacts it (which is why you see cooling systems for sensors in digital cameras used in astronomy). The more you test for them, the more you'll likely have, especially without any light hitting the sensor.

You just don't see hot pixels with most newer cameras because of built in noise reduction to subtract them out on longer exposures, since the ones that occur at typical shutter speeds have already been mapped out at the factory.

So, make sure your long exposure noise reduction is turned on. That's what it's there for (to remove hot pixels on longer exposures).

It works on photos taken longer than 1 second (where hot pixels usually show up). The way it works is by taking two photos. The first photo is of your subject. Then the camera automatically takes a second image using the same settings with the shutter closed.

The camera then notes the positions of any hot pixels in the "dark frame" (second) exposure, and maps them out of the actual exposure. It knows where to find them because the dark frame exposure was taken at the same time, using the same settings, with the camera at the same temperature. In most cases the hot pixel locations will be the same in both images using this technique.

But, the longer the exposure, the greater the chance it will miss some of them.
JimC is offline   Reply With Quote
Old Jun 22, 2011, 2:20 PM   #3
Senior Member
 
big_potato's Avatar
 
Join Date: Jun 2005
Posts: 487
Default

>>> Well, that was a 30 second exposure at ISO 360

Not a surprise to me to find there are hot/dead pixels with this exposure settings.

In fact I came across some folks in another forum who actual made a:

"Discovery of the Century"

to find out that D7000 had some tens of hot/dead pixels

at ISO1600 for 15 sec.


If you try hard to find some imperfect bahaviours of the sensor, you will certainly be able to find them someday .... but at the cost of burning the sensor.

I mean I worry that the hot/dead pixels would be there forever, once you have tested them out.

I think the general rule applies here:

If it doesn't break, don't test it just for the sake of knowing.
big_potato is offline   Reply With Quote
Old Jun 27, 2011, 5:50 PM   #4
Senior Member
 
deadshot's Avatar
 
Join Date: Apr 2006
Location: Hertfordshire UK
Posts: 759
Default

What's interesting here is that the instruction manual on the D5000 shows you how to set "high " normal" and "low" noise reductions but doesnt tell you what Nikon consider the relevent I.S.Os to be.
It just states that noise reduction kicks in at I,S.O 800.
So we can guess Low= 800/1200 Normal 1300/2000 High 2100 and above????
__________________
D5100 +18-200mVR Nikkor lens.
SB400 Flash, ML-L3 Remote.
SB 700 Flash
Holster + Shoulder Bag.
Beike carbon 4 section tripod/monopod
Gorillapod SLR Zoom + BH1 ball head
Panasonic FZ1000
Panasonic FZ200
Nissin D i40 Flash
+ SLR Gorillapod
deadshot is offline   Reply With Quote
Old Jun 27, 2011, 6:42 PM   #5
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

Those High, Low and Normal settings refer to the amount of Noise Reduction being applied, not the iSO settings it will be applied at (it sounds like it works at anything above ISO 800 if that's what the manual says).

But, High ISO Noise reduction settings wouldn't have anything to do with the helping OP with his problem. High ISO Speed is designed to reduce noise at higher ISO speeds, and is not the same thing as tbe Long Exposure Noise Reduction I discussed in my previous post.

Long Exposure Noise Reduction (a.k.a., Dark Frame Subtraction) works at any ISO speed and is designed to use dark frame subtraction on longer exposures to remove hot pixels (by taking a second photo with the shutter closed and looking for pixel locations that are bright, then mapping those locations out of the actual exposure by interpolating pixels from adjacent pixels to replace them). With most cameras, it kicks in at shutter speeds 1 second or longer if it's turned on. See my previous post for more details about it.

It works because in most cases when two photos are taken using the same settings (the camera is taking a second shutter closed photos when it's turned on), with the camera at the same temperature, the locations of the hot pixels will be the same in the dark frame (shutter closed) image as they'll be in the actual photo of your subject.

So, all the camera needs to do is look for any bright pixels in the dark frame exposure to determine where they are in the real photo of your subject (because everything should be dark in the shutter closed exposure, so it notes the locations of any pixels that are not dark and assumes they're bad in the actual exposure, too). Now, different cameras will use different thresholds for determining if a pixel location is bright enough to remove (replacing it with values from adjacent pixels in the actual photo of the subject). But, in most cases, Dark Frame Subtraction (a.k.a., Long Exposure Noise Reduction) does a pretty good job.

That's a totally different technique compared to reducing noise at higher ISO speeds.

Usually, the only control you'll have over how long exposure noise reduction is applied is a setting to enable or disable it. IOW, Long Exposure Noise Reduction is usually an On/Off setting, and tells the camera if you want it to use Dark Frame Subtraction on Longer Exposures or not. I'd leave it on if you don't want hot pixels on longer exposures (as in the OP's photos keeping the shutter open for 30 seconds).
JimC is offline   Reply With Quote
Old Jun 28, 2011, 6:53 AM   #6
Senior Member
 
deadshot's Avatar
 
Join Date: Apr 2006
Location: Hertfordshire UK
Posts: 759
Default

Thanks JimC,
I wonder if using long exp.s creates hot pixels the are there for good(get stuck) as I never noticed any on my D40 until after I had used some long exp.s on high I S Os,then they were always there even after I sent my D40 back to Nikon for them to be cleared. Nikon eventually replaced the sensor. My Son in law has an Olympus that has a feature that purports to get rid of them and it seems to work as once done they appear to be gone for good.
P.S Just as a matter of info the D5000 manual states that the long exposure n.r will operate if ON at speeds slower than 8 secs,
__________________
D5100 +18-200mVR Nikkor lens.
SB400 Flash, ML-L3 Remote.
SB 700 Flash
Holster + Shoulder Bag.
Beike carbon 4 section tripod/monopod
Gorillapod SLR Zoom + BH1 ball head
Panasonic FZ1000
Panasonic FZ200
Nissin D i40 Flash
+ SLR Gorillapod

Last edited by deadshot; Jun 28, 2011 at 6:57 AM.
deadshot is offline   Reply With Quote
Old Jun 28, 2011, 9:13 AM   #7
Administrator
 
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378
Default

Hot pixels are normal on long exposures, and you don't want to add those to a hot pixel table unless they become stuck (where they always register a value that's too high, regardless of shutter speed). Otherwise, could end up with a lot of unused pixels on all exposures (it's not uncommon to see hundreds of hot pixels on longer exposures with Long Exposure Noise Reduction turned off).

I doubt long exposures would cause the problem.

Note that when they do become hot (too bright) at typical shutter speeds (not just during long exposures), then they're referred to as "stuck", and Nikon can update a bad pixel table in EEPROM so that they're always mapped out (with their values replaced by taking values from adjacent pixels during the interpolation process).

As you noticed, some manufacturers have a bad pixel remap routine built in that looks for brighter pixels and adds them to a table in EEPROM. Some Sony models (for example, the A580) and most Olympus models have a menu choice for that feature.

Some models also run a bad pixel remap routine automatically from time to time (some Sony models do this once/month).

With Nikon models, you have to send them back to the factory, where they have software that adds any bad pixels to a table in EEPROM. I've got some older sofware that can remap them for some coolpix models, and I just might have a copy of service software for the D100. :-)

But, I doubt you'll find a solution for most newer Nikon models. If they become annoying, a trip back to the factory will be needed.

All these routines do is use a dark frame shot (shutter closed exposure) to identify the locations of any bright pixels. Then, add them to a bad pixel table in EEPROM so that they're mapped out with their values replaced from adjacent pixels when taking photos. That's the same technique the camera uses for Long Exposure Noise reduction, only it doesn't add the pixels it finds on the dark frame exposure to EEPROM (instead, it's just a temporary list it uses just for the exposure taken just prior to the dark frame exposure, where it replaces any pixels in it that were bright in the dark frame exposure)
JimC is offline   Reply With Quote
 
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -5. The time now is 5:17 AM.