Open Photography Forums  
HOME FORUMS NEWS FAQ SEARCH

Go Back   Open Photography Forums > Digital Camera Discussion > All other DSLR's and Four Thirds, 4/3

All other DSLR's and Four Thirds, 4/3 All DSLRs excluding Canon and Nikon mounts ie Sigma, Pentax, Olympus, Sony, Leica R Back DSLRs and 4/3 System

Reply
 
Thread Tools Display Modes
  #1  
Old September 7th, 2009, 01:28 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default New Sony A200 Hotpixels problem - Serios Build quality issues

Hi All,

I have recently purchased a brand new Sony A200 from an authorised dealer with proper documents and a 3 year warranty.

To my horror the camera has ~20 hotpixels (verified on the 77th shot, if someone is interested I will post the picture). Is it normal for a brand new DSLR to have such a high number of hotpixels?

Please advice.

Regrads,
Reply With Quote
  #2  
Old September 7th, 2009, 03:29 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Hi All,

I have recently purchased a brand new Sony A200 from an authorised dealer with proper documents and a 3 year warranty.

To my horror the camera has ~20 hotpixels (verified on the 77th shot, if someone is interested I will post the picture). Is it normal for a brand new DSLR to have such a high number of hotpixels?

Please advice.
Hi Rohit (or is Nanda your first name?),

Can you give a bit more detail about exposure time, ISO, and how you classify a pixel as 'hot'?
A link to a full size image may also be helpful in diagnosing what's going on.

Cheers,
Bart
Reply With Quote
  #3  
Old September 7th, 2009, 04:16 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Dear Bart,

You are correct, my first name is Rohit.

I am attaching a shot for your reference. It is the 77th shot from the camera at ISO100 and 2.5 sec shutter. (I recently went to a night safari and a 2.5 -3 sec shutter was common in pictures.)

The image is a test picture so that while testing the hotpixels come out separately. You will see bight green and purple pixels. Further the camera has gone to the serice centre 3-4 times for software upgrades and pixel mapping but it has not helped at all.

Reply With Quote
  #4  
Old September 7th, 2009, 05:56 AM
Cem_Usakligil Cem_Usakligil is offline
Senior Member
 
Join Date: Jun 2006
Posts: 5,940
Default

Quote:
Originally Posted by Rohit Nanda View Post
Hi All,

I have recently purchased a brand new Sony A200 from an authorised dealer with proper documents and a 3 year warranty.

To my horror the camera has ~20 hotpixels (verified on the 77th shot, if someone is interested I will post the picture). Is it normal for a brand new DSLR to have such a high number of hotpixels?

Please advice.

Regrads,
Yes, this is very normal. Read this article about hot pixels and just relax.

Cheers,
__________________
Kind Regards, Cem

flickr
website
Reply With Quote
  #5  
Old September 7th, 2009, 07:26 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Dear Bart,

You are correct, my first name is Rohit.

I am attaching a shot for your reference. It is the 77th shot from the camera at ISO100 and 2.5 sec shutter. (I recently went to a night safari and a 2.5 -3 sec shutter was common in pictures.)

The image is a test picture so that while testing the hotpixels come out separately. You will see bight green and purple pixels. Further the camera has gone to the serice centre 3-4 times for software upgrades and pixel mapping but it has not helped at all.
Okay,

First of all, as exposure times get longer there will be a gradual increase of noise caused by thermal effects. This noise should be minimal with exposures shorter than appox. 1 seconds. For longer exposures the noise will increase, that's why some camera's allow to subtract a dark frame which will eliminate most of the pattern noise, but not the random noise (the procedure will probably increase the random noise).

Hot pixels are pixels that are systematically always brighter than they should be, and there will be more of them and with higher intensity with longer exposure times. There is 1 characteristic that stays the same though and that sets them apart from random noise, they have the exact same pixel coordinates with each (long) exposure. So that would take at least 2 exposures to prove.

There exist also "stuck" pixels which always have more or less the same value, or "dead" pixels which are alway more or less zero.

In your example there are a couple of pixels that are relatively hot, there are 16 with an amplitude of more than 100, and a total 63 are hotter than 50. Whether that's a high number or not, depends on how other cameras with the same sensor do. Also important is how your DRO settings, contrast, and sharpening levels were, as that may influence the behavior of hot pixels.

When you feel that the amount of hot pixels is excessive, you should build a good case for Sony so they know what to look for. I suggest to make several exposure pairs with increasing exposure time (e.g. 1/15s, 1/8s, 1/4s, 1/2s, 1s, 2s 4s). Cover the viewfinder, and the lens, or better yet put the body cap on and shoot without lens. That will exclude light from outside the camera. For each exposure pair, the hot pixel positions should be the same, and approx. at the same level, thus excluding random noise. Thoughout the sequence the intensity will increase and the positions will be the same.

When you see the intensity increasing, and the hot pixels do not show at the shortest exposures, then mapping them out will probably be problematic. Stuck, and very hot pixels, may be candidates for mapping out, and you will know the pixel coordinates from the test sequence.

For images where the hot pixels are still objectionable, you can resort to postprocessing.
There are 2 free utilities you could try:
http://www.mediachance.com/digicam/hotpixels.htm and
http://www.mediachance.com/digicam/blackframe.htm

Again, whether the amount of hot pixels is substandard for the sensor in your camera model is hard to say without comparison to other units. However, with a bit of methodical testing you can at least pinpoint the worst offenders, and to some extent repair them in postprocessing.

Apparently (according to DPReview) your camera offers a noise reduction option for exposures longer than 1 second, it would probably help to reduce some of the systematic hot pixels at longer exposure times.

Hope that helps a bit,
Bart
Reply With Quote
  #6  
Old September 7th, 2009, 08:47 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Yes, it is correct as the exposure time increases the number of hot pixels also increase. There are none currently below 1 second but start at ~ 2 secs. (The attached shot is at ISO100 with 2.5 sec shutter)

The hotpixels are at the same place always but at longer exposure new pixels add up.

As I mentioned previously, at the service centre we took a test shot at ISO 100 with NR at 15 sec shutter speed that had huge purple cormers on the top. (All test pictures included the one posted is taken with the lens cap and eyecap.) This pictures are currenlty in my CF card and I donot have the card reader and the camera is with the sony service centre. Trying to arrange the card reader and will definitely post the purple corner pictures in some days.
Reply With Quote
  #7  
Old September 7th, 2009, 09:20 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Yes, it is correct as the exposure time increases the number of hot pixels also increase. There are none currently below 1 second but start at ~ 2 secs. (The attached shot is at ISO100 with 2.5 sec shutter)

The hotpixels are at the same place always but at longer exposure new pixels add up.

As I mentioned previously, at the service centre we took a test shot at ISO 100 with NR at 15 sec shutter speed that had huge purple cormers on the top. (All test pictures included the one posted is taken with the lens cap and eyecap.) This pictures are currenlty in my CF card and I donot have the card reader and the camera is with the sony service centre. Trying to arrange the card reader and will definitely post the purple corner pictures in some days.
Purple corners may be the result from "Amplifier glow" (assuming your lens cap doesn't leak a bit of light). You can check if it is still in the same position when you turn the camera upside down. If it is in the same place and same strength, then it is probably generated by the camera. The slightly higher temperature of the signal amplifier(s) generates a thermal 'signal' with long exposures. I'm not sure about the Sony but it is a known phenomenon with some other cameras (e.g. some Canon DSLRs have it in one corner), but usually with longer exposures.

Cheers,
Bart
Reply With Quote
  #8  
Old September 7th, 2009, 11:28 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Yes Bart, you are correct. I first felt if it may be due to light coming in from the corners of the lens cap. We tried taking the picture by switching off the room light in the service centre but that did not help.

The purple corners did increase in intensity and area when the exposure increased. Is this Ok?
Reply With Quote
  #9  
Old September 7th, 2009, 02:11 PM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Yes Bart, you are correct. I first felt if it may be due to light coming in from the corners of the lens cap. We tried taking the picture by switching off the room light in the service centre but that did not help.

The purple corners did increase in intensity and area when the exposure increased. Is this Ok?
The people in the service centre should be able to confirm if there are electronic components that heat up in those spots (remember that the image is projected upside down). The thermal noise increases exponentially with temperature and time. I don't know the Sony cameras well enough to confirm it, but from what you describe it seems consistent with "Amp glow". A "Dark frame subtraction" should be able to cure part of that issue, either as an in-camera noise reduction, or as a postprocessing step.

Cheers,
Bart
Reply With Quote
  #10  
Old September 7th, 2009, 09:37 PM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Actually the service team guys are unable to understand the issue. The camera has gone into 3 rounds of software upgrades and pixel mapping.

We did feel that it could be due to the heat, hence we also tried clicking photos under the direct AC blow and without holding the camera so that chances of heating are minimised.

The test pictures were taken with long exposure NR switched on as that is the default setting in Sony when the camera is turned on. As per my understanding the NR function does a black frame subtraction to reduce hotpixels automatically.
Reply With Quote
  #11  
Old September 8th, 2009, 02:17 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Actually the service team guys are unable to understand the issue. The camera has gone into 3 rounds of software upgrades and pixel mapping.
As explained, you'll only have some success from mapping out, if the pixels are "stuck" or "dead" or consistently (at all exposure times) hotter than they should be. One cannot successfully map out random noise.

Quote:
We did feel that it could be due to the heat, hence we also tried clicking photos under the direct AC blow and without holding the camera so that chances of heating are minimised.
Don't underestimate the internal temperature of a camera. It might take a while to cool it on the inside as well, and it will re-heat when being used.

Quote:
The test pictures were taken with long exposure NR switched on as that is the default setting in Sony when the camera is turned on. As per my understanding the NR function does a black frame subtraction to reduce hotpixels automatically.
That would be the usual scenario for most cameras. What remains is random noise, and that's just a function of all internal components working together. When the random noise shows clearly visible spikes, then noise reduction (e.g. NeatImage or NoiseNinja) software may not be able to improve the result. In that case, first use one of the tools I mentioned above. Also remember that repetitive opening, editing, and saving, will make the quality of JPEGs deteriorate. I therefore recommend to save the treated images as PNG or TIFF, or use the result as a layer in Photoshop to replace the "hot" pixels.

Cheers,
Bart
Reply With Quote
  #12  
Old September 8th, 2009, 03:04 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Thanks a ton Bart. Your reply was really helpful.

I have another query, if the hotpixels come at the same place always can that be cleaned by pixel mapping?
Reply With Quote
  #13  
Old September 8th, 2009, 03:55 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Thanks a ton Bart. Your reply was really helpful.

I have another query, if the hotpixels come at the same place always can that be cleaned by pixel mapping?
Not really, because a hot pixel has a variable brightness (varies with exposure time, temperature, and subject brightness) unlike a stuck or dead pixel. A variable brightness pixel cannot be distinguished from actual scene content in a single shot. If the tools of the service centre could map out selected pixels, then that would also affect shorter exposures where the pixels are not considered hot. It would sort of solve the hot pixels but create other issues at different conditions. I don't even know if the software used by the service centre can fix anything elso than stuck and dead pixels.

That's why e.g. Astro photographers go through lengths to separate random and systematic noise patterns. Systematic patterns can be suppressed, but random patterns can only be reduced by taking e.g. the average of multiple exposures (which assumes static objects, no motion between the exposures). There is specialized software for such postprocessing procedures, e.g. IRIS or ImagesPlus, but I think it's too much specialized work for most people.

With todays high Mpixel cameras, there is a lot that can be done in postprocessing without too much deterioration of the images.

Cheers,
Bart
Reply With Quote
  #14  
Old September 14th, 2009, 09:24 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Dear All,

Please advice, how do I solve this issue. Photo attached. I have compressed the phot for easy upload.

Regards,
Reply With Quote
  #15  
Old September 15th, 2009, 03:57 AM
Bart_van_der_Wolf Bart_van_der_Wolf is offline
pro member
 
Join Date: Jun 2006
Location: Netherlands
Posts: 4,053
Default

Quote:
Originally Posted by Rohit Nanda View Post
Dear All,

Please advice, how do I solve this issue.
Hi Rohit,

Since there seem to be no other takers, here's my take on it.

That looks like some serious amp glow! If that manifests itself with exposure times measured in a couple of seconds, that's poor engineering. Luckily there are things that can be done, unfortunately the repair with postprocessing is not trivial.

What's needed is what astrophotographers call a flat field correction.
Here's an example of how that could be done with the free IRIS software (for Windows OS only):
http://www.astrosurf.com/buil/iris/tutorial4/doc14_us.htm

Other software uses somewhat different approaches, but they all amount to the same; Mathematical division (after normailzation to 1.0 for the brightest pixels) of the image data by a smoothed version of the uneven background. It works best if all images have a linear gamma, so before a 1/2.2 Gamma adjustment for display.

You can do a poor approximation with Photoshop by using a blurred image copy of a dark exposure (same exposure time as actual image, but with lens cap on), and use the "Apply image" command with the multiply function and checking the "Invert" option for the blurred background image. The blurred background image should be normalized first, e.g. with levels. The result will be far from perfect, but better than the original. For better results you should use the astrophoto approach which, amongst others, adjusts for gamma, or spend some time with curves to adjust the luminosity of the background image to compensate for the gamma. You can also experiment with a blurred, normalized, and inverted background layer in "Linear burn" mode.

Cheers,
Bart
Reply With Quote
  #16  
Old September 16th, 2009, 01:27 AM
Rohit Nanda Rohit Nanda is offline
New Member
 
Join Date: Sep 2009
Posts: 8
Default

Thanks a ton. That really helps.

Regards,
Reply With Quote
Reply

Bookmarks

Tags
hotpixels, quality issues

Thread Tools
Display Modes

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 Off
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 05:34 AM.


Posting images or text grants license to OPF, yet of such remain with its creator. Still, all assembled discussion 2006-2017 Asher Kelman (all rights reserved) Posts with new theme or unusual image might be moved/copied to a new thread!