Reading mode: Light Dark

Can "DeadPixelTest" be trusted entirely ?

Go to last post
F

fotografo

Member
Messages
47
Reaction score
0
Location
Ottawa, CA
I'm testing my new Rebel Xt with the conditions to find possible dead/hot pixels. After a series of tests, I'm concentrating with the settings which are likely to capture more (ISO = 1600, exposure = 30 seconds).

Used Raw files converted to 2 formats:

Exif-TIFF (8 bits per channel)
TIFF (16 bits per channel)

6 tests done in similar conditions are giving results not entirely consistent:

Test # 1: Exif-TIFF, viewfinder eyepiece not covered: 70 hot pixels

Test # 2: Exif-TIFF, just a piece of paper over viewfinder eyepiece: 71 hot pixels, 0 dead

Test # 3: Tiff: eyepiece cover used: 88 hot pixels and 1 dead pixel (!!!)

Test # 4: Tiff (same settings above, to confirm dead pixel): 76 hot pixels, no dead pixel this time

Test # 5: Tiff: Set myself inside a dark closet, to ensure complete darkness put camera inside a black cloth and shoot: 75 hot pixels, 0 dead

Test # 6: Same original RAW as Test # 5 converted to Tif with 8 bit/ch: 75 hot pixels, 0 dead

So, statistics show that there are some pixels not good, about 75, but in one case a pixel was given as "dead". Was it really dead if shown only once on that state ?

I still can return this camera, but this would be the second time (the first one I returned just because the sensor was dirty, although at the time I made no tests with the program "DeadPixeltest").

I also know that there is no such thing as a perfect sensor, although I have seen people getting better results. I don't worry too much for the hot pixels, but I definitely worry about the pixel pronounced dead in one of the tests. I have till Friday to return this camera, after that I have to use Canon Technical Service if I want this fixed or improved (I understand they do not change the sensor, just do a "remapping", sounds liek "revamping"...).

Opinions, anyone ?
 
fotografo said:
Used Raw files converted to 2 formats:
Click to expand...
are you sure that the converter doesn't screen some of them away (there are reports that a pixel was dead with one corverter, but disappeared when using Adobe's converter).
fotografo said:
I also know that there is no such thing as a perfect sensor,
Click to expand...
there isn't.
fotografo said:
although I have seen people getting better results. I don't worry
too much for the hot pixels, but I definitely worry about the pixel
pronounced dead in one of the tests. I have till Friday to return
Click to expand...
after a year, you'll have 10 of them. so why bother? dead pixels go away as soon as you resize, are not visible in prints and nobody notices them at full resolution on screen (owner excluded).
fotografo said:
this camera, after that I have to use Canon Technical Service if I
want this fixed or improved (I understand they do not change the
sensor, just do a "remapping", sounds liek "revamping"...).
Click to expand...
they designate the pixel as dead, so the camera ignores information retrieved at this pixel site and simply uses neighbooring pixels alone to calculate the final result at this pixel site. it's probably cheaper for canon to show the middle finger to people who want new sensors, than to actually give them new sensors.
fotografo said:
Opinions, anyone ?
Click to expand...
i'd learn to not care. like with viewfinder dust, rattling flash guide rails, supposedly "plastic-like, non-durable" exterior, and just shoot pictures with it.

--

Nekdo je moral Josefa K. o'crniti, zakaj ne da bi bil storil kaj slabega, so ga nekega jutra prijeli.

 
Jurc:

I guess you're right ... It's just kind of psychological thing, because I still can exchange or return this camera this week, and if I can get someting better, I'll try. But the problem is, I could not possibly know if the next camera will have a better sensor: it might be worse. Or the sensor could be a little better, but then something else maybe not working properly and in the last camera it was.

I just made a new test using ISO = 100 with 30 seconds and got 4 hot pixels. So definitely the ISO is the main factor here. Also, faster exposures reduce this (with ISO = 100 and 1 second I got no hot pixels).

I'm converting to TIFF using the program that came with the camera ("Raw Image Task"). Where I can find the "Adobe converter" you mentioned ? It might be worth comparing with that.

Thanks for your valuable input on this issue.

Cheers,

Antonio (a.k.a. fotografo)
 
I know its been said, but how the heck do you find these dead pixels? I shot a frame at these specs and its one great big black frame. I scroll around in PS, but I keep getting lost. How do you all do this? Or am I just too impatient?
 
There is a program called "DeadPixeltest" you can apply to your TIFF files, which is FREE and can be obtained from:

http://www.starzen.com/imaging/deadpixeltest.htm

There is a file with instructions on it. You have to shoot several images with the lens cap on, lens set to manual mode (otherwise it cannot focus in the dark and the photo is not taken), and dial mode set to Manual. People usually shoot several photos using a combination os ISO/exposure, from 1 sec to 30 sec.

You shoot in RAW format, and use the program "Digital Photo Professional" which comes with the Rebel XT, or "Raw Image Task", usually accessed though "Zoom Browser EX", menu "Edit ~ process RAW Images...".

Then save the file in TIFF format, because "DeadPixeltest" program cannot use RAW files. Run it, and it will show how many dead or hot pixels you have and its coordinates. using those coordinates you can locate where they are. I just opened the TIFF image with Photoshop enlarged it to maximam (1600%) and sacan it viusally. Since the hot or dead pixel are against a black background it won;t take lon for you to spto those.

However, as people have pointed out already,you can get many hot pixels with a 30 seconds exposure and ISO 1600. Woth lower ISo and faster shutter speed, hot pixels diminish dramatically.

You can alos take a look at another thread which discussed this issue at length (most of my knowledge comes form that thread):

Bad pixel survey
 
Oops, didn't have the time to correct spelling mistakes, before it was posted. I'll do it now:

-----------------------------------------------------------------------------------------
There is a program called "DeadPixeltest" you can apply to your
TIFF files, which is FREE and can be obtained from:

http://www.starzen.com/imaging/deadpixeltest.htm

There is a file with instructions on it. You have to shoot several
images with the lens cap on, lens set to manual mode (otherwise it
cannot focus in the dark and the photo is not taken), and dial mode
set to Manual. People usually shoot several photos using a
combination os ISO/exposure, from 1 sec to 30 sec.

You shoot in RAW format, and use the program "Digital Photo
Professional" which comes with the Rebel XT, or "Raw Image Task",
usually accessed though "Zoom Browser EX", menu "Edit ~ Process RAW
Images...", with the desired file selected.

Then save the file in TIFF format, because "DeadPixeltest" program
cannot use RAW files. Run it, and it will show how many dead or hot
pixels you have and its coordinates. using those coordinates you
can locate where they are. I just opened the TIFF image with
Photoshop enlarged it to maximum (1600%) and scan it visually.
Since the hot or dead pixel are against a black background it won't
take long time until you spot those.

However, as people have pointed out already, you can likely get many hot
pixels with a 30 seconds exposure and ISO 1600. With lower ISO and

faster shutter speed, hot pixels diminish dramatically. Personally I do not shoot a lot under these conditions, so it's not too much to worry.

You can also take a look at another thread which discussed this issue at length (most of my knowledge comes from that thread): "Bad pixel survey".

Hope this helps !

Antonio

---------------------------------------------------------------------------------------------
fotografo said:
There is a program called "DeadPixeltest" you can apply to your
TIFF files, which is FREE and can be obtained from:

http://www.starzen.com/imaging/deadpixeltest.htm

There is a file with instructions on it. You have to shoot several
images with the lens cap on, lens set to manual mode (otherwise it
cannot focus in the dark and the photo is not taken), and dial mode
set to Manual. People usually shoot several photos using a
combination os ISO/exposure, from 1 sec to 30 sec.

You shoot in RAW format, and use the program "Digital Photo
Professional" which comes with the Rebel XT, or "Raw Image Task",
usually accessed though "Zoom Browser EX", menu "Edit ~ process RAW
Images...".

Then save the file in TIFF format, because "DeadPixeltest" program
cannot use RAW files. Run it, and it will show how many dead or hot
pixels you have and its coordinates. using those coordinates you
can locate where they are. I just opened the TIFF image with
Photoshop enlarged it to maximam (1600%) and sacan it viusally.
Since the hot or dead pixel are against a black background it won;t
take lon for you to spto those.

However, as people have pointed out already,you can get many hot
pixels with a 30 seconds exposure and ISO 1600. Woth lower ISo and
faster shutter speed, hot pixels diminish dramatically.

You can alos take a look at another thread which discussed this
issue at length (most of my knowledge comes form that thread):

Bad pixel survey
Click to expand...
 
Thank-you. Sometimes it feels like evrybody else knows what they are talking about except me. This explains alot. I had gotten the gisdt of how to take the frame, but never figured out how to locate the dead pixels. I scanned around in PS but never found any white dots. It was all black.
 
fotografo said:
Jurc:

I guess you're right ... It's just kind of psychological thing,
because I still can exchange or return this camera this week, and
if I can get someting better, I'll try. But the problem is, I could
Click to expand...
i agree that this is a psychological thing, that's why i said there is no real reason go be dealing with this. one has to let it go :)
fotografo said:
not possibly know if the next camera will have a better sensor: it
might be worse. Or the sensor could be a little better, but then
something else maybe not working properly and in the last camera
it was.
Click to expand...
that's one good reason to stick to a camera that has the mechanics working properly.
fotografo said:
I just made a new test using ISO = 100 with 30 seconds and got 4
hot pixels. So definitely the ISO is the main factor here. Also,
faster exposures reduce this (with ISO = 100 and 1 second I got no
hot pixels).
Click to expand...
well, of course. this is absolutely predictable. the higher the iso, the more amplification of signal that come off the sensor. with this, the noise get's amplified, slightly hot pixel get even hotter. the same with longer exposure - the longer it operates, more time for defects to express themselves.
fotografo said:
I'm converting to TIFF using the program that came with the camera
("Raw Image Task"). Where I can find the "Adobe converter" you
mentioned ? It might be worth comparing with that.
Click to expand...
i think it's called ACR, there should be plenty of mentioning it on this forum. i use RawShooter Essensials 1.1.1. at the moment.

--

Nekdo je moral Josefa K. o'crniti, zakaj ne da bi bil storil kaj slabega, so ga nekega jutra prijeli.

 
In fact, it is difficult to find less relevant test for a camera than making a 30 sec exposure with ISO1600 of a lens cap.

If you do not see hot pixels in your everyday shots then there are none.

--
Regards

alexeig

http://www.pbase.com/alexeig
 
Jurc:

Indeed, your comments are right to the point.

The hot pixels only will show on long exposure and high ISO, which are the conditions we all normally don't shoot 90% of the time and up.

When I had a film camera I made a lot of night shots, and they were really bad, even with tripod, and never questioned the camera but my ability. The images were inferior to the shots we can make with the Canon Rebel XT or other digicams, no matter how many hot pixels we get.

The only reason I made all those tests is because I'm a few days of the time window I could return or exchange the camera, and I don't want to miss that deadline without doing my homework. I believe more people in this forum might have similar thoughts when dealine approaches (Should I keep this or try another one ?). addmittedly, it is mostly a psychological thing. I'm pretty sure when dealine is over, I won't even think on it again !

Just for reference, follows the results of my tests.

ISO 100 200 400 800 1600
Speed
1/30' 0 0 0 0 0
1” 0 0 2 6 5
30” 3 9 9 30 70/71/88/76/75

Since I have a Canon Rebel XT, I used the "Raw Task Converter", to save Raw files as TIFF (16 bits). I don't have the 'Adobe RawConverter' nor "file Viewer Utility' used by other people (I guess they have the DRebel). I made more tests at 30 sec. and ISO 1600. Got less hot pixels (70/71) with TIFF-8 bits, even with eye viewfinder not covered. The 88 hot pixels may be explained because I just recharged the battery before that test, and it was hotter (temperature counts).

Thanks for your accurate comments.

Antonio
 
I agree with that. Most of the time we don't shoot @ ISO 1600 with 30 secs. I admit I did this (annoying !) tests because I still can return or exchange the camera without any discussion with store clerks, but next week I could not. Deadline triggers a psychological anxiety, because you think "Hmmm ... for the same money I paid I could have a better one". The problem ?
I can not know in advance if the next one will be better ... :-(

Cheers,

Antonio
 
I did a really quick test: All at iso 1600
1/60 0/7
1/100 0 Dead/7 Hot
5 sec 0/11
30 sec 3 dead/ 98 hot

Should I worry about that 300 second exposure? (despite the previous post saying not to worry)
 
My guess is that 98 hot pixels is pretty normal. The 3 dead might be a concern, but try to repeat the test, because I also got 1 dead pixel which didn't show in similar tests afterwards. If you get consistently the 3 dead pixels, AND you are still in the return window, consider exchanging it. But as other people (and myself) have said, you cannot know if the next camera would be better, because the sensor is not the single part that can be wrong. I also understand that later on (BEFORE warranty expires), you can ask Canon technical service to "re-map" dead pixels, which means, they use adjacent pixels readings for dead ones. Also, appears that even if you don't have dead/hot pixels now, you can have those appearing later on.

The only thing for sure: Dead/Hot pixels are to be lived with in the digital world.

Antonio
 
slatchley said:
I did a really quick test: All at iso 1600
1/60 0/7
1/100 0 Dead/7 Hot
5 sec 0/11
30 sec 3 dead/ 98 hot
Click to expand...
It looks like the hot pixel threshhold is set so low that at iso 1600 the read noise of about 7 pixels is high enough to be called hot. Try using iso 400 instead or increase the hot pixel threshhold.
slatchley said:
Should I worry about that 300 second exposure? (despite the
previous post saying not to worry)
Click to expand...
No, the program reports very hot pixels as bad- they are not bad, just "very hot" caused by the very long + very high iso exposure (or possibly an alpha particle hit- which becomes more likely with long exposures).
 
This shot is perhaps the closest real life shot that the test would be good for (taken iso1600, 30sec)



By the way the hot pixels were automatically removed in this image by the RAW processing software I used.

But I do agree it is generally a useless test. One I haven't done - either you have a problem if they show up in 'everyday' type shots, or you don't and they only show up in very long low iso exposures or moderately long higher iso shots.

Al
 
noisebeam:

Could you tell what is the Raw Application you used to remove them ?

Not sure if "Raw Image Task" or "Digital Photo Priofessional" that came with the Rebel XT can do that. BTW, what is the yellow area in your image (below) ? Just curious ...

Thanks in advance.

Antonio
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top