Canon 500D stuck/hot pixels fix

Hi everyone,

I've had my 500D for many years now and it is a great bit of kit. I've recently been taking more night/long exposure shots and have noticed a large increase in the number of stuck or hot pixels that appear as white, red or blue dots on my shots which are very noticeable on dark backgrounds.

I tried a setting recommended by another user in these forums to switch on 'Long exposure noise reduction' under the 'Custom Functions' menu setting. This does appear to remove or fix the pixels in question, at least when I have tested it with the lens cap on (untested outside yet).

Why do these pixels appear? Is it an age of the sensor thing and generally unavoidable? I rarely shoot in RAW and wonder if I should in future, as editing jpeg files with all these busted pixels is rather time consuming. See attached images for examples.

Recent night shot.

Long exposure lens cap on, ISO:200

Long exposure, lens cap on, noise reduction custom setting on.

Thanks for any help or thoughts!

Cheers,

R.

In all honesty, I do not see anything really abnormal here. I took some long exposures at night in January and I ended up with a similar result than yours, especially in very long exposures (20-30 seconds).

Fortunately, these are quite easy to remove with post processing.

If you shoot in RAW and open the CR2 in Canon's Digital Photography Professional (DPP) a free program for Canon users... there is a slider for "Chromatic Aberration" which this might be and another for Reducing Color Moiré.  These are found under the "Detailed Adjustment Tool Palette"

Even new 70D and 80D cameras can do this during long exposures.  Definitely use the long exposure noise reduction.  This is a pain as the camera will take a 'dark frame' after each picture, the same length of time as the original, and then subtract the noise from the image.  It really is a pain if you shoot 2 and 3 minute exposures like I have done at times in bulb mode.  But it works.

One other thing, Canon has an undocumented feature that maps out dead pixels.  Put the body cap on, then set the camera for manual sensor clean, let that go for about a minute.  (This is the same setting you would use to open the shutter for swabbing the sensor).  I did this to fix a bad pixel on a brand new T2i awhile back.  It was so bad that the dead pixel showed up as a white spot - even on the LCD in live view.

karl mohr wrote:

Canon has an undocumented feature that maps out dead pixels. Put the body cap on, then set the camera for manual sensor clean, let that go for about a minute. (This is the same setting you would use to open the shutter for swabbing the sensor).

+1

ric_dep wrote:

karl mohr wrote:

Even new 70D and 80D cameras can do this during long exposures. Definitely use the long exposure noise reduction. This is a pain as the camera will take a 'dark frame' after each picture, the same length of time as the original, and then subtract the noise from the image.

One other thing, Canon has an undocumented feature that maps out dead pixels. Put the body cap on, then set the camera for manual sensor clean, let that go for about a minute. (This is the same setting you would use to open the shutter for swabbing the sensor).

That's interesting Karl thanks very much for your answer. I think any option that will mask these pixels before post is a good one. As you say, I don't think these are dead pixels, but I will try out the pixel mapping feature you mentioned.

Thank you all for your answers also!

I tried this technique but it didn't help me for long exposures.

bclaff wrote:

Some cameras will map out bad pixels if you perform one or two consecutive sensor cleanings; so that's worth a try.
If you haven't done it here's the page from the manual:

If you're curious to document your bad pixels, particularly before/after trying this technique, contact me by email or Private Message (PM).
I have a statistical technique to objectively report bad pixels.

FWIW, sensors do "age"; one culprit is cosmic rays, they are rare but do damage pixels.

The actual mapping function actually is very close to where you are, but instead of 'clean now' you should choose the menu choice just below that one 'clean manually'. This is the function that flips up the mirror and opens the shutter for an actual physical cleaning of the sensor with a moistened swab or some other method. Leave in this mode for 30 seconds to a minute. This will probably only map out a blatantly bad pixel - the long exposure 'spots' you are seeing probably won't be affected.

benjilafouine wrote:

ric_dep wrote:

karl mohr wrote:

Even new 70D and 80D cameras can do this during long exposures. Definitely use the long exposure noise reduction. This is a pain as the camera will take a 'dark frame' after each picture, the same length of time as the original, and then subtract the noise from the image.

One other thing, Canon has an undocumented feature that maps out dead pixels. Put the body cap on, then set the camera for manual sensor clean, let that go for about a minute. (This is the same setting you would use to open the shutter for swabbing the sensor).

That's interesting Karl thanks very much for your answer. I think any option that will mask these pixels before post is a good one. As you say, I don't think these are dead pixels, but I will try out the pixel mapping feature you mentioned.

Thank you all for your answers also!

I tried this technique but it didn't help me for long exposures.

I'm not sure if the 500D has this feature, but my 550D has 'long exposure noise reduction' under  custom functions in the menu - that's your best bet for this problem.

Not nearly so useful as mapping the bad pixels out since this doubles the exposure time for every image taken.

Unfortunately, long exposure noise reduction, is probably the best way to deal with this.  I know waiting for the 'black frame' is annoying and time consuming.  I also sometimes do 3 and 4 minute exposures.  As you could imagine waiting for the black frame is even more intense.  I have since started working with 2 cameras when I do this - alternating which camera is on the tripod.  (At least you can move the camera when the black frame is exposing).

I was actually surprised when I first noticed this happening on a brand new 70D during 20 and 30 seconds daytime exposures.  I don't think there's anything wrong with your camera - it's just a fact of life.  Maybe going full-frame might help, my friend with a 5D3 doesn't complain about this.

As far as the pixel mapping, I fixed my new-in-box T2i when I first got it.  This is a heavily cropped 'before' image fresh out of the box....

bclaff wrote:

karl mohr wrote:

Not nearly so useful as mapping the bad pixels out since this doubles the exposure time for every image taken.

Unfortunately, long exposure noise reduction, is probably the best way to deal with this. I know waiting for the 'black frame' is annoying and time consuming. I also sometimes do 3 and 4 minute exposures. As you could imagine waiting for the black frame is even more intense. I have since started working with 2 cameras when I do this - alternating which camera is on the tripod. (At least you can move the camera when the black frame is exposing).

I was actually surprised when I first noticed this happening on a brand new 70D during 20 and 30 seconds daytime exposures. I don't think there's anything wrong with your camera - it's just a fact of life. Maybe going full-frame might help, my friend with a 5D3 doesn't complain about this.

As far as the pixel mapping, I fixed my new-in-box T2i when I first got it. This is a heavily cropped 'before' image fresh out of the box....

I strongly disagree.

Long Exposure Noise Reduction (LENR) is not only time consuming but you're likely to get a black pixel at the "bad" pixel location.
Whereas mapping is likely to give you the average of nearby pixels; that's a more reasonable outcome.

Remember, the topic is "bad" pixels not noise from long exposures.

Those spots/blemishes of the OP's example photos look like noise to me. But by all means, the OP could send it to Canon for checking and mapping, doubt it would help much, but if LENR is too inconvenient he could try....