long exposure- noise reduction = amp glow

[neilp1]

hi all

Somehow, i am currently in a bit of an argument over on dpreview regarding an amp glow issue i have with my A570is, running long exposures with the noise reduction feature off. A dark frame taken at 15seconds, ISO400, NR off show a significant (to me) patch of purple amp glow in the upper left corner, visible on the LCD immediately after capture. I use these settings for astronomy photos, and was really excited when the NR off function became available, but for anything over 30 seconds at iso 200, my dark frame subtraction isnt giving the results id like.

Anyone else see amp glow on the camera LCD immediately after capture when using these settings on the A570? I have a feeling maybe my sensor is a little on the 'bad' side of average but would like to verify.

for clarification, this other chap contests the finding that the amp glow is visible on the camera LCD. he maintains that i *must* be post processing... and it has now developed into a bit of an argument. anyway, the thread isnt hard to find.

some pictures-

Image 1 is the 15 second iso 400 capture of a sporadic meteor (completely by accident, was testing how much city light pollution i had from my location)

http://picasaweb.google.com/neilp2006/Ampglow/photo#5140268217080895682

Image 2 is a 15 second dark frame taken using a alu foil lens cap, in the field about 3 minutes after IMAGE 1 was taken


http://picasaweb.google.com/neilp2006/Ampglow/photo#5140268217080895698


Image 3 is image 1 with dark frame image 2 subtracted.


http://picasaweb.google.com/neilp2006/Ampglow/photo#5140572506923882738


according to this other guy, image 2 is post processed by me, then i posted it online and lied about its authenticity for no reason at all.

anyone fancy running a quick test and letting me know how noticeable the amp glow is?

cheers

[intrnst]

OK, no panic.
There is a Purple Alien living in your cam's corner!

Serious, NR by frame-subtraction inside the cam is useless.
(devs, sorry about that)

See:

• Dark-frame at chdk.wiki

• Abana opinion at forums.dpreview

• Divalent end of story at forums.dpreview
  

You might try this PP approach:

• http://www.mediachance.com/digicam

[neilp1]

OK, no panic.
There is a Purple Alien living in your cam's corner!

Serious, NR by frame-subtraction inside the cam is useless.
(devs, sorry about that)

Cheers for the reply


Yeah, i know, thats why i turned the NR off. I have a program for generating images of astronomy subjects that does the dark field subtraction out of the camera. But surely i shouldnt be seeing this amount of purple amp glow noise under these conditions. i would expect some, which is why the dark fields are done (also the hot and dead pixels) but im kinda concerned about the amount im seeing.

Just looking at image 2, the dark-field exposure in the upper left... thats too much noise, right? or am i expecting too much.

If you look at image 3, you can see an artifact where the software has gone too far with the correction and made that part black, whereas the rest of the image background is grey tone.

[urbient]

hey .... ihad the same problem with my canon A640 ... 15 seconds in iso 800 at night ... i did shots of jungle in moonlight and i had the purple stuff on both left corners ... i would even think of the lightleak as a reason as the people stated its also a purple color thing ... maybe ure cam has a gap in the case that lets some light from a side source inside ? ... its really difficult to find out whats the problem i think ....

good luck

chris

[neilp1]

Hey

Thanks for the post.

I dont suppose you have any of those pics available, do you?

Im guessing that i wont exactly be able to send it back to canon for a replacement if the only way to get the effect is through a software hack that we arent supposed to even use.

i put the camera on the 10 second timer (15sec, iso400, foil lens cover on, noise reduction OFF) and put it in box in a drawer at work.... unfortunately, its still there and to the same extent. So its looking ike its not a light leakage issue. To bad, because that would have been a nice easy fix

thanks again for you reply

[intrnst]

i put the camera on the 10 second timer (15sec, iso400, foil lens cover on, noise reduction OFF) and put it in box in a drawer at work.... unfortunately, its still there and to the same extent. So its looking ike its not a light leakage issue. To bad, because that would have been a nice easy fix

Hum... my previous guess was at the leakage path approach.
But, have you saw this thread before? Maybe it helps.

• Long time exposures without purpe corner at dpreview

[barberofcivil]

I've been having a look at this.  A couple ideas.

I know you have said that the exposures were both unprocessed and the same exposure time.  The dark frame reduction seems to work but a bit too much, like it is more exposed than the original.

1) Are all your other settings (that could affect exposure) the same (AV, white balnce, etc.)?
2) Since the glow is purple, it seems to indicate that perhaps it's not a light leak, but rather from heat concentrations around the CCD.  If this is the case, even with identical settings for everything, maybe what is actually causing the glow isn't the same between the original and the dark frame shot (i.e., warmer or cooler enough to make a visible difference, especially at long exposure times.

3) A third thought just occured to me.  Seeing as it's purple, perhaps it is coming from light inside the camera in or close to the ultraviolet spectrum.  I don't know how sensitive the CCDs are to wavelengths outside the visible spectrum.

Do you have your exact settings for this?  I have an A570 at work, but I haven't been able to get the glows, and would like to see if it can be replicated.

[neilp1]

I've been having a look at this.  A couple ideas.

I know you have said that the exposures were both unprocessed and the same exposure time.  The dark frame reduction seems to work but a bit too much, like it is more exposed than the original.

1) Are all your other settings (that could affect exposure) the same (AV, white balnce, etc.)?
2) Since the glow is purple, it seems to indicate that perhaps it's not a light leak, but rather from heat concentrations around the CCD.  If this is the case, even with identical settings for everything, maybe what is actually causing the glow isn't the same between the original and the dark frame shot (i.e., warmer or cooler enough to make a visible difference, especially at long exposure times.

3) A third thought just occured to me.  Seeing as it's purple, perhaps it is coming from light inside the camera in or close to the ultraviolet spectrum.  I don't know how sensitive the CCDs are to wavelengths outside the visible spectrum.

Do you have your exact settings for this?  I have an A570 at work, but I haven't been able to get the glows, and would like to see if it can be replicated.

Hi

Yes, the exact settings for the shots are as follows-

Manual, 15sec, f2.6, iso 400, white balance is at cloudy. Noise reduction OFF. High quality JPEG. Thats all there is to it. Both the actual shot and the dark frames were taken at the same time, within 5 seconds of each other. i use the custom timer, taking 2 shots 5 seconds apart (enough time so i can get the cap on).

i can see the purple glow on the LCD immediately after capture, so it definately isnt a processing artifact.

Let me know how you get on.

If you try it at 65sec with 400-800 iso, youll definately see it if your camera is functioning the same as mine. If you dont, then it only leads me to the conclusion that  maybe i got a bad apple... 

[barberofcivil]

I've been having a look at this.  A couple ideas.

I know you have said that the exposures were both unprocessed and the same exposure time.  The dark frame reduction seems to work but a bit too much, like it is more exposed than the original.

1) Are all your other settings (that could affect exposure) the same (AV, white balnce, etc.)?
2) Since the glow is purple, it seems to indicate that perhaps it's not a light leak, but rather from heat concentrations around the CCD.  If this is the case, even with identical settings for everything, maybe what is actually causing the glow isn't the same between the original and the dark frame shot (i.e., warmer or cooler enough to make a visible difference, especially at long exposure times.

3) A third thought just occured to me.  Seeing as it's purple, perhaps it is coming from light inside the camera in or close to the ultraviolet spectrum.  I don't know how sensitive the CCDs are to wavelengths outside the visible spectrum.

Do you have your exact settings for this?  I have an A570 at work, but I haven't been able to get the glows, and would like to see if it can be replicated.

Hi

Yes, the exact settings for the shots are as follows-

Manual, 15sec, f2.6, iso 400, white balance is at cloudy. Noise reduction OFF. High quality JPEG. Thats all there is to it. Both the actual shot and the dark frames were taken at the same time, within 5 seconds of each other. i use the custom timer, taking 2 shots 5 seconds apart (enough time so i can get the cap on).

i can see the purple glow on the LCD immediately after capture, so it definately isnt a processing artifact.

Let me know how you get on.

If you try it at 65sec with 400-800 iso, youll definately see it if your camera is functioning the same as mine. If you dont, then it only leads me to the conclusion that  maybe i got a bad apple... 

Good news (or is it bad?)  I was able to replicate it on my A570, it is also present on my S2 (though it isn't as bright with the same settings it seems), both in the same location, a bright one in the top left corner and one much less visible near the top center.  It seems that this may be related to the CCD itself and not the camera model.

One interesting thing I found, is that the internal dark frame subtraction virtually removes it, so it still seems that there is some slight exposure difference between your image and manual dark frame for some reason.

I have no idea what is causing it, but I would tend to think it is internal to the camera, either an internal light (not likely) or thermal sensitivity of some sort (or even UV light from something is possible to).

[neilp1]

Good news (or is it bad?)  I was able to replicate it on my A570, it is also present on my S2 (though it isn't as bright with the same settings it seems), both in the same location, a bright one in the top left corner and one much less visible near the top center.  It seems that this may be related to the CCD itself and not the camera model.

One interesting thing I found, is that the internal dark frame subtraction virtually removes it, so it still seems that there is some slight exposure difference between your image and manual dark frame for some reason.

Yeah, i never noticed it with the standard pre-chdk 15 second exposures. Only with the NR turned off. But thats to be expected. Ill just dark frame subtract out of the camera with the software i have.

I have no idea what is causing it, but I would tend to think it is internal to the camera, either an internal light (not likely) or thermal sensitivity of some sort (or even UV light from something is possible to).

good/bad news?? hmm, depends... good that it can be replicated under the same conditions, but kinda bad for the astrophotos that id like to do with it.

Although that kind of imaging is 3rd down the list so its tolerable. I have tried some software solutions for removing it, and have got a pretty good PS action that i made that can be used to get rid of it. Doing a normal darkframe subtraction followed by a little more PS and its pretty much taken care of. Cant really argue with it, since its a $140 camera, performing well out of its documented specification range. 

Thanks for verifying it for me. You do see this in the LCD screen right after capture, correct? This was the main reason i got into the discussion with the other chap- he maintains that seeing it on the LCD is 'impossible'.

Cheers!

[barberofcivil]

Good news (or is it bad?)  I was able to replicate it on my A570, it is also present on my S2 (though it isn't as bright with the same settings it seems), both in the same location, a bright one in the top left corner and one much less visible near the top center.  It seems that this may be related to the CCD itself and not the camera model.

One interesting thing I found, is that the internal dark frame subtraction virtually removes it, so it still seems that there is some slight exposure difference between your image and manual dark frame for some reason.

Yeah, i never noticed it with the standard pre-chdk 15 second exposures. Only with the NR turned off. But thats to be expected. Ill just dark frame subtract out of the camera with the software i have.

I have no idea what is causing it, but I would tend to think it is internal to the camera, either an internal light (not likely) or thermal sensitivity of some sort (or even UV light from something is possible to).

good/bad news?? hmm, depends... good that it can be replicated under the same conditions, but kinda bad for the astrophotos that id like to do with it.

Although that kind of imaging is 3rd down the list so its tolerable. I have tried some software solutions for removing it, and have got a pretty good PS action that i made that can be used to get rid of it. Doing a normal darkframe subtraction followed by a little more PS and its pretty much taken care of. Cant really argue with it, since its a $140 camera, performing well out of its documented specification range. 

Thanks for verifying it for me. You do see this in the LCD screen right after capture, correct? This was the main reason i got into the discussion with the other chap- he maintains that seeing it on the LCD is 'impossible'.

Cheers!

No problem!  Yes it was visible on the LCD after taking the photo.  The problem also seems to not just be limited to the A570, the S2 is (was) a higher end camera in it's day, so it's not just the cheaper cameras.

Can anybody else replicate on other models?

Perhaps a good addition to the CHDK would be to add an option for darkframe where it is saved as an image but not applied.  This way you could easily get a darkframe matching the original (without the need for a lens cap even) but apply it manually if you'd like.  Perhaps I'll add this to the request board.

[Kalli]

Hi
A620: purple field in top left corner,    height: ~800 pixel, width: ~155 pixel, shutter at 65".

I suppose a heat source might be located in the camera close to that erea (resistance, IC...)

[GrAnd]

Hi
A620: purple field in top left corner,    height: ~800 pixel, width: ~155 pixel, shutter at 65".

I suppose a heat source might be located in the camera close to that erea (resistance, IC...)

Flash?

Or the picture on the sensor is 180 degrees rotated? Then it can be battery supply IC.

[barberofcivil]

Hi
A620: purple field in top left corner,    height: ~800 pixel, width: ~155 pixel, shutter at 65".

I suppose a heat source might be located in the camera close to that erea (resistance, IC...)

Flash?

Or the picture on the sensor is 180 degrees rotated? Then it can be battery supply IC.

I think neither.  The location is pretty much the same for both my A570 and S2 which have completely different bodies, the S2 has a pop-up flash.  The battery compartment seems somewhat removed from the CCD location.  There is also the issue of some glow at the top middle.  I could only see this by bumping up the brightness of the images.

[neilp1]

cheers guys!

I too only see the came in the centre top of the pic if i enhance levels, but the one in the left is there even without processing. Id love to see an internal schematic of the camera 'wiring'... im pretty sure its a component issue, likely the amplifier. I had a hardwire modded 300d that had the exact same purple glow in the same location before the mod, which involved relocating the amp about an inch further away fromt he sensor. Post mod had no glow in that area.

Thanks all to have verified this, i was getting a little paranoid it was just me. Post processing gets rid of it pretty good, so im not too concerned since it seems my specific camera is performing similarly to others.