Amp GLow on A570 cameras...

Started Dec 4, 2007 | Discussions
neilp1 New Member • Posts: 15
Amp GLow on A570 cameras...

Now that it is possible to do long exposures (65 seconds) with the in camera noise reduction turned off (Chdk builds), has anyone else noticed just how bad the amp glow is on these powershot cameras?

I occasionally use these features for taking pictures of meteors, or star trails, and have noticed considerable amp glow in the top left of the ccd. With a dark frame subtraction, i can take it away, but it is so extensive that i fear that some real signal in this area is being reduced or even eliminated. in fact, it so over-corrected in the top left that the background is going black

Anyone fancy posting some dark frame pics at 65 sec (lcd off) to see how bad my results are in comparison? I would very much appreciate any help in figuring out if my camera is 'normal' or is doing pretty badly.

cheers

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

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

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

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

Which of any of these images have you processed?

I think you will find that ANY camera's dark-frame at ANY shutter speed will show amp-glow effects like you are experiencing. During normal editing operations like histogram-equalize, auto-correction, or any tool that's trying to find a white-point automatically. You are, in effect, amplifying the low-level warmer pixels into values that stretch into the median values.

I think that you'll find if any part of your image are properly exposed that any corrections done won't try to stretch those low-level warm pixels into values that you can see.

On one of my digital cameras I find an interesting pattern (not Canon, an older camera), that when taking 30-second exposures of a black-frame, and I do a histogram-equalize on it, I can actually see several chips in their exact location and shape on the opposite side of the board to which the CCD is mounted, just from the minor warmth they generate. None of which interferes when using the black-frame normally without any correction done to it. Nor do those warm areas show up during actual use for long exposures. Only when they are intentionally (and wrongly) amplified will you even know they are there.

To let you know, I just did a similar thing with a black-frame from an S3 IS at 65" exposure. Without any processing done to it it looks perfectly black and normal. Just the meager peppering of warm pixels in random locations. BUT, if I do any auto-correct to it, any histogram-stretch, histogram-equalize, auto-white-point, etc. I get almost the exact same pattern that you are seeing. I expected that and I'm not surprised, because I know it means absolutely nothing during real use. Anyone can turn up the sound from their high-quality stereo so loud that there's nothing but distortion and noise throughout. That's what you have done. I only did it because I wanted to see if I can see any chips on the board behind the CCD, sadly, none show up. That would have been fun. And again, it wouldn't have bothered me in the least. Because I know in real use, they have no effect.

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

Hi

Thanks for your reply

The only one of those images that has been 'processed' in any way is the 3rd of the set, which is image 1 with the dark frame (image 2) subtracted. Images 1 and 2 are exactly as seen on the camera LCD immediately after capture... and definately before the images even get close to an image manipulation software. Noi other adjustment was made to image 3, no levels adjustment or curves etc.

RobHeath wrote:

Which of any of these images have you processed?

I think you will find that ANY camera's dark-frame at ANY shutter
speed will show amp-glow effects like you are experiencing. During
normal editing operations like histogram-equalize, auto-correction,
or any tool that's trying to find a white-point automatically. You
are, in effect, amplifying the low-level warmer pixels into values
that stretch into the median values.

I think that you'll find if any part of your image are properly
exposed that any corrections done won't try to stretch those
low-level warm pixels into values that you can see.

not quite sure about this, since i havent adjusted either of image 1 or 2...

None of which interferes when using the black-frame

normally without any correction done to it. Nor do those warm areas
show up during actual use for long exposures.

well, these are 'actual use' examples of long exposures...

Only when they are

intentionally (and wrongly) amplified will you even know they are
there.

I understand the point you make about making exposure adjustments and how that will stretch out the signal to the median range, but these images are before manipulation. I was very surprised to see this amount of amp glow visible on the camera LCD which is why i brought it up.

To let you know, I just did a similar thing with a black-frame from
an S3 IS at 65" exposure. Without any processing done to it it looks
perfectly black and normal. Just the meager peppering of warm pixels
in random locations. BUT, if I do any auto-correct to it, any
histogram-stretch, histogram-equalize, auto-white-point, etc. I get
almost the exact same pattern that you are seeing. I expected that
and I'm not surprised, because I know it means absolutely nothing
during real use. Anyone can turn up the sound from their high-quality
stereo so loud that there's nothing but distortion and noise
throughout. That's what you have done.

Nope, no levels adjust, curves or brightness etc. exactly as it come from the camera

Your 65" black frame example above- This was done with the noise reduction turned off? If so, this finding only reinforces my feleing that i have a hyperactive amp glow problem

thanks again for your reply

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

neilp1 wrote:

Hi

Thanks for your reply

The only one of those images that has been 'processed' in any way is
the 3rd of the set, which is image 1 with the dark frame (image 2)
subtracted. Images 1 and 2 are exactly as seen on the camera LCD
immediately after capture... and definately before the images even
get close to an image manipulation software. Noi other adjustment was
made to image 3, no levels adjustment or curves etc.

Impossible. If those were black frames (shot properly) right from the camera they would be black. Not gray. Let me guess, you are shooting in RAW and importing them in a RAW utility that does an auto-white-point balance.

not quite sure about this, since i havent adjusted either of image 1
or 2...

No, you may not have intentionally adjusted them. But whatever editor or method you used to get them from camera to posting on the net surely did.

well, these are 'actual use' examples of long exposures...

I'm sure they are. After the blacks have been wildly over-amplified.

Your 65" black frame example above- This was done with the noise
reduction turned off? If so, this finding only reinforces my feleing
that i have a hyperactive amp glow problem

I turn off noise-reduction permanently now after I see how much having it turned on damages images. (Shame on Canon for doing that. Probably just another method where they could convince people that they need to buy DSLRs. I smell a class-action lawsuit for installing noise-reduction that actually creates noise in images. How underhanded and sneaky can they get.)

Do this test again, only do it IN the camera. Shoot a dark frame, then look at it through the EVF. You won't see a hint of gray or any cast in the corner. This is how it will look before any processing done to it. Your images have been obviously processed.

Either that or you're using a really poor lens-cap that's letting in light. Do it in a dark room if you can. With a 65" exposures I can get nice clear images from the shadows in a dark room, light levels so low that I can't even see anything with my own eyes, and this is even when using ISO80. You need absolute black to do this test and to get any accurate black-frames for post-processing noise reduction.

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

One other question. Why on earth are you using a 65" exposure as your dark-frame on a 15" exposure? You don't do that. You use the same time exposure for the dark-frame as you do your original exposure. Otherwise the sensor noise from the two images won't match.

Applying a 65" dark-frame to a 15" image is like applying a dark-frame subtraction routine 4 times. Only worse because the same warm pixels in a 65" dark frame won't even show up in a 15" dark frame.

Just this alone lets me know that you're not quite familiar with what is going on and how to do it properly.

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

Ok, didnt intend on getting into a pi* ing contest over this, but anyways

IMAGE 2 is a 15 second dark frame.

IMAGE 1 is a 15 second frame of the sky at the back of my yard

IMAGE 3 is IMAGE 1 (actually about 15 x 15 second frames stacked to get the star trails) with dark frame subtracted.

Clear??

I never applied a 65 second dark frame to the 15 second exposure. I only mentioned the 65 second exposure because seeing the amount of amp glow i was getting at 15, inferred that it would be next to useless in terms of a high level of amp glow, with the specific camera i have

other points-

all shot in highest quality JPEG, noise reduction off. camera was outside cooling for 20 minutes. Ive been doing CCD imaging of the night sky for about 3 years, i know about cooling the sensor and the difference between processing artifacts, amp glow and hot pixels.

Dude, before you use words like impossible then imply i know nothing of what im talking about, put the brain into and maybe think that i am actually being truthful when i post.

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

Do this test again, only do it IN the camera. Shoot a dark frame,
then look at it through the EVF. You won't see a hint of gray or any
cast in the corner. This is how it will look before any processing
done to it. Your images have been obviously processed.

seriously dude, i already said these where this bad on the LCD. There is no grey in the dark frame (um, thatll be because its the dark frame) but i can very clearly see the purple amp glow at the top left corner. IN CAMERA

Either that or you're using a really poor lens-cap that's letting in
light. Do it in a dark room if you can. With a 65" exposures I can
get nice clear images from the shadows in a dark room, light levels
so low that I can't even see anything with my own eyes, and this is
even when using ISO80. You need absolute black to do this test and to
get any accurate black-frames for post-processing noise reduction.

Um, well, my dark frame is black, not grey, so im guessing my lens cap is good enough. But i still see the purple in the corner

The first 2 pictures have not even seen an image editing program. last time im going to say it.

How about, if all your going to do is bash and not trust anything i say, you maybe dont try to 'advise' me any further. I understand the concepts ive been doing, i understand the problem ive been looking at. Why would i lie about any of the steps ive taken? doesnt make sense, really. Just because you have never seen amp glow, in camera, with a dark field exposure, doesnt mean it doesnt happen. Sensors arent produced exactly alike- maybe i got a bad one that slipped through because canon applies automatic noise reduction and image adjustments and would never have tested the particular sensor in my particulr camera under the conditions i did.

Oh, and by the way, if you have the dark frame you say doesnt show any amp glow whatsoever, how about posting it. But hey, how can i trust thats its straight out of camera?? oh, i see, why would you lie, and why wouldnt i trust you?

touche?

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

Here you go. S3 IS, 65s exposure, noise-reduction turned OFF, downsized to 640x480 using Lanczos 8 for best detail retention, JPG compression 90% used for saving.

When comparing this against the original and a histogram equalize or other auto-correction tools are applied the pattern is very similar to the full-size image so little is lost in this down-sized one (for that testing purpose). All you'll see of warm-pixels is when you amplify them. Though I do admit, I lucked out and got an exceptionally quiet CCD chip in mine, luck of the draw.

Here it is with a quick "auto-correction" tool applied to save you the trouble of applying any "magnify the noise" tools yourself, but you are free to use any histogram or white-point tools yourself to see a similar image (albeit slightly different each time).

Now do you see why your images don't make sense to me?

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

RobHeath wrote:

Here you go. S3 IS, 65s exposure, noise-reduction turned OFF,
downsized to 640x480 using Lanczos 8 for best detail retention, JPG
compression 90% used for saving.

When comparing this against the original and a histogram equalize or
other auto-correction tools are applied the pattern is very similar
to the full-size image so little is lost in this down-sized one (for
that testing purpose). All you'll see of warm-pixels is when you
amplify them. Though I do admit, I lucked out and got an
exceptionally quiet CCD chip in mine, luck of the draw.

Here it is with a quick "auto-correction" tool applied to save you
the trouble of applying any "magnify the noise" tools yourself, but
you are free to use any histogram or white-point tools yourself to
see a similar image (albeit slightly different each time).

Now do you see why your images don't make sense to me?

i would if my darkframes looked so obviously blown out like yours do. take a closer look at mine and you will clearly see that those are nothing alike. your hot pixels are screaming, and they would also be screaming in mine if i had done any sort of image manipulation. But since i havent, they dont

Like you said, you lucked out an got a quiet chip in yours. Maybe i didnt and i got a noisy chip in mine.

Whats the ISO on those pics? I was talking about 400. At 80, the glow is a lot less, but still visible. Im guessing yours are at 80. Try one at 400.

I appreciate the help in sorting this out, but i got kinda defensive when it was implied that i dont know what im talking about, and worse, that i had lied about what i had posted.

Image 2 is straight out the camera. I can SEE the purple noise on the LCD. Oh, and i would never use the word impossible in a post... unless youve checked every ccd chip out there under those conditions and havent seen it, then you cant be sure enough to use 'impossible'

cheers

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

actually, after a quick try in PS... i messed with the levels until i got a picture which looks like yours... but damn, i still dont have as many hot pixels as you. So your quiet sensor response is kinda bunk.

cheers

http://picasaweb.google.com/neilp2006/Ampglow/photo#5140650104098019586

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

There's a difference between mathematically changing a pixel with a value of 5 to 255 as opposed to one of 35 (warm) to 255 (hot). You don't understand how auto-correction tools work, do you. They take the lightest possible pixel in an image and apply the math to turn that brightest spot on the whole sensor to 255, amplifying the rest accordingly. Those "hot pixels" as you like to call them have values all below 5, even after a 65 second exposure. You won't even find a sensor that quiet on most DSLRs. But go ahead, believe what you want if it'll make you feel better.

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

ok, taking it slow... i may not know exactly how auto correction algorythms work, but then, since none were applied, why is it relevant??

the hot pixel issue ha got nothing to do with auto correction or software... Its a hardware issue caused by a malfunctioning pixel. It going to be there ALL the time, increasing in intensity as time of exposure increases. Thats why darkframes are used to subtract them out, since they are in a defined place on the sensor every time you push the button. In the old days before software done it, it was done through a hardware route known as pixel mapping and bypass. go read about it.

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

RobHeath wrote:

You won't even find a
sensor that quiet on most DSLRs. But go ahead, believe what you want
if it'll make you feel better.

great! them maybe my sensor is actually better than i thought...

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

you might want to check this out

http://forums.dpreview.com/forums/readflat.asp?forum=1021&message=20953375&changemode=1

i particularly like post 8

i didnt find it before because the search function is off

enjoy the read

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

neilp1 wrote:

Im guessing yours are at 80. Try one at 400.

The previous one was at ISO 100 (65 seconds exposure). Considering that 65 seconds at ISO100 is 2 full stops shutter speed longer, and 2 full stops ISO speed shorter from 15 seconds ISO400 they're fairly equivalent.

Here's 15 seconds at ISO400. Just as you wanted to see, the same settings you used.

ISO800 at 65-seconds just barely starts to show some minor warmer areas with a few brightest values going up to 25. The main small patch in the corner peaking at the value of 12 (this is on a scale of 0-255). Nothing worth a worry, at all. Any dark-frame subtraction in any images taken with that will easily handle those values.

I'm glad you posted this and wanted test results from my camera. I had no idea that my sensor was this quiet, even at ISO800 and 65 second exposures. For a CCD chip this is admirable. This is going to make meteor recording and other astrophotography a breeze. Considering my older camera can easily record stars as dim as magnitude 8.5 with f/2.0, ISO200, 30-seconds, I'm guessing I can easily go another magnitude further than that with single exposures. If it wasn't -10 F. (-24 C.) outside right now I'd go and test it on some constellations and brighter nebulous regions.

You might want to see about getting a different camera.

RobHeath Forum Member • Posts: 59
Re: Amp GLow on A570 cameras...

neilp1 wrote:

you might want to check this out

http://forums.dpreview.com/forums/readflat.asp?forum=1021&message=20953375&changemode=1

i particularly like post 8

i didnt find it before because the search function is off

enjoy the read

Apples vs. Oranges : CCD vs. CMOS

OP neilp1 New Member • Posts: 15
Re: Amp GLow on A570 cameras...

RobHeath wrote:

I'm glad you posted this and wanted test results from my camera. I
had no idea that my sensor was this quiet, even at ISO800 and 65
second exposures. For a CCD chip this is admirable. This is going to
make meteor recording and other astrophotography a breeze.
Considering my older camera can easily record stars as dim as
magnitude 8.5 with f/2.0, ISO200, 30-seconds, I'm guessing I can
easily go another magnitude further than that with single exposures.
If it wasn't -10 F. (-24 C.) outside right now I'd go and test it on
some constellations and brighter nebulous regions.

thanks for that, im so happy for you. But my overall point at this stage was not how the camera was functioning (i can see the purple amp glow on the LCD at 15 sec and decided about 4 posts ago you were not interested in actually helping out with some constructive thoughts) but trying to figure out why you thought id lie and post PS'd pics as originals.

You might want to see about getting a different camera.

thanks again. Seeing how im more convinced my particular camera is on the noisy/hot end, i might just do that.

edited to remove some thoughts on internet forum etiquette and the dangers of flaming someone you know nothing about.

cheers

Keyboard shortcuts:
FForum MMy threads