Bright spots in infrared photos (Gimp processed)

(ron) #1

I have taken some near total darkness infrared photos with a Sigma SD14 (really like the camera even tho it’s pretty dated at 2007) :slight_smile:

I have the camera hooked up with an interval timer, taking one minute long exposures every other minute. The bright spots are scattered in the frame, and that scattering is different (different number, position, and direction of spots) in each of the interval timed frames.

So, what am I looking at? Note the photos have been cropped, enlarged, shrunk, and brightness/contrast adjusted, but nothing else. Here’s the shots:

(Edit:) got permission denied message, so here’s the shots on my site:

Link to the infrared shots

Late note: this has been resolved to be stray light, in an insufficiently darkened “dark room”.

(Morgan Hardwood) #2

Interesting, please upload a raw file. You can use if the forum doesn’t let you attach the file.

(ron) #3

Hi Morgan,

I had been taking jpegs, so here’s one of the original jpegs:

Link to brightspot IR photo

BTW, I’d never used before. I hope it’s OK :slight_smile:

I should add some info about the setup. The exposures are 1 minute long and controlled by an interval timer. I’m using an IR filter to block all visible light (720 nm cutoff). The internal IR filter has been removed from the camera, to enable IR. Additionally, I’m using an adjustable neutral density filter on top of the IR to adjust the exposure time optimally.

The lens is a 43mm Fujifilm, the ISO was set to 200, and the aperture to 3.5. The camera was set to bulb mode, and the lens to manual. The lens cleanliness is perfect (as close as can be).

  • Ron

(Hermann-Josef) #4

The bright pixels could be “cosmic ray” events in the detector. These are very common in (long-exposure) astronomical CCD-images. However, they need not be of “cosmic” origin. Slightly radioactive glasses, e.g. in the filter, can also cause these spots.

The number density should increase linearly with exposure time. The hot pixels come in just three colours: red, green, purple. Why purple and not blue, I do not understand. Does the camera use a Bayer-matrix? The hot spots are not always one pixel in size, so this is puzzling with this interpretation - if the Bayer filters are covering just one pixel of the detector.


(ron) #5

Hi Jossie,

The number density should increase linearly with exposure time.

Thanks for the suggestion. That adds another metric for me to use. I realized that I should be taking raw, to eliminate some oddity in the jpeg engine, although that doesn’t seem (intuitively) to be likely. The raws eat up my memory quickly on the interval setup, but I think I have to do some just to check.

The Sigma uses the Foveon sensor, which is not at all like Bayer. It measures photons by the depth penetration into the silicon. New fangled stuff that never really caught on (maybe soon tho).

(Hermann-Josef) #6


In this case it may well be that adjacent pixels are affected by one “hit”.

Just for fun, here is an example for severe cosmic ray hits in one of my HST images. At left the raw image, at right the processed sum of several images with cosmic ray hits removed:


(Thomas) #7

Did you take control shots with lens cover?

(ron) #8

Surely that should’a been my first thought, right? Thanks for the tip, I’ll go into the dark room and see what a lens cover shot does for me …

(Edit - )
I took both raw and jpeg images at 50, 100, 200 ISO with IR filter + ND + lens cap in the darkened room. There were zero bright spots in any of the shots.

(ron) #9

@jossie : Thanks for the interesting photograph of quasar jet. The star is assumed from position? How did you filter the cosmic noise?


(Hermann-Josef) #10


I am surprised about your result. Would have expected to see no difference. Is your darkroom better shielded than the place where you took the first images?

Concerning the HST-image:
No, the star is not assumed from position but as a real star “survived” the removal of the cosmic ray event.
The events are removed via a median filtered image as follows: One always takes several exposures with the telescope shifted by small amounts in between (called dithering). Then all the images are positionally aligned with one another (imagine a cube with each image in a layer. Viewed from above, all real objects are viewed at the same position.). For each pixel across all images (vertical direction in the stack) the median is calculated and gives the value for this pixel in the median filtered image. With the cosmic ray events being random in position and intensity, the median image will not contain the cosmic ray signals. Then each individual image is compared with this median image on a pixel by pixel basis. Pixels deviating more than a given threshold from the median are cosmic ray events and are replaced by the scaled median (scaled e.g. for exposure time). All cleaned images are finally added up to give the clean sum image. All this is standard astronomical image processing.


(ron) #11


I just looked at your profile. It’s a pretty neat thing to get the astronomical process description from an actual astronomer such as yourself. Thank you!

As far as the room goes, it has window covering that is very thick when drawn closed, and special valences that block the tops and sides of openings. On a bright day, it’s very close to pitch black inside the room (I cannot see my hand in front of my face, but maybe just hints of outlines around some room objects). The bright spot photos were all taken between 1:00 AM and 4:00 AM in the morning, which I assumed would be “dark enough” relative to anything that could get in from the outside.

I took care that all internal sources were eliminated (power strip lights and the like).

Is it possible that a few stray photons from starlight or moonlight could have the effects I’m seeing?

(Hermann-Josef) #12


When talking about shielding I was thinking of concrete etc. But the idea with the cosmic ray events seems to have been ruled out by your recent experiment.

Since you are observing in the IR it may well be that there still was an IR light source you are not aware of because you could not see it. This is, I think, what the experiment with the lens cap might tell us.


(ron) #13

All of this may just be an obvious result of my setup (in hindsight). But, it’s not something I’d know until I took all the tiny steps :slight_smile:

I guess I don’t know what really really low photon count photos look like. What is the IR content of starlight and moonlight?

(Hermann-Josef) #14


Moonlight is reflected sunlight and thus has its maximum in the green wavelength range. Starlight is much, much less intense and has various spectral distributions. But heat sources emit IR radiation. So a light bulb, although turned off, may emit IR radiation for some time, ect. etc.


(ron) #15

So, based on what you’re saying, the IR probably isn’t coming from celestial bodies.

I still don’t understand the results that I’m seeing. It seems that any stray photons would not be so random (completely different in every shot). And - they seem to be clustered, but not in the same spot or spots. There is very little motion at four in the morning.

(Hermann-Josef) #16

@Ronaldlees Good morning Ron,

for curiosity I also would like to understand what you are seeing. You are using a filter to cut-off visual light with wavelengths less than 720nm. On the other hand, the bright spots have colours of red, green and purple. If I understand correctly, the detector in the camera derives the colours from the penetration depth of the photons. So if you see a blue signal, there must have been blue photons – right? I would expect that IR-photons being much less energetic, would not penetrate as deeply as green or blue, thus one would expect just a red signal from any IR-radiation. Please correct me, if I am wrong.

So to make the finding very clear: In your dark room, you see bright pixels when the lens cap is removed. You see no bright pixels with the lens cap on – everything else being the same?

Yes, I would be very surprised, if the IR would come from a celestial body. Was the camera pointing upwards or against something in the immediate surrounding? What were you hoping to see in the IR-exposures?

In summary, it is extremely puzzling, that it is only widely scattered single pixels or small pixel groups with a signal. From a real object one would expect are more continuous distribution. Dark current can be excluded, since the spots disappear with the lens cap on.


(Morgan Hardwood) #17

Hi @Ronaldlees

Interesting discussion.

The reason I asked for a raw file is to be able to see the actual photosites to gain insight into the nature of the artifacts, to eliminate the source of the artifacts as being something post-capture. Would be great if you could reproduce the artifacts in a raw photo.

On a side note, I downloaded a sample Sigma SD14 raw file and while RT reads it correctly, it seems to lack white levels, raw crop and input profile. Would be great if you could follow these steps and send us sets 1 and 2:
Do you also have access to a color target?

(ron) #18

Hello again Morgan,

I have uploaded five raw files to the same site as the previous files:

I don’t have the low noise option for raw, so maybe that invalidates the tests? Anyway, I included ISO-50 and ISO-100, and I should note that the SD14 is a low ISO camera. People don’t recommend much above ISO 200. These raw files are shot into a bright light bulb to overexpose all the frames.

I’ll wait till 2:00 AM or so, to make sure the conditions are the same as before, to take some raws of the dark room.

I wonder who pays for filebin? Don’t see any ads.

  • Ron

(ron) #19

Hello again Jossie,

So if you see a blue signal, there must have been blue photons – right? I would expect that IR-photons being much less energetic, would not penetrate as deeply as green or blue, thus one would expect just a red signal from any IR-radiation …

Well, I’m not sure about that one. Since the first files were JPEGs, they would have been run thru the SD14’s JPEG engine, which would have interpreted those photons in some way (unknown to me). Since the camera does have marketing related to IR usage, I suspect the JPEG engine does handle the raw IR data in some sensible way, just don’t know what it is…

  • Ron

(Hermann-Josef) #20


Good evening Ron,

Wikipedia tells me that the sensor in your camera has the blue-sensitive layer on top, in the middle the green-sensitive and at the bottom the red-sensitive layer. So it is the other way around than I supposed.

Nevertheless, if there is a blue signal I would expect that there must have been blue photons to produce that.

I have downloaded the ISO50 image but what am I supposed to see? This is what I got: