View Full Version : Ghost images of bright edges DMK21AU04.AS

June 19, 2009, 00:28:46
Hello All,

I have a question about the DMK 21A04.AS camera.

I'm having an issue with the performance of the camera. It seems that the camera produces 'ghost' images of contrasty spots in the images. These ghost images are shifted a couple of pixels to the right of the actual object causing it. This ghost object only shows when there is a bright edge in the actual object;
- this can be the edge of a planet
- or it can be from star or even a bright nebula in a deepsky image. It seems to be somehow related to either the gain or brightness setting of the camera (the gamma was set to 100, I believe that is the default value).

Here is an example of two days ago showing a resized stack of about 70 images of M57 taken in prime focus of my newton telescope without ANY filters used. This stack has not been sharpened in any way, it has only been made a little bit brighter. You can see several ghost stars next to the four brightest stars (shifted to the right), but there is also a slight ghost image visible of M57 itself.


What is the cause of these artifacts? It's not a reflection, because the ghost images only show edges (like an edge detection filter!), and not the bright stars themselves. Also, I did not use any filters to get this image, it was taken in prime focus (it has been resized to 2x to make the effects more clear). My modified Toucam Pro II with the same ccd (icx098bl) never showed this issue. I'm guessing it has something to do with the electronics of the camera.

I hope you can help me out!

Kind regards,


ps. I know I'm not the only one with this issue; I talked to several other people (owning a DMK21, and DMK41), and they also show this effect on planetary images (edges of the planets overlayed on the planet), and on moon images (extra edges of the crates just next to the crates). It is always shown next to a contrasty spot.

Another example of the effect seen on Jupiter:

The 'new' edge was manually removed on the dark side of Jupiter, but on Jupiter itself you can still see it. These images were rotated 180 degrees, so the actual ghost image is actually to the right of Jupiter (like in the deepsky image) Note that this might look like the well known onion ring effect, but it certainly is not that. The onion ring effect is only visible inside the planets, and never outside it! It's also only visible after sharpening, and this is even visible in the raw stack of M57.
The animation consists of 17 recordings (all stacks of about 1500 frames each). The extra edge stands out due to sharpening effects, but it's already there to start with. As you can see it also shifts position between the recordings, this might be due to changing gain settings (I was pretty much playing around with different gain settings on every recording).

Stefan Geissler
June 19, 2009, 10:14:34
Hello Emil.

as already written by email, the source can be the used software too. However, the original images must be checked, whether there is a movement visible, that may is calculated wrong in the stacking software.

In the camera is no programm running, that creates ghost images. Also the CCD is the same as in the TOU cam (or what this name was).

However, please let me know the gain and exposure values. I will try to reproduce the problem. I have no telescope, thus I must simulate the starts.

My engineer asked, whether the behavior changes, if you use a different frame rate. That means, whether the distance betwee the the spot and the shadow changes.

Stefan Geissler
June 19, 2009, 10:51:46

I just tried to simulate your problem using a back background and a laser pointer as star. Also
Registax4 was used for stacking. The lens was a very high end one. I got no shadows under different
conditions: Iris closed, Iris opened, different exposure and gain values.

June 19, 2009, 14:02:46
Hello Stefan,

Thank you very much for your response. Perhaps the following image can show the effect a little bit better. You see several stacks of frames; 1 frame up to 200 frames.
The actual ghost effect is very dim, barely visible in one single frame (it hardly gets above noise level, but if you look closely you can still see it). When more frames are stacked, you can see it more cleary. The final frame shows an oversharpened image clearly showing the position of the ghost effect.

The animation is in its original size, but it is cropped.


I tried different software stacking programs, including one I wrote from scratch myself, so I know that it really is just an averaging effect that is being performed when stacking.

- In single frames the ghost effect can hardly be noticed, due to a very low signal/noise ratio. However, in the attached image of Jupiter a trace of it can still be seen in the single frame. When more frames are stacked (when noise is reduced), the ghost signal becomes apparant.
- It's not related to shutter speeds; it shows up in 4 second recordings, 1/60s recordings, and 1/30s recordings.
- It does not always show up. it's always visible in the long exposures I created, but then I had the gain setting opened fully. In the Jupiter images I played around with the gain setting, exposure and brightness settings, I believe this might have something to do with the effect actually showing up.
- Judging from my own recordings, the effect is not always in the same position (you can tell your engineer that!). I don't notice a shift in the long exposure images, I used exposure times 1s, 2s and 4s, and the effect appears to be on the same position each time (all exposures had gain set to full). In the Jupiter images it's does change position, but I had not written down any gain settings for those recordings (but I know that I played around with them!).

If you need more information, please let me know!

ps. I used the default imaging recording software that came with my camera (IC Capture), and the Y800 codec.

Stefan Geissler
June 19, 2009, 14:22:06
Hi Emil,

Thank you for your post. I would like to know, which frame rates use use. The engineer asked for the ghost image position depending on the frame rates.

June 19, 2009, 15:18:42
Hi Stefan,
I selected the maximum framerate at all times, and I then just changed the exposure setting (which then slowed down the framerate as well...).

I really can't figure out which exposure settings I used, because I did not write that information down. The only thing I can tell you about that now is that I used somewhere between 1/78s and 1/15s for Jupiter images, with the sample image of Jupiter of my previous post probably at the slower side of this (1/30s - 1/15s).

If weather permits, I want to see if I can reproduce the ghost image under different conditions (perhaps by simply imaging a bright star in different exposures, or by trying to image Jupiter again).

Hope this helps, let me know if you need any other information

Stefan Geissler
June 19, 2009, 15:23:43
Hi Emil,

the frame rates are the interesting part, because they have some influence on the speed of the camera and the read out of the CCD. This the only position, where a problem can occur.

However, since the problem is so faint in "normal" images, I am not sure, whehter we can do anything about this.

I am waiting for the new images (they are still very good!).

June 19, 2009, 16:22:07
Stefan, Thank you.

I know it's possible to create good images using the DMK camera's, I'm just hoping that this one 'problem' could be fixed. I would like to get the maximum out of this camera ;-)

Let's hope the night will be clear so I can do some more testing. I will report back when I have gathered more information. In the mean time, if you have some suggetions for tests I could perform (would changing the USB cable matter? upgrading firmware? changing software? brightness settings/gamma settings?), I'll be glad to hear them.

One quick question; The brightness setting almost has no effect on the actual brightness of the previewed, and captured, image. This makes it difficult to not clip the histogram when imaging objects on the dark background of the night sky. Is there a way to boost up the brightness somehow, so there are no or little pixels with zero brightness?

Stefan Geissler
June 19, 2009, 16:36:44

As far as I can see, there is no way getting rid of the ghost images, since we have no idea, what causes them.

The brightness slider adstust dark power setting on the CCD, thus there is not so much too see. You may use the gamma setting, that calculates the brightness on the image in the camera. That means, the original data from the CCD are manipulated by the camera. This is the only way, but it does nothing on the CCD, only on the read out image data.

June 22, 2009, 02:10:31
I'm just in the middle of some testing (and it's vey cold outside.....);

It appears the effect only shows up when 60fps is selected in the Device Settings dialog. All other framerates work just fine. The effect might be somewhat dependend on the gain setting; higher gain settinngs show the effect much more clearly.
It's definately not dependend on the exposure setting. The brightness and gamma settings also has no real effect on it.

I have tested on the bright star Deneb in prime focus of my F/4.5 telescope (at 1140mm focal length). I used all kinds of exposure settings (from 1/10000 to several seconds), and the position of the effect does not seem to change according to any settings.

So, my question now is; why does it show at 60fps, but not at 30 or lower framerates? And, could it be fixed (I just keep on asking this question hehe). Could it possibly have something to do with usb-cables/ports or is it really in the camera?



ps. attached is the finished image of M57. The effect was paint-brushed away, but next time it wont be necessary, because I will use shorter FPS for deepsky objects :)

I'm heading back inside to warm up a bit, can't feel fingeerresrss

Stefan Geissler
June 22, 2009, 10:00:22
Hi Emil,

first of all, I hope, you got the sensitivity back in your fingeerrrrs :-)

However, the 60 fps frame rate was the first suspect of the engineer. From your experiences it is acknowleged. I think these ghost images are a side effect of this high frame rate. I do not think, it can be avoided, since we do not know, how these ghost images are created in the CCD.

The M57 image is brilliant from my point of view! Thank you for posting.