PDA

View Full Version : Bayer pattern



Daniel_Leclerc
January 13, 2014, 11:08:58
Hi Stefan and colleagues,


Help! Help! Help!

I am using a DBK21AU618.AS camera. I started using Y800/Y800 capture with external debayering last spring, when it became possible to do it under Windows 8. I got images with NO BAYER PATTERN at all, between June 2013 and October 2013. I was very hapy!!!

I came back to astrophoto in December 2013, with the same DBK camera. All my new images show a strong Bayer pattern, using the same conditions for capture (same iccf file for settings) and same processing steps. It looks like something changed, but I cannot figure it out. My Windows 8.0 got updates since June 2013, could they affect the way the images are debayered? I am using FireCaptureDebayer but the same problem happens when external debayering is done in AutoStakkert or in PIPP.Could it be related to some capture parameters that are not covered by the iccf file and that I should readjust? I am lost!!!

Thanks for helping!

-Daniel

Daniel_Leclerc
January 13, 2014, 11:11:14
I would like to specifiy that I want to use Y800/Y800 because it is the only effective way in my hands to reach a "real" speed of 60 fps, which is important for me in order to "beat the bad seeing conditions" that we have here.

-Daniel

Stefan Geissler
January 13, 2014, 16:12:56
Hello Daniel,

in Windows 8 is a "secret" security update. Therefore, we had to rewrite the driver. Please use the version "v2.6.5.1014 Device Driver for The Imaging Source USB 3.0 Cameras [1.93 MB]" from http://www.theimagingsource.com/en_US/support/downloads/.
It works with USB 2.0 as well as with USB 3.0 cameras.

Daniel_Leclerc
January 14, 2014, 02:18:17
Thanks Stefan for this info. I installed it this evening and I will try using it when I will get a clear sky. Is there a way to get notified when new drivers are available for my DBK21AU618? It is a bit sad to discover the existence of new drivers after experimenting a problem.
-Daniel

Stefan Geissler
January 14, 2014, 10:38:20
Hallo Daniel,

we do not have a notify service for new drivers or software.

Daniel_Leclerc
January 24, 2014, 11:47:07
Hello Stefan,

No luck. I tried the new driver, but I still get a strong checkerboard pattern in my stacked externally-debayered images. I was not getting this problem in June 2013-October 2013. I started getting this problem after October 2013. Any other ideas? Could there be a specific unreported problem that I am the first one to report? I Wonder if this could be related to recent changes in Windows 8.0 . I like to use external debayering because it allows me to capture at 60 fps. I would like to be able to get images without this strong checkerboard, as it was the case last summer and beginninng of fall 2013.1666 I try to attach my last image (January 20th, 2014)

Stefan Geissler
January 24, 2014, 11:59:11
Please let me know the serial number of your DBK 21AU618.

Do you know by incident the firmware version in the camera? or the 21AU618.AS series two firmwares are available. Version 1021 shows a faint Bayer mosaic pattern at 60 fps, while all other possible frame rates are fine. The version 1028 reduces the Bayer mosaic pattern extremely, but in 30 fps two prominent vertical lines appear on the sensor.

Daniel_Leclerc
January 25, 2014, 19:52:22
Hi Stefan,

Because I did not have a strong Bayer mosaic pattern before end of 2013, I suspect it is probably not an intrinsic property of my camera.
The serial number is 44210087. I do not know how to find the firmware number.

As I was trying to figure out the firmware number, I found some info on my hard disk, related to the DBK21AU618-related "events". It shows the date of installation of new drivers. I think the solution may be there. I see 4 dates:

2013-01-13 (seems to correspond to my first trial with the camera)
2013-05-22 (a little bit before I started to use Y800/y800 and external debayering)
2013-10-16 (after this date, I started having a strong Bayer mosaic pattern after stacking my avi files)
2014-01-13 (still having strong strong Bayer mosaic pattern after stacking my avi files)

I know, for sure, that I was able to get wonderful avi files with Y800/Y800 and external debayering, without Bayer mosaic pattern, on June 19th and October 11th. These two dates are meaningful because images that I did on these two days were the best captures I made with my camera. This is consistent with the idea that I installed a "good" driver on May 22th and "bad" drivers on October 16th and January 13th. "Good" and "bad" refer to the appearance of strong Bayer mosaic patterns my stacked externally debayered images.

This is suggesting that the driver that I installed on May 22th 2013 was not showing the strong mosaic pattern and the 2 updates done in October 2013 and January 2014 were showing the problem. Should I try to install the driver that I previously installed on May 22th? I see a folder named "icy800capturesetup", that was created on my hard disk on May 22th. I suppose it contains the relevant driver. Should I try this? Am I going to get in trouble if I use this, given the changes that were done recently to Windows 8.0? I need you advice please, because I do not want to solve a issue and get in trouble because I created a new problem making this change!

I hope this detective work will help to solve the issue and I also hope it could help you to help others... Thanks again for your help. I am waiting for your advice before moving on.

-Daniel

Daniel_Leclerc
January 25, 2014, 20:00:06
... Stefan, if the driver that I installed in May 2013 is not appropriate for using with current version of Windows 8.0, should I try using it in "compatibility mode" ( I see that I can do it as a mimic of Windows XP if I right-click the icon of IC CAPTURE 2.3). Other suggestions? Do you think I have to wait for you and your colleagues for making a fix to the most recent driver, because I am experiencing a problem that was not reported by others before? Not sure I can figure out the best thing to do without your help.

-Daniel

Stefan Geissler
January 27, 2014, 09:58:17
Daniel.

If you use Windows 8, you must use the the v2.6.5.1014 USB driver. Microsoft added a secret security update, that automatically changes all unknown video format in AVI Capture to RGB24. Therefore, the Y800 of the old driver was changed into RGB24. Then the files were no longer usable programs like Registax.

The compatibility mode does not help.

The firmware in your camera is 1028, therefore no firmware update is necessary.

I do not think, it is a driver issue, because Y800 video format is completely unchanged by the driver. It is forwarded to AVI capture as it comes from the sensor. No image processing at all.

Are you sure, you used the correct Bayer pattern for debayering? (I do not know it).

Daniel_Leclerc
January 27, 2014, 11:07:59
Hi Stefan,

I am very surprised about your sentence specifying that " the files were no longer usable programs like Registax." Two weeks ago (and every week since several months) I was using my files in RegiStax and AutoStakkert2! (although I am getting this Bayer mosaic pattern if do a normal use of the wavelets).


I am sure about the Bayer pattern. I am using GB for debayering. I have two evidences that it is the right choice:
1. It is the only one, among the 4 choices, that is giving me the correct colors. The 3 others are obviously generating other colors.
2. I used GB debayering in May 2013 and October 2013, during the "good period" with my DBK21AU618 camera, and my resulting images were great, without this exagerated Bayer mosaic pattern. Last week, I confirmed that I can redo now the treatment of AVI files obtained in June 2013 without seeing the ugly Bayer mosaic pattern.


Since I seem to have everything that I need on my computer, is it possible that I have something that interferes with the process? Should I remove something from my computer? (something that interferes with the correct outcome)? On the other hand, if it is not a driver issue, could it be a software (IC Capture) issue?

-Daniel

Stefan Geissler
January 27, 2014, 16:31:31
Hi Daniel


May 2013 and October 2013, during the "good period" with my DBK21AU618 camera, and my resulting images were great,
Aha, what changed in or since October?

This is the question.

Stefan Geissler
January 28, 2014, 10:08:06
Hello Daniel,

I wonder, how we can proceed in this case. The situation is not satisfactory: You are disappointed about the images, I do not know, what happened with the camera.

A good start could be following:
You photograph a white sheet of paper or a white wall with not lens or telescope on the camera. Simply point the open camera aperture to the paper or wall

Use an exposure time as short, so the image is not over exposed.
Use the white balance values 32 and disabled auto white balance.
The image should look a little bit brown / red, because we have no IR cut filter.
Use Y800 video format, but enable debayering in IC Capture with GB pattern and bilinear debayering, so you see a colored image.
Set 60 fps
I my tests, I am now able to see a faint Bayer pattern of the red pixels. Because the camera is a DBK. There is no IR cut filter, so the camera is very sensitive to red and infra red.
You can magnify the effect by raising gain.
Also the sensor heats up over the time, so you will receive IR radiation from the sensor too.

This happens with driver 2.4.14 and 2.6.5. No difference in the image is visible.

This explains the prominent red and brown pixels on your posted Jupiter image.

After I found this pattern, I put an IR Cut filter on the sensor and the image becomes green and the Bayer pattern is no longer visible.

Now we have to find out, why the red pixels were not as prominent in the past.
Some questions:
Did you play around with white balance settings?
Did you change your optical setup?

Daniel_Leclerc
January 31, 2014, 04:10:55
Hi Stefan,

When you mention the green image, I assume that you turn the debayering on.
When you mention the “white balance” to 32, I assume that you have the “white balance blue” in mind. Taking that in account, I can reproduce what you explained.

In June 2013, I saved the iccf file (configuration file) that gave me my best results with Saturn. I used the same iccf file for Jupiter in October 2013. It was fine, no strong Bayer mosaic pattern. Then, I updated my IC Capture software and the driver. My problems started there. No change in the optical setup. No change in the white balance settings since the same iccf file was used. I am sure there were also some updates to Windows 8.0 in the same period.

Before to read your last message, I started some tests, similar to the one you proposed, with a wall as a target, no scope, in my house. I realized that the choice of the Debayer algorithm, (BEFORE to switch the debayering off) has an effect on the Y800/Y800 capture with debayering off. I did captures of 20 seconds. I did the external debayering with IC capture Debayer, and I stacked the avi file with AutoStakkert2!. Then, I used the “conv” file for each one, I enlarged it and I saved a screen capture of everyone. All captures and treatment were done in parallel, in exact same conditions. I observed, on a reproducible way, an effect on the outcome by:
- The debayering algorithm that was used before debayering is turned off
- The compatibility settings used for running IC Capture
- The driver that is used with the camera
- The version of IC Capture that I used

I was surprised, because your previous comments made me supposing that the debayering algorithm should not have an effect on the Y800/Y800 capture. I also previously understood from your comments that the driver should not affect the Y800/Y800 capture. I got some images with more obvious checkerboard, some with more homogenous color (probably acceptable, but not perfect) and some with apparent lines instead of a checkerboard.

Based on that, I wonder if I unintentionally changed the debayering algorithm in October 2013, since this is not taken in account by the iccf file. Since I also observe an effect of the driver and the IC Capture version, these may also have contributed to my problem, I guess. I chose two conditions that gave more homogenous color in this test, in order to try them with a real celestial object, when I will have a chance to do so.

Any idea after reading this? Is this unusual to see the changes that I observed, when I changed the 4 points that are listed above?

Best wishes,

-Daniel

Daniel_Leclerc
February 2, 2014, 16:48:21
1668Hi Stefan, I try to attach a jpg file showing some of the examples discussed above. Sorry for the bad quality of the image and limited number of panels, I was not able to upload big files that would show better details. Do you think that my observations, related to the 4 points discussed in my previous post, mean that I have a unusual problem linked to my camera/computer and not shared by other users OR do you think I am just testing parameters that were not tested by others before?

I have the feeling that my external debayering pattern at 60 fps, showing an obvious Bayer mosaic pattern, may have been caused by an accidental change in the Debayering method when Debayering was ON in IC Capture. I would like to get your opinion about all that, and your feeling about potential guidelines that I should consider for deciding among the 4 points that are mentionned in my previous post.

Best wishes,

-Daniel

Stefan Geissler
February 3, 2014, 09:54:58
Daniel,

I spend another 2 hours with my colleagues and tested older drivers and older IC Capture versions. I can see the pattern with all versions at 60 fps.

- The pattern is enhanced, if the images are stacked.

- The pattern will will be less visible, if an IR cut filter is in use.

I wonder, whether your older images have this pattern too, but less visible, because the light conditions were different.

I do not think, the camera changed its behavior over the time.

Daniel_Leclerc
February 3, 2014, 11:03:56
Hi Stefan,

Thanks for investigating this issue.

I tried to answer to a post where you were asking me about the driver versions and you were also asking info about the wavelets. It looks like your post was removed as I was trying to answer. I will answer here. The nice images that I have been taking of Saturn and Jupiter at 60 fps with Y800/Y800, without getting the strong Bayer mosaic pattern, were done using the usbcam_2.5.8_tis driver (downloaded in May 2013). I recently used the usbcam_2.6.5_tis driver (downloaded in January 2014).

The “wavelets” are part of the tools in RegiStax and Iris, for examples. I am not a mathematician and I do not understand this very well. However, I can say that the ultimate goal of amateur astronomer doing planetary or Moon imaging is to “beat the bad seeing” by getting several nice images that are sharp enough, among other images that are not really looking at focus because of air turbulence. Images are stacked one above the other one using AutoStakkert or RegiStax. The “wavelets/” are then done for doing adjustments in the stacking using various “layers” and getting the optimal arrangement of the data for having an outcome looking like an image that is as sharp as possible but still looking natural.

About your question related to conditions of capture: I started getting my Bayer problem when I was using same parameters from an iccf file that I created in June 2013. It worked fine until October 11th 2013. Parameters during capture were the same except for the gain, that I need to adjust for every capture. Planets are not always as brilliant, depending of their distance, that is changing, as well as conditions of the sky.
About your question related to the presence of some Bayer mosaic pattern in my “good” captures done before October 11 th 2013: yes, if I zoom on my final images, I can see some trace of the Bayer mosaic pattern, but it is really very minor compared to the strong pattern that I am getting since October 2013.

I will investigate the effect of the IR/UV cut filter with planetary targets, when the sky will be good enough. However, usage of this filter is not a perfect solution. Amateur astronomers have shown that IR/UV filters are helping to more easily get correct colors, but these filters are obviously giving images that are not as sharp as without the filter. In some instances, several amateur astronomers even use an IR-pass filter when they do Moon imaging, for getting sharper images. Moon imaging is part of my hobby. I would like to get images at 60 fps (Y800/Y800 captures) that are not showing the strong Bayer mosaic pattern.

I also made comparisons between IC Capture 2.2 and the recent IC Capture 2.3.382.1796. In my hands, for trials with the wall in my basement, the new 2.3.382.1796 version gave me less of the strong Bayer Mosaic pattern.

The internal debayering algorithm that is chosen before to turn debayering off, also affects the outcome. Assays with the wall in by basement show, so far, that EDGE SENSING (before to turn debayering off) is showing less of the strong Bayer mosaic pattern (observed in the file obtained from external debayering).

The compatibility settings for running IC Capture also affect the outcome when I shoot the wall in my basement. So far, it looks like Windows 8 gives me the best images, with less of the strong Bayer mosaic pattern.

Best wishes,

-Daniel

Stefan Geissler
February 3, 2014, 16:07:13
Dear Daniel,

Thank you for your extensive post.

However, the debayering does not matter at all. You may do the same with YUY2 video format. You will see the very same results and no debayering on the computer is used. It is all done in the camera.

You could use AMCap for display of video, then you should see the same effect.

Daniel_Leclerc
February 5, 2014, 02:23:31
Dear Stefan,

Thank you for your investigations and your response. I will change my strategy for getting rid of the checkerboard pattern that I obtain when I do the external debayering, based on your last message.

Last night, I captured several avi files for Jupiter )(Y800/Y800, 60 fps). I made 10 avi files with IR/UV cut filter and 10 avi files without the filter. I can confirm your conclusion when I am using a celestial target: after stacking, the Bayer mosaic pattern is much less obvious with the IR/UV cut filter.

However, I still see it in some areas of the image, even if I used the filter. I played a bit with capture conditions (BG or GB; various settings for red and blue balance) and I figured out that it is becoming more obvious with some of these settings. Do you have an idea about the most important parameters for minimizing this mosaic pattern? I would like to find conditions for minimizing the presence of the checkerboard after stacking, using optimal capture conditions. Also, I would love to get images with minimal checkerboard pattern, when I do not use a IR-cut filter.

-Daniel

Stefan Geissler
February 5, 2014, 09:36:25
Hi Daniel,

If you want to do Y800 / Y800 capture, you should not use the Debayering at all, otherwise you will receive RGB24 videos instead of Y800 videos

If you change the Bayer Start pattern in the post image processing, you will receive wrong colors. Therefore, there is only one valid setting.

I must admit, I have no recommended settings, but you may have a look into other astronomy forums, where other astronomers have suggestions.

Daniel_Leclerc
February 5, 2014, 17:11:19
Hi Stefan,

I adjust my colors, before starting the capture, with Debayering ON, and I turn it OFF just before to start the capture with the red button in IC Capture. I was wondering if the selection of the Bayer pattern, before to turn Debayering OFF, was affecting the data that is saved during the capture that is done after turning Debayering OFF. I understand from your answer that it has no effect.

This is bringing a new question: I know that I have to choose GB as the Bayer pattern for the external Debayering step. Does it mean that I SHOULD always adjust my colors, before to start capturing with Debayering OFF, with the Bayer pattern set to GB?

Sorry if my questions are looking very basic, I try to understand these things correctly for doing better captures.

Best wishes,

-Daniel

Stefan Geissler
February 7, 2014, 08:47:04
Daniel


This is bringing a new question: I know that I have to choose GB as the Bayer pattern for the external Debayering step. Does it mean that I SHOULD always adjust my colors, before to start capturing with Debayering OFF, with the Bayer pattern set to GB?


Good question. If your post processing software is not able to adjust the white balance, then this is a good idea. Also the the white balance controls in the camera work on the analogue image data directly, as far as I know, so it is a good idea to performe the white balancing, before you start to capture. (This brings a new question: How to place a white sheet of paper in front of Jupiter :-) ) I guess, you may ask other astronomers, how they do this.



Sorry if my questions are looking very basic, I try to understand these things correctly for doing better captures.

Never excuse for questions. An Asian philosopher (Buddha?) said: Better ask and be embarrassed for one minute, than never ask and be embarrassed for the rest of your live.

You are welcome.

However, my astronomical photography skills are quite poor, therefore, I am not in any case the correct person to ask.

Daniel_Leclerc
February 15, 2014, 04:18:17
Thanks Stefan,

I followed your suggestion. I have a friend doing external debayering with a camera "similar to mine" and I tried using his parameters for capture. Let me explain some details.

This friend is using a DFK21AU618. I understand that this camera is practically the same as a DBK21AU618, beside the filter on the DFK. He got some dust behind the filter and he decided to clean it . Since that, he is using the camera without the filter. In his hands, without the filter, captures with Y800/Y800 at 60 fps are very good, without any important Bayer pattern after performing the external debayering. He can see some Bayer pattern, but it is not so important, compared to the one I get on my side.

I am getting a strong Bayer pattern when I use his iccf file parameters for setting my capture conditions. We both use similar approaches for stacking the avi files.

The serial number for my DBK21AU618 is: 44210087.
The serial number for the DFK21AU618 is: 34110327.
I must say that we are not using the same telescope. Mine has an aperture of 8 inches and his scope is 11 inches. It would be nice to try my camera on his scope, and vice-versa, but we are not in the same city and we did not do this trial. Based on these facts, do you have any idea that could help me for getting captures not showing a strong Bayer pattern when I do not use a UV-IR filter with the DBK on my 8-inch telescope? You previously mention differences in various firmwares for similar cameras and even for cameras of the same model. Could this be involved in our specific case? Is there something related to this issue that I should/could consider?

Thanks,

Daniel

Stefan Geissler
February 17, 2014, 09:10:20
Hello Daniel

For the 21AU618.AS series two firmwares are available. Version 1021 shows a faint Bayer mosaic pattern at 60 fps, while all other possible frame rates are fine. The version 1028 reduces the Bayer mosaic pattern extremely, but in 30 fps two prominent vertical lines appear on the sensor.

But we already tried this, did we?

Stefan Geissler
February 17, 2014, 10:04:05
Daniel

I would like you, to go in contact with me by email again. I spoke with my supervisors and we want to look on it on our own and may exchange the camera.

Daniel_Leclerc
February 17, 2014, 15:06:28
Hello Stefan,

Thanks again for your help. Based on the serial number, you were able to deduce that my DBK has the 1028 version of the firmware. I was wondering if the DFK of my friend (serial number 34110327, but he removed the IR-cut filter) is using a different firmware than my DBK, since he does not get a so important Bayer pattern. Can you tell from his serial nuber? It may also be the optical system (we are using different telescopes) that makes the difference between his results and mine, I suppose.

You can reach me by email at deleted to protect you from spam mails

-Daniel

Stefan Geissler
February 17, 2014, 15:27:48
The camera 34110327 uses the 1021 firmware. No stripes at 30 fps, but therefore a bayerpattern. This was the green pixel pattern, as far as I know. So you may try it and see, what happens.

Daniel_Leclerc
February 17, 2014, 16:05:53
I never played with firmwares... How can I change the 1028 for the 1021, keeping the possibility to get back to the 1028? Are these files or packages (?) available from the TIS web site? Is there a simple procedure that I can follow? This is all new to me. :-)

Stefan Geissler
February 18, 2014, 09:44:30
The Update, I would sent to you , contains both firmwares, so you can switch between them.

Daniel_Leclerc
February 26, 2014, 10:26:24
Hi Stefan,
Did you try to send me the firmwares and some instructions? I did not receive them and I wonder if there was some problems with the email.
-Daniel

Daniel_Leclerc
February 26, 2014, 10:27:00
Hi Stefan,
Did you try to send me the firmwares and some instructions? I did not receive them and I wonder if there was some problems with the email.
-Daniel

Stefan Geissler
February 26, 2014, 15:57:10
Daniel

If I remember correctly, I offered to exchange the camera.

Daniel_Leclerc
February 27, 2014, 11:20:01
Hello Stefan,

I am confused because your last message was about a proposal for testing the 2 firmwares that you were going to send me (with some instructions). I thought you were suggesting to test this before going for a physical exchange of the camera. I guessed that you were going to send me this by email, but I may have misunderstood, as I do not know much about these firmwares issues. I assumed that they concern computer files that can be transferred to the camera.

Best wishes,

-Daniel

Stefan Geissler
February 28, 2014, 11:46:16
Daniel

please generate a support request at http://www.imagingcontrol.com/en_US/support/case/

For fields, you do not know what to fill in, select simple the first item...

Daniel_Leclerc
March 1, 2014, 16:04:17
Hi Stefan,

I tried... the web platform does not accept "DBK 21AU618 44210087", "44210087", nor other variations that I tried. What should I enter for the serial number?

Best wishes,

-Daniel

Stefan Geissler
March 3, 2014, 11:58:00
I searched your email address and sent something to you