View Issue Details

IDProjectCategoryView StatusLast Update
0002099CCdciel[All Projects] Generalpublic19-04-27 21:14
ReporterThomas JaegerAssigned ToPatrick Chevalley 
PriorityurgentSeveritymajorReproducibilityalways
Status resolvedResolutionfixed 
PlatformPCOSLinuxOS Version64bit
Product Version0.9 
Target Version1.0Fixed in Version 
Summary0002099: Change of binning causes a false image geometrie
DescriptionWith a new version of CCDCiel or possible the INDILib came a new severer issue. I recognized this issue during stacking. My Bias and Dark had a other image geometry than recent imaged lights.

Steps To ReproduceTo reproduce this, change binning to bin4x4, make a image, than change it to bin1x1 an the image geometry does not return to the maximum again.

Press reset in image Frame
1x1 -> Camera frame x=0 y=0 width=5496 height=3672
4x4 -> 1396x... (shown right in status bar, not in log)
1x1 -> Camera frame x=0 y=0 width=5472 height=3672

-----------------------------------------------------------------------------------------------

A second issue it the incorrect logging of the "Camera Frame ..." log.
2019-03-23T13:06:41.732 2: Set Binning 4x4
2019-03-23T13:06:41.732 3: ZWO CCD ASI183MM Pro: Set binning 4x4
2019-03-23T13:06:41.936 2: Start single preview
2019-03-23T13:06:42.350 2: Camera frame x=0 y=0 width=5472 height=3672 <- wrong! should bei 1396x...





TagsNo tags attached.

Activities

Thomas Jaeger

19-03-23 13:25

reporter  

Log_20190323_130330.log (4,634 bytes)

Patrick Chevalley

19-03-23 18:57

administrator   ~0005479

Hi Thomas,

With INDI the ccdframe value are always in unbinned pixel, so if you only change the binning the box with the frame size must not change, same for the values in the log.

In the log we see the frame size change immediately when the camera is connected:
2019-03-23T13:03:39.121 2: Connecting Camera INDI "ZWO CCD ASI183MM Pro" ...
2019-03-23T13:03:39.138 2: Camera frame x=0 y=0 width=5496 height=3672
2019-03-23T13:03:39.140 2: Camera cooler : True
2019-03-23T13:03:39.726 2: Camera frame x=0 y=0 width=5472 height=3672
2019-03-23T13:03:41.509 1: Camera connected

Then it change many time between this two values.

Do you have an idea why? what is the correct one?
It is possible this is when loading the default configuration. Can you check in ~/.indi/ for a file with the camera name ending with .xml and .xml.default.
Open each file and look for CCD_FRAME , then WIDTH and HEIGHT, do you find this values?

Also 5472/4=1368 and 5496/4=1374, are you sure you see 1396 for the binned image width?

An interesting test is to change binning and take test exposure using only a simple Indi client, not ccdciel.
From the indi client, connect the camera, in the option tab for Configuration click Load, for Upload click Local.
Then in Main Control set the Expose Duration=1, this take an image.
Repeat after changing the binning in the Image setting tab. Also check for Frame size in this tab.
Check the result in the files ~/IMAGE_xxx.fits

Thomas Jaeger

19-03-23 20:46

reporter   ~0005480

In my case, I changed the binning with the CCDCiel/Preview/Binning
The correct geometry comes when I press CCDCiel/Frame/Reset. It is 5496x3672px.

>Then it change many time between this two values.
This came from my pressing frame/reset to reproduce it again and again.

>Also 5472/4=1368 and 5496/4=1374, are you sure you see 1396 for the binned image width?
I was small on the status bar, I will check this.

I understand the test with the Indi Driver, I will do this tomorrow.

Thomas

Thomas Jaeger

19-03-24 09:27

reporter   ~0005481

An interesting test is to change binning and take test exposure using only a simple Indi client, not ccdciel.
From the indi client, connect the camera, in the option tab for Configuration click Load, for Upload click Local.
Then in Main Control set the Expose Duration=1, this take an image.
Repeat after changing the binning in the Image setting tab. Also check for Frame size in this tab.
Check the result in the files ~/IMAGE_xxx.fits

I did this test.
1. After load in the INDI Client the image geometry was 5496x367px what is the right value for my camera (After reloading defaults.png), I made a image, the result was right.
2. Setting binning 4x4 is crazy, the downloaded image has the right geometry of 1368x918px but the INDI Frame setting changes to 5472x3672px what is wrong. (After switching to 4x4.png)

3. Going back to 1x1 binning results in a Image of 5472x3672px

This is not a failure of CCDCiel but what can I do? Should I report it to Jasem Mutlaq or is there a change to reset a config file in a proper manner?

Patrick Chevalley

19-03-24 11:37

administrator   ~0005482

I see you report it in the Indi forum, wait for Jasem response.

You can try to save the default when the width is 5472, so it start with this value. But a "reset frame" will set it back to the max reported size of 5496.

Is 1368x918px the camera specification for bin 4x4?
In this case this can explain why it set this size, 4*1368=5472.
Otherwise it must be 1374x918, 4*1374=5496.

Thomas Jaeger

19-03-25 15:39

reporter   ~0005483

I checked this before I reported this. Reset frame works right and set it back to the right bin1x1 size.
I usally take 4x4 frame loops to center my objects. My last photo session was affected with this bug. I had to crop my masterdark and masterbias to rescue this session.
Thomas

Patrick Chevalley

19-03-26 22:16

administrator   ~0005492

I think the origin of the problem is now clear:
https://www.indilib.org/forum/ekos/4517-can-not-calibrate-images-captured-with-ekos-stellarmate-in-pixinsight.html?start=60#37008

But I am not sure if this can be fixed in INDI.
So I make a fix that must work on every case. When the ROI cover almost all the sensor size it do a CCD_FRAME_RESET before to set the new binning. The effect is to reset the binning to 1x1 and the frame to the max size, so the subsequent binning command always start at the same point and not at the value rounded by the previous binning.
https://github.com/pchev/ccdciel/commit/d0d69046e4f3d9cca75ff933c01c350be76ba26e

I make a test version you can try immediately:
https://vega.ap-i.net/tmp/ccdciel_0.9.54-1492_amd64.deb
Can you confirm if this solve the issue.

Thomas Jaeger

19-03-27 18:06

reporter   ~0005505

Thank you very much for your effort.
I tried your test version and it works.
Thomas

Patrick Chevalley

19-03-27 21:24

administrator   ~0005507

Thank you for testing Thomas.

I let the issue open for now to see how this evolve in INDI. It is probably best if this can be solved in driver and not request specific sequence in every application.

Thomas Jaeger

19-03-30 11:23

reporter   ~0005519

Last night it was clear and I took 0.9.54-1492 out in the field. There was only one situation in which your special version fails. Besides this it worked wonderful.
I have commended the log.
Jasem has not yet replied to my bug report,

Log_20190329_194741.log (24,005 bytes)

Patrick Chevalley

19-03-30 22:25

administrator   ~0005523

Thank you for the test.
From the log it look like some messages get delayed and maybe execute in the wrong order.
I will change that to be sure the command to reset the maximum size is executed before to continue.

Sure it will be better if the computation was fixed in the driver without the need to reset the size.

Patrick Chevalley

19-04-01 22:15

administrator   ~0005539

I make a change to be sure the server receive the frame reset before to set the binning:
https://github.com/pchev/ccdciel/commit/aa90c6c4c239ed0681fed5410471b7afbdf5b5b9

A new test version is available with this change and also the zero exposure time if you want to test it:
https://vega.ap-i.net/tmp/ccdciel_0.9.54-1512_amd64.deb

Thomas Jaeger

19-04-02 18:22

reporter   ~0005542

It is not clear at the moment in germany. I have tried to reproduce this issue in the lab (without teleskop) with 0.9.54-1492 but it won't. I will use 0.9.54-1512 as soon as it cleared here.
What can I do to speed the fix in the driver. I recognized that is also present with my old ASI1600.
Thomas

Patrick Chevalley

19-04-02 20:53

administrator   ~0005544

I don't know what to do to get a fix, I cannot work on this myself because I not have an ASI camera.
And no one look interested despite there is also a bug report with Ekos:
https://www.indilib.org/forum/ccds-dslrs/4906-asi183mm-pro-partial-image-download-with-ekos.html
Maybe bump this message and your.

I am confident that 0.9.54-1512 fix the issue but it is bad to have to do this kind of trick at the application level.

Thomas Jaeger

19-04-07 09:48

reporter   ~0005583

>I am confident that 0.9.54-1512 fix the issue but it is bad to have to do this kind of trick at the application level.

Hello,
I was in the field last night and did M97 and SextansA with 0.9.54-1512. Your fix works fine, thanks. I made a new indi forum entry under "CCD and DSLR". My last entry was on general. Unfortunately Jasem and the other developers seem to be on vacation, may on the NEAF.
I don't know.
Thomas

Thomas Jaeger

19-04-19 22:40

reporter   ~0005602

Until now, there is absolutely no response in the INDI Forum. Jasem may be busy. I sent him a message, I will wait.
Is it possible, that you let this fix in further releases after 0.9.54-1512?

Thomas Jaeger

19-04-21 14:51

reporter   ~0005603

After I sent a private message, I got some response from Jasem:
"If there is a solution to be done at the driver level (I looked into that and couldn't find something that is good to implement). Maybe Patrick from SkyCharts has an idea about this?"

Patrick Chevalley

19-04-21 21:54

administrator   ~0005604

I give a response for Jasem in your post in the INDI forum, I think it is better to use the forum to discuss this point.

Thomas Jaeger

19-04-27 12:45

reporter   ~0005616

Thank you for contribute the discussion in the INDI Forum. This entry can be closed.
Thomas

Patrick Chevalley

19-04-27 21:14

administrator   ~0005617

Great, thank you for testing!

The final fix:
https://github.com/pchev/ccdciel/commit/50b9ae2c4d0a9fda28b61c03e3226c9906d7166a

Issue History

Date Modified Username Field Change
19-03-23 13:25 Thomas Jaeger New Issue
19-03-23 13:25 Thomas Jaeger File Added: Log_20190323_130330.log
19-03-23 18:57 Patrick Chevalley Assigned To => Patrick Chevalley
19-03-23 18:57 Patrick Chevalley Status new => feedback
19-03-23 18:57 Patrick Chevalley Note Added: 0005479
19-03-23 20:46 Thomas Jaeger Note Added: 0005480
19-03-23 20:46 Thomas Jaeger Status feedback => assigned
19-03-24 09:27 Thomas Jaeger File Added: After reloading defaults.png
19-03-24 09:27 Thomas Jaeger File Added: After switching to 4x4.png
19-03-24 09:27 Thomas Jaeger Note Added: 0005481
19-03-24 11:37 Patrick Chevalley Note Added: 0005482
19-03-25 15:39 Thomas Jaeger Note Added: 0005483
19-03-26 22:16 Patrick Chevalley Status assigned => feedback
19-03-26 22:16 Patrick Chevalley Note Added: 0005492
19-03-27 18:06 Thomas Jaeger Note Added: 0005505
19-03-27 18:06 Thomas Jaeger Status feedback => assigned
19-03-27 21:24 Patrick Chevalley Note Added: 0005507
19-03-30 11:23 Thomas Jaeger File Added: Log_20190329_194741.log
19-03-30 11:23 Thomas Jaeger Note Added: 0005519
19-03-30 22:25 Patrick Chevalley Note Added: 0005523
19-04-01 22:15 Patrick Chevalley Target Version => 1.0
19-04-01 22:15 Patrick Chevalley Note Added: 0005539
19-04-01 22:15 Patrick Chevalley Status assigned => feedback
19-04-02 18:22 Thomas Jaeger Note Added: 0005542
19-04-02 18:22 Thomas Jaeger Status feedback => assigned
19-04-02 20:53 Patrick Chevalley Note Added: 0005544
19-04-07 09:48 Thomas Jaeger Note Added: 0005583
19-04-19 22:40 Thomas Jaeger Note Added: 0005602
19-04-21 14:51 Thomas Jaeger Note Added: 0005603
19-04-21 21:54 Patrick Chevalley Note Added: 0005604
19-04-27 12:45 Thomas Jaeger Note Added: 0005616
19-04-27 21:14 Patrick Chevalley Status assigned => resolved
19-04-27 21:14 Patrick Chevalley Resolution open => fixed
19-04-27 21:14 Patrick Chevalley Note Added: 0005617