Timelapse only capturing frame every two minutes

[CleverJay]

My night time exposure setting is 58000ms with a 2000ms delay, so that I can hopefully acquire one frame per minute. Allsky is only capturing one frame (image) every two minutes. I will try setting the delay to the default 5ms tonight, but that doesn't seem like long enough for Allsky to write the image. Any thoughts or solutions?

[EricClaeys]

@CleverJay Look in /var/log/allsky.log to see what the timings are. For example, how long between start a picture and when it is finished saving? It takes some amount of time to save the picture, which depends on what camera you have. How long is it actually sleeping between shots?

If you have a ZWO camera, what is your max nighttime exposure? That determines how long it waits between shots.

[CleverJay]

Thanks Eric, I am using the RPi HQ camera. I've attached a three frame excerpt from my log file. By reducing the delay time to 10ms, I was able to get the frames to write at an average of 65 sec for a 58 sec. exposure. I will adjust my exposure time to 50000ms and hopefully that will give me one frame per minute. I am wondering about the LIBCAMERA_LOG_LEVELS='ERROR,FATAL' message though.
log_excerpt.txt
.

[EricClaeys]

@CleverJay You can ignore the LIBCAMERA_LOG_LEVELS message. It is simply telling the libcamera-still command (which actually takes the picture) to only output ERROR and FATAL messages.
The libcamera-still command takes several seconds to start up, so that adds time between pictures. Using 50 second exposures should work. Possibly eve 53 or 54 seconds.

[CleverJay]

Thanks Eric, I thought that we had this solved, but last night with the exposure set at 50000ms and delay at 10ms I was getting 50 sec timeouts between exposures (see the attached log excerpt.) Any ideas?

log_excerpt.txt

[EricClaeys]

@CleverJay Can you attach /etc/raspap/settings_RPiHQ.json ?

[CleverJay]

Hi Eric, see attached (had to change to .txt file)
settings_RPiHQ.txt
.

[EricClaeys]

@CleverJay I'm pretty sure there's a bug in the way the software calculates how long to sleep at night. I assume it's sleeping for 30 seconds during the day (your daytime delay setting)?
In config/config.sh, if you set CAPTURE_EXTRA_PARAMETERS="--nightmean 0.3" it'll use a different auto exposure method and the nighttime sleep setting should work ok.

[CleverJay]

Thanks Eric, I will try this and let you know if it works.

[mpinnuck]

This is the same as what I am experiencing.
I posted a similar QA before seeing this post.
My Raspistill is taking over 70secs for a 30sec capture.
Similarly during the day millisecond exposures take the best part of a minute. The exposures are good, the images look ok, the time between logging the command and the saving messages is a mystery.

[mpinnuck]

I increased my debug level to 4.
You can see the delay appears to be in raspistill. Taking 40sec for a 1ms exposure.
All the post capture processing, mean calculation and adding text to the image only takes about 5 seconds.

Mar 11 11:21:24 allsky allsky.sh[6185]: > Capture command: raspistill --output '/home/pi/allsky/tmp/allsky-20220311112123.jpg' --thumb none --burst -st --timeout 1 --nopreview --mode 3 --exposure off --shutter 1 --analoggain 1 --exif IFD0.Artist=li_1_1 --awb auto --saturation 50 --brightness 55 --quality 100
Mar 11 11:22:04 allsky allsky.sh[6185]: > Focus: 23.614733
Mar 11 11:22:07 allsky allsky.sh[6185]: > exposure: 0.000 sec, mean: 0.4369, mean_value: 0.3000, diff: -0.1369
Mar 11 11:22:07 allsky allsky.sh[6185]: > MeanCnt: 12, historySize: 3
Mar 11 11:22:07 allsky allsky.sh[6185]: > i=1: mean_history[1]=0.4383 exp_history[1]=-718
Mar 11 11:22:07 allsky allsky.sh[6185]: > i=2: mean_history[2]=0.4384 exp_history[2]=-718
Mar 11 11:22:07 allsky allsky.sh[6185]: > i=3: mean_history[0]=0.4369 exp_history[0]=-718
Mar 11 11:22:07 allsky allsky.sh[6185]: > mean_forecast: 0.4353, values: 9, mean_diff: 0.1369
Mar 11 11:22:07 allsky allsky.sh[6185]: > ExposureChange: 45 (-1)
Mar 11 11:22:07 allsky allsky.sh[6185]: > ExposureTime_s too low - stop !
Mar 11 11:22:07 allsky allsky.sh[6185]: > FF aktiviert
Mar 11 11:22:07 allsky allsky.sh[6185]: > Mean: 0.436851, diff: 0.136851, Exposure level:-718 (0), Exposure time:0.00000100 s, analoggain:1.00
Mar 11 11:22:07 allsky allsky.sh[6185]: > Got exposure: 1 us, shutter: 0.0000 s, quickstart: 0, mean=0.436865
Mar 11 11:22:08 allsky allsky.sh[6185]: > Saving DAY image 'allsky-20220311112123.jpg'
Mar 11 11:22:09 allsky allsky.sh[6185]: Sleeping 30.0 seconds...
Mar 11 11:22:09 allsky allsky.sh[6185]: DEBUG: saveImage.sh DAY /home/pi/allsky/tmp/allsky-20220311112123.jpg EXPOSURE_US=1 BRIGHTNESS=55 MEAN=0.437 AUTOEXPOSURE=1 AUTOGAIN=0 AUTOWB=1 WBR=2.80 WBB=2.00 GAIN=1.00 GAINDB=0 BIN=1 FLIP=0 BIT_DEPTH=8 FOCUS=24

[mpinnuck]

I ran raspistill on the command line and executed in a couple seconds.
In capture.cpp I removed the call to "line 1727 focus_metric = (int)round(get_focus_metric(pRgb));"
I just set focus_metric = 20.0 for testing purposes.
Restarted allsky and the PiHQ captures only take a couple of seconds.
PiHQ has a large sensor 4056 x 3040, all the pixel crunching in get_focus_metric() appears to a bit to much.

Mar 11 17:29:36 allsky allsky.sh[962]: > Capture command: raspistill --output '/home/pi/allsky/tmp/allsky-20220311172936.jpg' --thumb none --burst -st --timeout 1 --nopreview --mode 3 --exposure off --shutter 1 --analoggain 1 --exif IFD0.Artist=li_1_1 --awb auto --saturation 50 --brightness 55 --quality 100
Mar 11 17:29:38 allsky allsky.sh[962]: > exposure: 0.000 sec, mean: 0.3120, mean_value: 0.3000, diff: -0.0120
Mar 11 17:29:38 allsky allsky.sh[962]: > MeanCnt: 24, historySize: 3
Mar 11 17:29:38 allsky allsky.sh[962]: > i=1: mean_history[1]=0.3241 exp_history[1]=-713
Mar 11 17:29:38 allsky allsky.sh[962]: > i=2: mean_history[2]=0.3169 exp_history[2]=-713
Mar 11 17:29:38 allsky allsky.sh[962]: > i=3: mean_history[0]=0.3120 exp_history[0]=-713
Mar 11 17:29:38 allsky allsky.sh[962]: > mean_forecast: 0.3071, values: 9, mean_diff: 0.0128
Mar 11 17:29:38 allsky allsky.sh[962]: > ExposureChange: 5 (0)
Mar 11 17:29:38 allsky allsky.sh[962]: > ExposureTime_s too low - stop !
Mar 11 17:29:38 allsky allsky.sh[962]: > Mean: 0.312799, diff: 0.012799, Exposure level:-713 (0), Exposure time:0.00000109 s, analoggain:1.00
Mar 11 17:29:38 allsky allsky.sh[962]: > Got exposure: 1 us, shutter: 0.0000 s, quickstart: 0, mean=0.312003
Mar 11 17:29:39 allsky allsky.sh[962]: > Saving DAY image 'allsky-20220311172936.jpg'

[EricClaeys]

@mpinnuck Mark, Your delay is different than @CleverJay's. His is/was due to the length of time Allsky sleeps between taking pictures, while yours is due to how long it takes to capture an image.

What type of Pi do you have? On my Pi 4 I haven't noticed any delay due to getting the focus metric. However, I just put in a change to only get the focus metric if the user wants it displayed ("Show focus" setting in the WebUI). Basically, the change was to prepend if (showFocus) to line 1727. If you get a chance, could you add that to capture_RPiHQ.cpp, then in the WebUI toggle that setting and see if you still get the drastic change in time?

[CleverJay]

@mpinnuck @EricClaeys I set my exposure to 53 sec with a delay of 10ms and with the nightmean set to 0.3 as you suggested. and I am still getting an approximately 53 sec delay between exposures. I would attach a log file excerpt, except I'm running an Bullseye update at the moment. @mpinnuck please note that I am running Bullseye that uses libcamera instead of raspistill , I don't know if that would make a difference.

[EricClaeys]

@CleverJay Look in the log file for the last Capture command, the stop Allsky and run the raspistill command on the command line - just copy from the log file and paste onto the command line. If there's a 2 > /dev/null in the command, do NOT use it. And prepend the command with time so it reports how long the command took in wall time and CPU time.

That will rule out the raspistill command as being the culprit.

Stop allsky, run sudo truncate -s 0 /var/log/allsky.log, then set the Debug Level in the WebUI (an "advanced" option) to 4 and let Allsky run for a couple minutes until you see the long delays. Then stop allsky and attach the log file.

Then look in the log file for how long it's sleeping between shots. If it's a big number, it might be the bug. Send me your ~/allsky/src/capture_RPiHQ.cpp file and I'll try to fix the bug and/or add some extra debug statements. I'll then give back to you and tell you how to use it.

If the sleep times are short, try lowering the exposure a couple seconds and try again, and keep doing that until you get one picture a minute.

We'll get this resolved so hang in there with me.

[mpinnuck]

I have a Pi 3A+.
Raspistill is not a problem it runs without delay on the command line.
My delay is due to the get_focus_metric() function.
When I remove it I get pictures every 30sec for 24 hours.
Get focus metric takes between 40-60 secs to run on my Pi 3A+.
If you set your debug level to 2 or greater the log reveals all.
The amount of Delay this routine will cause, will depend on your Pi version and sensor pixel count.

[EricClaeys]

@mpinnuck Mark, that's a huge difference compared to my 4B. When I get it running again I will time it with and without the focus function, but it was calling the focus funtion with no noticable delay.

I updated my Pi a couple days ago and got a "timeout" error near the end, and now it gets a kernel trap while booting, so it's currently unusable and I don't know how to fix it other than wipe the SD card and reload Linux.

[mpinnuck]

Sounds like your pi (sd card) is bricked, Just increase the debug level to 3 and run it, it will give you the timing. The difference is surprising when you look at the specs for the 3A+ and the 4b. The 4b is a desktop compared to the 3A+ especially if you have 4gb mem. I am sure the focus routine would not have got this far if it caused you the problems it has caused me. I am trying to get close to full exposure of the night sky with no gaps in satellites, meteor showers and planes. I am getting 30sec exposers with a 1-3 sec gap. The 3A+ only has 512 mb of ram, if it starts to page swap, the execution times could blow out significantly. The focus routine makes many passes thru 12 meg of pixels for a magic number that is not used. With the function removed i get captures every 30sec give or take a couple of secs, the function in add 40-60 secs between images. You can see my experience in the couple of log entries I posted. For focusing I stream video in a vnc session, to use the number in allsky for focusing is not practical. On a different topic have you come across any solutions for dew?

…

On 12 Mar 2022, at 10:15 am, EricClaeys ***@***.***> wrote:  @mpinnuck Mark, that's a huge difference compared to my 4B. When I get it running again I will time it with and without the focus function, but it was calling the focus funtion with no noticable delay. I updated my Pi a couple days ago and got a "timeout" error near the end, and now it gets a kernel trap while booting, so it's currently unusable and I don't know how to fix it other than wipe the SD card and reload Linux. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.

[EricClaeys]

I have a fan controlled by CPU temperature that blows air over the Pi and into the dome on one side and out the other side. Seems ro work well, but I am in the mountains of NX where it usually isn't humid.