Starlight Live V2.0 Now Available

[Paul81]

Hi everyone,

I've been a bit quiet recently due to various commitments but have been working in the background to finish off Starlight Live V2.0 - the biggest addition being multispectral video observing (the full list of updates is below). Will work may may through the posts on this forum I have missed!

OSX: 'Starlight Live' for Mac OSX - version 2.0

Win32: 'Starlight Live' for Windows - version 2.0

As usual let me know of any issues (hopefully some of the improvements to the star extraction algorithm hasn't broken anything for anyone, likewise the display processing performance updates...), or if you have any questions on how to use the latest features or what they mean...

The following updates have been made in V2.0 since V1.1:

- Added multi-spectrum stacking capability. Colour channels can be masked during stacking process.

- Added ability to select individual colour channels for display (multi-spectrum).

- Display Processing 'Reset' button now resets all colour channels to defaults as this is more intuitive for multi-spectrum.

- Added button on Image Export tab to save the current stacked image in FITS format (prior to display processing - i.e. the current raw stack).

- Added colour channel mask to the automatically generated filename when exporting raw FITS files.

- Added command line parameter '-load-image-rggb' to permit loading of Bayer RGGB encoded images. This can be used to later load and display a FITS file saved using the save current stacked image to FITS file feature (when image was colour).

- Image repeat dialog display mirrors the main display window, including any zoom applied plus overlays.

- Disable image registration during stacking by setting Max Displacement to 0 (i.e. images are just stacked without any alignment).

- Improvements to star pattern matching algorithm to increase robustness of correspondence (especially with Ultrastar images).

- Enhancements to star extraction algorithm.

- On key frames, image is divided into 4x4 grid and the 3 brightest stars are selected from each grid to limit the number of stars to be matched.

- When matching stars, the corresponding star is searched for inside a bounding area defined by Max Displacement plus 10% with respect to the position of the keyframe star. All candidates within this area are then used for correspondence matching.

- Fixed undo bug that caused exposure number to keep decrementing when 'undo last' is clicked after an undo is applied.

- Added keyboard shortcuts to zoom and pan the image display

+ or = : Zoom display in (2x each key press)

-      : Zoom display out (2x each key press)

LEFT   : Pan left.

RIGHT  : Pan right.

UP     : Pan up.

DOWN   : Pan down.

- Added keyboard shortcuts for the display processing tab. One of the functions (i.e. black level, white level) is selected via the numeric keys, and then this can be modified by other keys to decrement or increment the value instead of using the UI slider control.

1      : Select Black Level

2      : Select White Level

3      : Select Brightness

4      : Select Contrast

5      : Select Saturation

6      : Select Hue

{ or [ : Fast decrement

} or ] : Fast increment

< or , : Slow decrement

> or . : Fast increment

- Multiple performance enhancements to the display processing algorithms. These now use multiple CPU cores during processing. The more cores the faster the algorithms run.

The following are in active development for the next release:

More performance enhancements (display processing and stacking)

TEC support for cameras featuring TEC

Camera binning

Capture session directly to video (likely Web-M format)

Invert filter

Night mode

Allow multiple connected cameras

Allow different exposures on Focus / Alignment / Framing to other modes

Enjoy!

Paul

[Paul81]

Oops, I forgot to add I have also been working on trying to remove the need for taking darks (mainly hot pixel removal, but also amp glow gradients) but haven't been 100% satisfied with the results to date but will keep working as dark frames seem to be quite un-popular   (I guess it is an extra pain...)

[Astrojedi]

Paul,

As always thank you for your efforts. Will try out the new features if it stays clear tonight.

Not having to take darks makes a big difference (for me at least). Allows me to get up and running quicker and change exposure settings without worrying about hot pixels and noise from the darks frames (especially struggle with that on cold nights)

[DoctorD]

Hi Paul

Thank you for the major update, cannot wait to be able to try it out.

Hot pixel removal via a pixel mask or better still pixel replacement would be fantastic - my Lodestar-C suffers from quite a large number of hot pixels.

Hope you have a great 2016

Paul

Sent from my iPad using Tapatalk

[aparker]

Wow - that's a big release. You have been hard at work on this! I'm very excited to try out the multispectral features now that I have a filterwheel and a set of NB filters.

I would also be interested to compare bad pixel replacement to the current dark frame approach. Darks certainly work well for me, although the need to retake them does tend to limit my desire to switch exposure durations. As an aside, I have never been able to get the dark frame save-and-restore function to work, when I try I just get a blank black screen for light frames. So I always have to re-shoot them in the moment.

From what I understand about bad pixel removal (from reading the Nebulosity manual), it might still br necessary to make a new map or at least reset a threshold when changing exposure durations. Of course, simply identifying artifact pixels on the fly and masking them from the stacking algorithm would make a big difference to performance. The main reason I use darks is to prevent SLL from stacking using hot pixels rather than stars. From reading the 2.0 release notes, the revised approach to registration star choice may fix this problem anyway!

Thanks again for provided SLL to the community. It has certainly been a ton of fun for me to use so far.

[Paul81]

Thank you all for the kind comments. Thanks should also go to Don, Martin and Dom (especially Dom who sent me loads of vital multispectral data) for testing during development.

If anyone spots bugs or weird behaviour let me know. I will look into the load / save (not had issues myself but there could still be a bug there...).

I haven't done a lot of work on hot pixel removal to date, but I am aiming for an automatic approach. If you have no darks then it will apply the algorithm, if you do have darks then it will turn it off and use the darks instead (as they do a very good job of removing hot pixels). If you change exposure time then it will automatically re-establish the hot pixel map (i'll probably check it every N exposures too...).

I am aiming for both 'cosmetic' and functional - i.e. remove the hot pixels so you do not see them (i'd like to do amp glow gradient too..) in the stacked images, but if that proves to be problematic I'll scale it back to being just used in the star extraction process during stacking, so you would see hot pixels in the images but they will be discounted during stacking and not effect the image registration.

[MrOD]

Thank You!

Where is the "Donate" button !?

[HiloDon]

Thanks, Paul, for the great Christmas present. I think most everyone in the EAA community is recognizing your software as the best for near real time viewing and imaging. It is inspiring other software developers to follow suit for other cameras, which will help to make EAA more accessible to everyone.

Wishing a very Happy New Year to you and your family.

Don

[Hibou]

Paul, many thanks for this important upgrade. As a monochrome non-astronomical user, some of the improvements I can't use, but I particularly like:

1) That SL-2 works perfectly for the Ultrastar as well as the Loadstar. (I tested both mono cameras).

2) The possibility of switching off frame registration with Max Pixel Displacement = 0 (very important for me, and speeds up stacking greatly)

3) The export of the raw stacked data to a FITS file works well (an essential feature). I prefer to save the raw stacked data; I can filter it with imageJ.

4) Hot pixel filtering with the Median 3x3 Output Filter - great for cleaning up the display with little impact on speed.

5} Speed improvement using multiple cores, and the ability to pan and zoom with the keyboard.

Concerning your future improvements, I am particularly interested in:

1) Binning including asymmetric binning eg. Yx1, Xx2 (because I use a mirror that can shrink the image along one axis only)

2) Restricted region acquisition as in Artemis Capture i.e. restrict capture to Xmin<X<Xmax and Ymin<Y<Ymax (to screen out artefacts)

These two could fit under exposure control

3) Multiple camera acquisition. For example, the ability to acquire from two cameras simultaneously. At present I can only do this with ImageJ/ASCOM scripts

4) TEC support in principle, though I currently only use uncooled SX cameras

You mention automatic hot pixel removal. I would like to be able to switch that off, since I prefer to save raw data and correct it with imageJ. The Median 3x3 Output Filter already does a good job cleaning up the real-time display, which is all I need.

Thanks again for an excellent Christmas present, and best wishes for 2016 ! Alan.

[Hibou]

Oh, and I forgot for the wish list :-)

5) ASCOM camera support (Even for SX cameras some things can only be done with ASCOM)

6) The ability to launch acquisition with only basic command-line arguments and no interface or display (scripting).

Plenty there for 2016 :-) Alan.

[aparker]

Ran into some bugs (software repeatedly hung when stacking Ha captures) on my first time out with v2. I had limited time but was able to convince myself that the same issue did not happen with SL v1. Next time I can get to the scope I will do a controlled comparison between v1 and v2, save raw images, take screenshots, etc.

Paul is there anything else I can save to help document this?

Thanks

Alex

[Astrojedi]

Another bug report... Experienced unexpected behavior when stacking (straight up imaging no multi-spectral) with Ultrastar Mono on multiple occasions.

After the 4th or the 5th image the display window went dark (i.e. no image). Then in the next exposure an image showed up but it looked like the stack had reset but the stack image counter was still moving up i.e. did not reset. Note the pixel displacement was set to 3 and tracking was very stable. Eventually switched back to using v1.1 which worked fine.

[Paul81]

Hi Alan,

Thank you for your comments - some interesting suggestions which I have taken note of for the future work list :-) 

With regard to the auto hot pixel removal, it will disable if you take a single dark, but regardless (like darks currently) it does nothing to the raw FITS files saved (but would have an effect in the stacked FITS). I guess adding a checkbox to turn it off completely on the settings tab wouldn't be too difficult so will have a look at that too.

Paul

[Paul81]

Hi Alex,

Yep sounds like you have discovered some bugs in the reworked stacking algorithm! 

Not sure about the hang, but I was wondering if there is something wrong when it generates new key frames. Your max displacement is very small - can you try a larger value (say 16-32) and see if that has any effect.

The biggest help to debug issues is to have the raw FITS export enabled, and when you get a problem if i can grab the raw FITS (darks and the lights) I can then play them through the SW in the test mode under debug. Also helpful is what stacking settings you were using (mode and max displacement). Once you have the FITS, message me and we can arrange transfer. Please note, if you are using darks then I also need these so I can exactly replicate the sequence of data (basically enable raw export at the start of a session and its covered). Having the FITS files means I can find the cause of such issues really quickly - fixing them might be more tricky though 

Also helpful is the application log - start the app from the command line and add the '-enable-logging' argument. You will get a TXT file in your home directory which is a log. I plan on beefing the log with more debug output in the next version, or if I am really stuck I can augment a build and send it to you for trial. In general though, issues like this tend to be better debugged with the FITS files.

Keep me posted (likewise anyone else who runs into issues can you please do the same....)

Paul

[HiloDon]

Hi Paul,

I had the same problem last night that Alex and Hiten reported. I shut the stacking off and it wouldn't capture another image. It also wouldn't recycle. It was stuck on 3 Lights and 2 Stacks. It did work flawlessly for a number of images before the problem. I closed the program, but it wouldn't reopen, so I just used v1.1. I assume it will reopen after a computer reboot and will let you know if it doesn't.

Don

[Dave In Vermont]

Something has solved this from happening to any new person who wanted to test it out: My download failed. I don't know if you pulled the plug, Paul - but it got most of the way there,then died.

Good luck -

Dave

[Paul81]

Don - sounds like same bug as experienced by Alex. If you have any FITS data available let me know. I have a suspicion that the stacking thread is getting stuck based upon your bug report.

FYI I have also discovered whilst working on the next version that sometimes the display panning keys do not work (left, right, up, down). What is happening is that the key events are being passed to other controls on the dialogue (normally the slider controls). Should be a simple fix!

Dave - not sure what was happening there. Starlight Xpress host the downloads for me on their website. I just downloaded both versions with no problems so it might be worth trying it again?

[Dave In Vermont]

Okay, Paul - I'll run it again. If you don't hear from me, it was a success.

Dave

[MrOD]

Just a question - is it or would it be possible in any way to connect lodestar camera attached to remote computer to locally running StarlightLive?

[DoctorD]

Hi

I've had some success using a Raspberry Pi running Virtualhere USB over IP (free for one connection). Clients for OS X, Linux and Windows.

Managed to achieve download in about 5 seconds, so OK for remote viewing but slow for alignment and focusing.

I also use ser2net to provide remote access to the mount instead of paying for a second Virtualhere device.

HTH

Paul

[Paul81]

Yes there are ways of doing this. There are several projects on the web to transport USB via Ethernet (either wired or via WiFi) – using a Raspberry Pi or similar (see http://usbip.sourceforge.net/ or http://store.raspberrypi.com/projects/virtualhere as a starter). I do not know how well they work or what the lower bandwidth of typical WiFi would do to downloads and reliability.

The other alternative is to have a machine running Starlight Live connected directly to the camera out with the scope, and then use RemoteDesktop (Windows) or Screen Sharing (OSX) on another machine.

[Astrojedi]

Much faster with Raspberry Pi 2. I hook the camera up to the Pi and run VirtualHere. The Pi connects to an ad hoc network created using my laptop. This gives untethered operation.

For remote operation I use VNC to connect my observatory computer. I use a iPad as a VNC client.

[MrOD]

Great, i will try Raspberry pi 2 with wired and wifi network.

[Astrojedi]

Make sure you get a good quality wi-fi adapter for the Pi 2. The cheaper ones don't perform as well. Also make sure you have a battery/power source capable of supplying at least 1.5A. You will also need to make a small change to the Pi settings to let the USB ports get sufficient power to run the camera.

[Paul81]

I have managed to replicate the bug reported whereby stacking is going as normal and then suddenly on the next exposure you get a black image, after which upon the next exposure the image comes back again.

There is a bug in the part of the algorithm that creates a new key frame once the displacement goes past the limit set via max displacement. This was introduced as a result of multi spectrum (oops..  )

To work around this, you can use a larger max displacement to avoid new key frames being generated (I use values 16-32 pixels).

Will be fixed in V2.1!

I have also fixed the key shortcut issue.