Quick and dirty BPP in the field

[miska]

Hi !
I am using BPP to stack my DSLR images, and that works fine. It is quite a slow process though, even on my desktop.

Now, when I am in the field, I would sometimes like to do a quick and dirty stack of some images, just to see if my object is in my field of view or not (if it is not visible in one exposure - sky limited). My laptop is slower than my desktop, so BPP as it is seems too slow for this.
So is there a way to either accelerate (by a lot !) BPP to get a quick answer ? Or is there a better tool for this ?

Thanks !

[Alejandro Tombolini]

just to see if my object is in my field of view or not

It should be enough using ScreenTransferFunction process (STF). You also can adjust the values conveniently.

From STF Documentation:

Auto Stretch

Click this button to perform an automatic screen stretch. Ctrl-click this button to open the STF Auto Stretch dialog. On this dialog, you can modify the two parameters that control calculation of automatic STFs:

Shadows clipping

This controls the shadows clipping point of the computed STF. This parameter is expressed in sigma units, measured from the median of the target image. Recall that the median identifies the main histogram peak in deep sky images where the sky background strongly dominates their pixel value distributions. Hence, this parameter will normally have negative values, since we want to clip at the left side of the main histogram peak. The default value is ?1.25.

Target background

This is the desired mean background value in the normalized [0,1] range. ScreenTransferFunction will compute the required nonlinear midtones balance transformations to achieve this mean background level in the screen representation of the target image. The default value is 0.25.

Automatic STFs are computed separately for individual RGB channels when the Link RGB channels option is disabled; otherwise a unique STF is applied to the three RGB channels. By default, the Link RGB channels option is enabled. Separate STFs are useful to compensate for color balancing problems in raw data in most cases; however, always take into account that when separate STFs are being applied, you are not seeing, in general, a screen representation of the actual RGB color image that you have.

Saludos, Alejandro.

[miska]

Hi,
I did try stretching the values of a single sub, and it did not help. I was looking at very faint structures (Simeis 147) and I could see "something" on an individual image, but not enough to be sure that I was pointing at the center of the object.
Any other ideas ?

[pfile]

if you are using windows for image capture, i would look into astrotortilla - it includes the plate solver from astrometry.net and it can tell you exactly where you are pointing, so there's no need to inspect the image to find out if your framing is right.

rob

[miska]

Yes, plate solving is on my list. Unfortunately, I have not yet found a system that works offline, on a mac.
Thanks for the suggestion.
But nobody is coming up with an actual answer to my question: any way to rapidly combine (in an non optimal way) several images together ?
DSS does this much faster than PI - but DSS doesn't work on a Mac, and I'd like to stay within the PI environment.

[msmythers]

You have a number of things you could do in PI to solve your problem.

While the BatchPreprocessing script as it stands won't work without calibration you could use old calibration files possibly. Not the best for a proper image but it might work well enough. You could also rewrite the script to bypass the calibration process. That's the great thing about PI scripts, we have the possibility making a script do what we want.

The next is what Alejandro suggested. Use a single image, process it or just unlinked AutoSTF with a star chart if plate solving is not available to check framing.

If you have an internet connection available to your computer at the time of your capture you could take a single image and use ImageSolver and AnnotateImage scripts and plate solve within PI.

You can also do the processing that that BPP would do manually but bypass calibration. It's only 3 simple step. BatchDebayer, StarAlignment and ImageIntegration. You need at least 3 images but the process is very fast.  I've attached a screen dump showing the 3 steps.


Mike

[miska]

Thanks Mike, I'll have a look at the manual process...

Now, since I have lightroom on my Mac, I could also directly export TIFFs, and bypass the debayering. I could even imaging to downsample my  images (now either 20Mpix 6D or 36Mpix A7R) while exporting in LR, to speed up the subsequent image registration and  addition that could be done in PI.

IF I didn't do dithering (which I do), then it would be really simple: just bovinely stack all the frames. But now, I need to first register (but integer pixel shifts would be enough and presumably much faster) before the actual stacking.

Perhaps there is something in this LR downsampling idea. Hmmmm.

[gvanhau]

Hello
If your pointing is not so bad, you could use the Image solver script, using a local star catalog to get the center of your image.
You can afterwards  also use the anotation script to see where your DSO is in yoyr image.
Geert

[oldwexi]

Hi Miska!
I would not waste exposure time with exposing and stacking multiple exposures in the field.
It is  a matter of planning before you go outside.
The simplest way to find the center of the "invisible" object is to make a short exposure
for example 10 seconds after you have placed the scope relatively near to your target.
In this exposure only the brighter stars are visible. They show, especially with Simeis_147,
nice significant brighter star combinations.
Running PI Star Generator with the center of Simeis_147/Sh2-240
creates this significant star combinations also. Comparing the stars in the exposure and
the stars created in PI-starfield gets you very fast to the Point where you want to expose.
No Need to see your object.

Find attached a link to a screenshot which shows
the created starfield around Simeis_147, the process with the Parameter and
a planetariumsshot which also shows Simeis147 and its  significant stars.


I personally have moved away from imaging bright objects, i moved to Imaging molecular clouds and
dark nebula which are not visible also on a longer single shot.
So, the method to find the significant star combinations is a must for me to get the scope fast to
the needed point. Stacking is a NoGo for me.
Hope thats an idea for you.

Gerald

[miska]

Thanks for all the suggestions !
For sure, I will not have access to Internet while in the field. So is there a way to download (even a large, I don't care if it eats a few GB of space) plate solving catalog onto my Mac, beforehand, and then use plate-solving offline, within Pixinsight  ?
I did not see, in the ImageSolver script a way to force it to load a catalog in advance. Ideally, I would download the whole catalog and use ImageSolve offline. Alternatively, I could just download a piece of the catalog (near my object, but with a comfy margin).

Looks like this should work...

Thanks again for all the suggestions !

[miska]

And answering my own question: yes ! It looks like in http://pixinsight.com/download/ on can download the PPMX catalog, for offline use.
Looks like I'm good to go, will have to test this with the images I already have, but the workflow looks pretty clear...
Yay !

[gvanhau]

Hello
There are several catalogs you can download for  free. (GSC1.1, UCAC4, etc).
The image solver script doc is here: https://dl.dropboxusercontent.com/u/71653208/PublicScripts/ImageSolver.pdf
See  charpter 2.3 how to use a local catalog.

Regards
Geert

[pfile]

Yes, plate solving is on my list. Unfortunately, I have not yet found a system that works offline, on a mac.

i only suggested astrotortilla because it's easy to install, and 98% of people doing astroimaging use windows for image capture.

astrotortilla is just a wrapper for astrometry.net... but astrometry.net will run locally on the mac and the easiest way to install it is using homebrew - see the middle of this page:

http://astrometry.net/doc/build.html

also AstroPlanner is an amazing program and it also runs natively on the mac. it can help you find and frame objects:

http://www.astroplanner.net/

it is certainly possible to use ImageSolver to solve your images but not being a blind solver you have to give it good hints. if you are too far off your target then it will be very difficult to get it to solve. astrometry.net won't have that problem since it's a blind solver. but of course you have to configure it right and passing it an image scale hint is a very good thing to do to speed it up.

while i am a mac user since 1984, i eventually started using windows for image capture. this is because with a program like SGP, you just figure out what center coordinates you want with something like astroplanner, then just tell it to image there. it handles syncing the scope to the sky and finding the object by plate solving.

rob

[miska]

One very cool thing that I found, is that once the picture is plate solved and saved as fits, I can import it into Cartes du ciel, and overlay it with the sky map. Very graphical way to see where I am...

Now, I'm trying to:
- Overlay into CdC the isophotes of faint nebulae (it should be somehow possible, but don't know how), in addition to my picture.
- Figure out if I can easilly calculate by how much my pointing is off (in RA and DEC), and how to give these offset to my AZ-EQ6 mount hand controller (I don't think I can say "move X deg in RA and Y degrees in DEC").

I am still trying to avoid using a computer in the field continuously (like with EQmod). I just want an occasional use of the laptop, to check everything is fine.

But those are not PI questions anymore :-)

[pfile]

can't CDC just tell you the coordinates of the mouse cursor? after putting your image on, just mouse to where you want the center to be and write down the coordinates, then tell the mount to go there. or, if CdC is connected to the mount, does it have a "slew here" function? i thought some of these programs know how to talk to certain mounts, even without ASCOM.

anyway for me, it's just way easier to use the laptop and let it find the target... saves a lot of time. for instance, i set this up in astroplanner:



25 minutes later this was on my hard disk:



some people are starting to put these mini-computers (like intel NUC) right on the OTA itself so that the wiring is entirely self-contained. then you just use VNC or teamviewer over wifi to connect to the computer and control the mount. it's too bad most astro software is only on x86 windows, becuase there are some very small (credit card size) yet powerful arm-based computers available now, like the odroid XU4. that would make a great on-mount computer.

rob