Click here if you are having trouble logging into the forums
Privacy Policy |
Please read our Terms
of Service | Signup and
Troubleshooting FAQ | Problems? PM a Red or a Green Gu.... uh, User
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
I'm currently less of an astronomer than a "player with cameras". I have a Canon A630, with CHDK installed. I've previously had fun dabbling with panoramas and HDR.
Since CHDK allows me to take lots of shots ("fast intervalometer") and to suppress the dark frame subtraction that slows down long exposures on a Canon, I put the camera on a sturdy tripod, pointed at the sky, and let 'er go.
Since I have no way of tracking, I found (by trial and error) that at maximum zoom (134mm AKA 4x, don't laugh) that a 2 second exposure is sort-of-ok.
I took 5 minutes worth of 2 second exposures, which comes out at around 118 frames (there's some overhead in the camera recording data, so I don't have the "obvious" 150 frames).
On stacking these frames (which are VERY dim, since I used ISO 80, f4 2 second) the results are disappointing.
With all those frames I was hoping that noise would be reduced, and gray scale depth increased.
I have found that if I pre-multiply my data up a bit before Registax (I tried multiplying by 7) I get much better results.
Evidently they're brighter, but there's also far more colour depth, and hence less banding when I edit the curves in cinepaint.
It's (I guess) something to do with scaling, fixed point arithmetic, ranges and quantisation, but I don't know
enough about registax to be certain of the details.
Here are the results;
If I simply stack in registax, I then have to multiply the samples by 70 to get this (reduced to 8bit for web use)
If I pre-multiply by 7, stack, and then multiply by 10 (for a net multiplication by 70) I get this superior result:
Can anyone explain what's going on?
I'm really not interested in spending my way to a solution by buying telescopes, adaptors, trackers etc.
I simply want to see what's possible with a P&S camera and a laptop.
My aim is to dig as much light and detail out of the sky as possible, within my equipment constraints.
BugBear
Edited by bugbear (02/13/09 10:49 AM)
|
Charlie B
sage
Reged: 03/22/08
Posts: 264
Loc: Virginia
|
|
Interesting project! Try ISO 200 with 134mm f4. Also, download and use the free software Deep Sky Stacker here for this kind of stacking. Registrax is best for planets. DSS should be able to do a better job.
Charlie B.
--------------------
Meade SN-8, DS-90, AT-66
DSI Pro II (Schuler Filters), DSI-C, LPI, Canon XTi & modified 20d
AIP4WIN, Nebulosity, DSS, Registrax, GIMP
running on Dell 1420 with Vista
|
LLEEGE
Running out of Oxygen
Reged: 03/03/05
Posts: 9790
Loc: Cloud-chester,NY
|
|
I don't think there is any benefit to stacking JPEG images. It may reduce some of the random noise wont increase you SN ratio as the images are compressed. Can you shoot in RAW mode? I have heard of Canon hacks that allow some PS camera to shoot in RAW mode.
--------------------
"Okay! You draw the straws. I'm-a taking the parachute."
|
LLEEGE
Running out of Oxygen
Reged: 03/03/05
Posts: 9790
Loc: Cloud-chester,NY
|
|
And as Charlie said, Registax is geared to planetary imaging. You can download Deep Sky Stacker for free. There is also a program called star trails which is great for tripod mounted cameras. Startrails download
deep sky stacked download
--------------------
"Okay! You draw the straws. I'm-a taking the parachute."
|
Quads
sage
Reged: 09/07/08
Posts: 296
Loc: Central Sands, Wisconsin
|
|
These images of Orion were taken with my point and shoot SX100. I was zoomed in just enough to frame Orion. I haven't tried to zoom in on the sword yet, but will when the weather cooperates again. At that time I will piggyback the camera on my ETX, use about 10x zoom, and no longer than 10 second exposures. For the following pictures the camera was mounted on a regular photo tripod.
This first image is one 15 second exposure, ISO 400, f/3.2, resized for posting but otherwise unedited.
This second image is a stack of twenty-one 15 second exposures. (I think the bar at the bottom is my house roof, or heat from it?) Stacked with Deep Sky Stacker, light frames only. The camera takes dark frames itself and I didn't use any flats or bias frames. I used the Autosave.tif file that DSS makes. I increased the black level until there was fair contrast between stars and background. Increased the saturation a little. Then resized for posting.
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
Interesting project! Try ISO 200 with 134mm f4. Also, download and use the free software Deep Sky Stacker here for this kind of stacking. Registrax is best for planets. DSS should be able to do a better job.
Charlie B.
Yes, I gathered that- but DSS doesn't run under Linux, or even Wine.
As a passing aside, given the massive UNIX bias in academic astronomy (I'm thinking Sun/Solaris) for decades, I'm amazed that so much astro software is Windows, not *Nix now.
BugBear
Edited by bugbear (02/13/09 10:55 AM)
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
I don't think there is any benefit to stacking JPEG images. It may reduce some of the random noise wont increase you SN ratio as the images are compressed.
Not true at all, I'm happy to say.
Here's a single JPEG frame, multiplied by 70
I think you'll agree the stacked versions are happier :-)
Quote:
Can you shoot in RAW mode? I have heard of Canon hacks that allow some PS camera to shoot in RAW mode.
Yes - CHDK allows that mode. It slows down the transfer to the SD card a lot, and I've never had found a successful way to process the resulting file. dcraw "handles it" but the results look terrible.
Some analysis of the numbers:
(done via jpegtopnm IMG_5635.JPG | ppmtopgm | ppmhist -nomap)
lum count
----- -------
0 5898625
1 1955490
2 126303
3 7730
4 871
5 262
6 196
7 120
8 97
9 66
10 55
11 35
13 31
12 25
14 23
16 18
17 17
15 17
19 16
18 15
24 11
So by the time we're at color 24, there are only 11 pixels out of 8 million!!
It tails off after that... So I think each frame effectively is around 4 bits of damn noisy data.
BugBear
Edited by bugbear (02/13/09 11:07 AM)
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
These images of Orion were taken with my point and shoot SX100.
The 10x zoom should help! You'll get 2.5 times as many pixels as me.
BugBear
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
I should probably point out that the images I'm posting are NOT scaled down for web use - that's all the pixels I've got!
BugBear
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4497
Loc: Leeds, UK, 54N
|
|
Quote:
Not true at all, I'm happy to say.
Here's a single JPEG frame, multiplied by 70
And what it shows me is you have exposed a lot of nasty JPEG compression artefacts!
LLEEGE's advice is good advise. JPEG is a major handicap - but you should make some better progress. Registax is not a good solution for what you are trying to do for teh reasons given above.
Have you got GIMP? Manual stacking and using 100%, 50%, 33%, 25%, 20% etc opacity between the stacked layers achieves the proper stacking steps for an "average combine"
You will also do better keeping the in-camera noise reduction on. With JPEG format you won't be able to do any calibration with darks etc. Even if you could use DSS - it would just tell you it can't calibrate JPEGs! The reason is the camera does non linear processing when creating the JPeG and thus it impossible to match darks to the images you are taking. In camera noise reduction is applied before the non linear processing so you will get better results that way.
Quote:
As a passing aside, given the massive UNIX bias in academic astronomy (I'm thinking Sun/Solaris) for decades, I'm amazed that so much astro software is Windows, not *Nix now.
I'm not supprised at all. Academics write software for research purposes, for accuarate photometry, astrometry etc etc. Not much is done regarding creating pretty pictures from point abd shoot cameras!
However lots of ordinary joes have DLSR/P+S cameras and they have Windows machines - so there is plenty of demand for astrophotography software for Windows as a result. To answer your question - if there was sufficient demand you would see lots of linux astrophotography software.
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
Quote:
Not true at all, I'm happy to say.
Here's a single JPEG frame, multiplied by 70
And what it shows me is you have exposed a lot of nasty JPEG compression artefacts!
LLEEGE's advice is good advise. JPEG is a major handicap - but you should make some better progress. Registax is not a good solution for what you are trying to do for teh reasons given above.
Have you got GIMP? Manual stacking and using 100%, 50%, 33%, 25%, 20% etc opacity between the stacked layers achieves the proper stacking steps for an "average combine"
Can't work - with no tracking, my subjects move. I *need* alignment.
And I don't need Gimp - I'm an experienced programmer (I've even done image processing,for pre-press), and I'd already written a netpbm script (netpbm is GREAT) to do stacking...
Which worked right up until I found out how fast stars MOVE!
BugBear
|
Tonk
Post Laureate
Reged: 08/19/04
Posts: 4497
Loc: Leeds, UK, 54N
|
|
Quote:
I'm an experienced programmer
So am I  but the range of essential tools for processing astro images such as curves transforms, DDP etc etc etc is going to keep you busy on the script writing front  . I'm lazy and use the stuff written by Luc, Noel and Russel who you see occasionally on this forum
You should have fun with your approach. Keep at it and you will get results.
A real good tip is to create a barn door tracker. Two bits of wood, a hinge and a screw thread and you have a basic hand controlled tracker that will get you tracking wide field for a few minutes
--------------------
Televue 85, GM-8/Gemini, Canon 40D (unmodded), Canon 450D (modded w/Astronomiks clip-ins - UV/IR, OWB)
Coronado SM60/BF10, Baader Herschel Wedge
Leeds Sky Clock Ripon Sky Clock
|
Charlie B
sage
Reged: 03/22/08
Posts: 264
Loc: Virginia
|
|
IRAF is the package most used by the professional astronomy community and it runs on UNIX. The Hubble pictures we all admire were produced by tools built on IRAF. However, most professional astronomy only works with FITS files and there is a steep learning curve for the software.
You may want to look at ImageJ here. ImageJ is a package from NIH that is designed to do image processing for CAT scans. It also has astronomy plug-ins. There are some limitations, but, if you are a Java programmer, you should be able to do reasonably good at stacking. As been suggested, I believe I would try to save the photos as raw files and convert them to either TIFF of FITS files. The 12-bit format for the jpeg will limit your stacking results. You might do better with Registrax with TIFF or FITS. Registrax 5 is coming out soon and should handle TIFF files better.
Finally, there is Iris. It might run under wine, but I've never tried it.
Regards,
Charlie B
--------------------
Meade SN-8, DS-90, AT-66
DSI Pro II (Schuler Filters), DSI-C, LPI, Canon XTi & modified 20d
AIP4WIN, Nebulosity, DSS, Registrax, GIMP
running on Dell 1420 with Vista
|
DonR
scholastic sledgehammer
Reged: 11/15/06
Posts: 988
Loc: Georgia, USA
|
|
It looks look you're simply adding the pixel values from your sub-frames - try using a higher ISO and then a median combine instead. You need more data than 2 seconds at ISO 80 can provide.
I used Linux when I first started trying to process astro photos. Like you I was surprised that there was practically no software out there geared for that job, and everything I tried had issues with wine. I started using Windows under VMWare (the server version of VMWare is free if your flavor of Linux qualifies as a server OS), and settled on Iris software for processing astro images, which worked great. But then I found that there were issues with serial port control under VMWare that prevented me from controlling my mount, so finally I gave in and installed Windows as the primary OS.
Regarding the stars moving, most of us have to deal with that even with equatorial tracking mounts. I could track all night long with no pixel shifting if I wanted to, but that introduces problems with fixed pattern noise from the sensor, so I intentionally move the mount a little after every two or three frames. All of the astro specific software already mentioned in this thread handles that by registering the stars in the images before stacking. I have even done the registration manually in Photoshop, stacking layers and nudging them to achieve alignment before combining them. But with over a hundred layers, that would become very tedious.
--------------------
Don
----------
Atlas EQ-G
Orion 8" f/4.9 newtonian
Orion 127mm Mak-Cass
Orion Skyview Pro mount
Orion 80mm guide scope
Canon Digital Rebel XT
Meade DSI
Philips SPC 900 NC webcam
http://www.pbase.com/dtreed/astrophotos
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
As been suggested, I believe I would try to save the photos as raw files and convert them to either TIFF of FITS files. The 12-bit format for the jpeg will limit your stacking results. You might do better with Registrax with TIFF or FITS.
I've taken a set with RAW files. I'm and struggling to process them. Getting a colour transcription out of dcraw has so far evaded me, so I'm using:
dcraw -4 -D -t 0 -j
http://www.cybercom.net/~dcoffin/dcraw/dcraw.1.html
This gives me "pure" raw gray scale (with matrix artefacts).
I have discovered (by experimention) that the maxval in the output of this is 1023, which tallies with the D/A in my camera being 10 bit. I therefore multiply the data by 64 to bring it upto "full scale" 16 bit.
The resulting data looks pretty good: Here's the top of a histogram:
value count
==================
64 31
128 21
192 24
256 84
320 85
384 122
448 159
512 220
576 431
640 482
704 608
768 828
832 1402
896 1588
960 1663
1024 569
1088 2851
1152 3438
1216 4076
1280 4375
1344 7224
1408 5590
1472 8270
1536 9105
1600 12358
1664 11282
1728 12317
1792 12275
1856 16536
1920 14478
1984 18842
2048 14910
2112 17663
2176 14376
2240 14189
2304 26534
2368 11197
2432 10727
2496 7168
2560 9439
2624 7076
2688 5759
2752 4475
2816 4896
2880 3269
2944 3196
3008 961
3072 981
3136 1509
3200 1217
3264 873
3328 797
3392 444
3456 314
3520 261
3584 233
3648 141
3712 85
3776 60
3840 65
3904 38
3968 24
4032 16
4096 22
4160 6
4224 6
4288 13
4352 23
4416 2
4480 11
4544 7
4608 6
4672 11
Here the data scale so that 4672 is the maximum value (8 bit grayscale png for web use)
But I have so far failed to convert this data into anything that Registax will handle correctly. I've tried making both 8 and 16 bit grayscale/color PNG and TIFF frames, all to no avail.
Registax either crashes, or guess the colour/B+W wrongly.
All the frames I generate are readable in Gimp or Cinepaint (Gimp won't handle 16 bit data).
BugBear
Edited by bugbear (02/16/09 06:42 AM)
|
Charlie B
sage
Reged: 03/22/08
Posts: 264
Loc: Virginia
|
|
Quote:
dcraw -4 -D -t 0 -j
From reading the dcraw options, won't the -D option give a grayscale output?
Charlie B
--------------------
Meade SN-8, DS-90, AT-66
DSI Pro II (Schuler Filters), DSI-C, LPI, Canon XTi & modified 20d
AIP4WIN, Nebulosity, DSS, Registrax, GIMP
running on Dell 1420 with Vista
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
Quote:
dcraw -4 -D -t 0 -j
From reading the dcraw options, won't the -D option give a grayscale output?
Charlie B
Yes - I've completely failed to get sensible colours out of dcraw, so I stopped trying.
I (finally) got 16 bit frames that Registax would take (by generating all-channels the same colour from the dcraw output and the using a -force argument to pnmtopng).
pamcut -left 1517 -top 1149 -width 626 -height 584 $o | pamfunc -multiplier=64 | pgmtoppm white |pnmtopng -force > $tt
So here's the latest stacked shot:
By using raw, I've avoided the compression artefacts; in general it's not very pretty, but at least I've got some shape into the nebula - it's no longer just a bright "splat".
(add; shots were taken 2009:02:13 19:52:46, Diss, 52°23'N, 01°07'E, at the edge of a town)
BugBear
Edited by bugbear (02/16/09 07:36 AM)
|
Charlie B
sage
Reged: 03/22/08
Posts: 264
Loc: Virginia
|
|
Looks like it's coming along pretty good to me.
Regards,
Charlie B
--------------------
Meade SN-8, DS-90, AT-66
DSI Pro II (Schuler Filters), DSI-C, LPI, Canon XTi & modified 20d
AIP4WIN, Nebulosity, DSS, Registrax, GIMP
running on Dell 1420 with Vista
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Quote:
Looks like it's coming along pretty good to me.
Thanks, and thanks to all for the help. I tried dark frames (5 raw at each end of the sequence), but the result were worse, which seems odd.
I think that last shot is the best I'm going to get, at least until I take a longer frame sequence.
The conditions were also pretty favourable that night.
I find the results very satisfying, given that all I used was a compact camera I already had, a laptop, and free software.
BugBear
|
bugbear
member
Reged: 12/06/07
Posts: 39
|
|
Oh good grief! Nobody told me I look like a fool, and that "Jason H" has pretty much been there, done that, bought the T shirt...
http://www.cloudynights.com/ubbthreads/showflat.php?Cat=&Board=DSLR&Number=2662113&Forum=f85&Words=%2Bmechanical%20%2Btimer%20%2Bhome%20%2Bdepot&Searchpage=0&Limit=25&Main=345259&Search=true&where=bodysub&Name=&daterange=1&newerval=2&newertype=y&olderval=&oldertype=&bodyprev=1#Post2662113
1 canon P&S, 1 home depot timer, 1 wonderful shot of the Orion nebula!
BugBear
|
|
5 registered and 11 anonymous users are browsing this forum.
Moderator: Charlie Hein, knuklhdastnmr
Print Thread
|
Forum Permissions
You cannot start new topics
You cannot reply to topics
HTML is disabled
UBBCode is enabled
|
Thread views: 1112
|
|
|
|
|
|
|
Previous
Index
Next
Threaded
Edit
Reply
Quote
[Re: 
