|
Michal1
sage
Reged: 07/25/10
Loc: Czech Republic, Central Europe
|
computer processing of film images
#4249047 - 12/14/10 07:42 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I'm creating this thread so that we can deal with image software and processing procedures special for film astrophoto. What image processing software do you use? How do you process your pictures?
OK, I will start. Currently I use this combination of programs: Zoner - for most of the basic operations: curves, levels, resizing, channel mixing Image Analyzer (my loved ) - A small freeware program, but contains a number of usefull features rare in other applications. Unfortunatelly it works internally only with 8-bit per color. This program has it's own forum, where you can ask it's developer and suggest new features. Matlab - I use it for everything what I can't do with another software. It's usually a lot of work. In fact it is programming of own software. DenoiseMyImage - A dedicated denoising freeware appplication. It's said to be one of the best. Registar - A well known program for celestial image registration and stacking. I don't like Photoshop (I'm the only one in the world?). Mainly because it's a gigantic software with only several functions usefull for me (which I can do with Zoner). I mind I can't understand the principle of it's functions (what I can in Image Analyzer).
I promissed to write my method for background subtraction in Image Analyzer. It works only for pictures without extended objects like the Milky Way or large nebulae. It in two main steps - extracting the bakground and proper subtraction. I will describe the process in the latest version of IA (1.32). * Open the picture to be flattened and make a copy: File-Open, Edit-Copy & Edit-Paste * Deletion of smaller stars: Operation-Morphological operations-Erosion. Repeat typically 2-4 times. * Deletion of remaining stars, dark speckles and larger objects by clonning: Tool selection button-Clone. Set the appropiate size by the button aside. * The image brightness decreased by the erosion. It can be restored by dilatation: Operation-Morphological operations-Dilatation. Reapeat to reach the original background brightness. * Delete the renewed stars by clonning. * Blur the background: Operations-Filters-Smooth & set the blurr radius. It must be the smallest possible to perserve all the background features but all the grain and star remnants must disappear. A good starting point is usually about 60. * Now comes the background subtraction. It is done via the dialog: Operations-Pixel format conversion-Combine images-User expression. You need to select 'p1' as the backround image and 'p2' the original. Now is needed to enter the combining mathematical expression (no restrictions here!!). Who don't want to experiment or don't like maths simply copy this formula
(p2-0.95*p1)/(256-0.95*p1)*236+20
which gives usually a good result and IA will remember it for the future. If want to know more about this formula read further. This method has no "physical" foundation, but it works. The basic idea is to subtract the background but save the values of bright stellar centres at 255 and at the same time set the values at the bakground level to 0. The values in between are uniformly stretched to this range. It leads along to the basic formula (p2-p1)/(255-p1)*255. The problem is the program is worried about possible division by zero, so 255 has to be changed to 256. A fully subtracted image looks offten boring flat, that's why I subtract the background only partially. Sutracting of e.g. 95% of the background is implemented by replacing p1 by 0.95*p1. The next bug is the part of original's backround is lower than the smoothed backround so the resulting value is lower than zero and is truncated to 0. To prevent this, you need to add a constant e.g. 20 and multiply by 236(=256-20) instead of 256. So this is that magic formula.
|
Rick Thurmond
sage
Reged: 10/30/06
Loc: California
|
Re: computer processing of film images
[Re: Michal1]
#4249611 - 12/15/10 02:07 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Michal, Thanks for starting the topic. I'm the other one in the world who doesn't like Photoshop, but I'm finding it useful. I've been having Dave align images for me because he has Registar and I don't because I have a Mac. That program does an excellent job. As soon as I get back an aligned image, I change the color profile to BestRGB so I can make adjustments to the individual channels without them interfering with each other. Then I use Curves layers to brighten or darken parts of the images. I always do it on each channel individually. Once I've done all I can in Photoshop I move the image into Aperture where I can do some final color tweaking and cropping, the I export a low-res version for the web and a high-res version to print. Rick
|
Dave Kodama
super member
Reged: 10/04/10
Loc: CA, USA
|
Re: computer processing of film images
[Re: Rick Thurmond]
#4251678 - 12/16/10 04:09 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I use Photoshop too. My main problem with it is the high cost of the software and upgrades. This is true of all of the Adobe products. These days I'm spending more money for Adobe software than any other software.
However, Photoshop is so dominant that it's hard to get along without using it. All of the books, tutorials, videos, and conference talks assume the use of Photoshop and I've found the GradientXterminator add-in to be very useful. In addition, photoshop has proven to be quite reliable even on very large files. I've used it on files as large as 3GB without experiencing a crash. But I certainly agree with Michal that the lack of information about what the filtering routines are doing has annoyed me right from the start.
I also use Registar for image stacking and creating mosaics. However, it is only available in a 32-bit version and will run out of memory on large files. I can stack fully overlapping 4x5 images scanned at 2400 dpi, but will fail to register them if there is a lot of non-overlapping area due to its memory limition. As far as I know the authors do not plan to upgrade it to 64-bit operation (I asked last year).
And finally, I have found Adobe Lightroom to be essential for managing collections of images and doing the "finishing" work on images for the web and printing.
I haven't used any sophisticated digital image processing type programs since I don't do any analytical work such as astrometry with my images.
Dave
|
Michal1
sage
Reged: 07/25/10
Loc: Czech Republic, Central Europe
|
Re: computer processing of film images
[Re: Dave Kodama]
#4509635 - 04/10/11 06:28 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Is there any free software which can be used for backround subtraction for 48-bit images? All of my software fails: Image Analyzer works with 24-bit images only, Matlab is slow and runes out of memory with my images and Iris seems to generate step-like artificial background. I can't find a CCD software which could create artificial bakground. All of them are designed to straightly remove (divide or subtract?) the background. They don't take into account the nonlinearity of film. I know there was the Pleiades program suitable for our purposes, but it can't be found to download now. Thanks for any suggestions. Michal
|
Nebhunter
Carpal Tunnel
Reged: 10/04/03
Loc: Frostbite Falls
|
Re: computer processing of film images
[Re: Michal1]
#4509907 - 04/10/11 09:15 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I'm waiting for Jim to pipe in as he uses Pix in Sight mostly. I have the early version which is no longer available for download. Take a look at their website and visit the tutorial page. It gives an excellent view of how, and what happens with each step. http://pixinsight.com/examples/index.html
Here you can read how it does the dynamic extraction - but with the older version. http://pixinsight.com/tutorials/obsolete-LE/DBE-example/en.html
The fact that this program was developed specifically for astrophotography says a lot to me. When I look at many of the images with CCD on long exposures - the good ones - I often find Pix in Sight listed as one of the tools.
I went from CS2 - to 4 as my Mac upgrade to the newer OS prevented CS2 from working. There is no way I'm spending more money on going to version 5. I use is for working my landscape photos, but my new 9000f scanner is so good that I'm doing next to no post processing. So a basic program would probably serve me fine.
I will probably go with the new version of Pix in Sight. Learn how to use it with the tutorials - and get the best from my work that I can. Poking around in CS4 is just frustrating. Gads, did I just type all that? What was the question again? Going to walk the dogs.
igor
|
Nightfly
scholastic sledgehammer
Reged: 06/20/07
Loc: Eastern Maine, USA
|
Re: computer processing of film images
[Re: Nebhunter]
#4509946 - 04/10/11 09:33 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I use PixInsight mostly for the DBE and Green Gun. I curve and level in PS. The DBE is a powerful tool and allows one to shoot wide open and take care of the vignetting later. Busy star fields and Milky Way shots make the DBE operation harder of course.
That being said, I use the old LE version that in no longer downloadable. If someone still have the original installable .EXE or .zip file that would get you there.
Jim
|
Hikari
scholastic sledgehammer
   
Reged: 01/05/11
Loc: Maine, USA
|
Re: computer processing of film images
[Re: Nightfly]
#4510048 - 04/10/11 10:29 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
You could try ImageJ which is free and designed for microscope images. There is also a pdf manual which you will need to find out if it can do what you want, but it will process 16-bit files.
|
Michal1
sage
Reged: 07/25/10
Loc: Czech Republic, Central Europe
|
Re: computer processing of film images
[Re: Hikari]
#4549954 - 04/28/11 03:37 PM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
I found the install package of the old free version of PixInsight along with the upgrading package to the version 1.0.2. Who miss this software, send me a PM.
|
okiestarman56
professor emeritus
   
Reged: 04/10/08
Loc: 36.9'33" N 95.23'29" W
|
Re: computer processing of film images
[Re: Michal1]
#4550811 - 04/29/11 12:24 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
wow this thread is going to be better than peanut butter, digital is fine but film is some thing more, I guess it's because I've shot film for 30 years and loved every minute of it.
|
Nebhunter
Carpal Tunnel
Reged: 10/04/03
Loc: Frostbite Falls
|
Re: computer processing of film images
[Re: okiestarman56]
#4556313 - 05/02/11 12:35 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
Glad to have you here. Pull up a chair and join in.
igor
|
Michal1
sage
Reged: 07/25/10
Loc: Czech Republic, Central Europe
|
Re: computer processing of film images
[Re: Nebhunter]
#5269156 - 06/13/12 08:27 AM
|
Edit
|
Reply
|
Quote |
Quick Reply
|
|
|
OK, a little of activity from me. I would like to tell you about IDRMyImage program, which I found to be very interesting and useful for me. I've used the program to enhance structures of Milky Way in most of my photos. It is able to visualize various dark lanes and star cloulds otherwise hardly noticeable. You can see an example of its work in the following pair of pictures:
processed
original
Several remarks. Its most important function is the Smart contrast button whose effect is similar to wavelet filter function of other programs but this one seems to suit better to MW. I've used the Single image IDR buttons ocassinaly. They allow one to decrease the difference between very bright and very dim parts of the image. I've used it e.g. to make this picture (along with smart contrast and saturation increase) when I wanted to make the faint nebulae visible:
It takes it 20min to process my 23Mpix images (scans of 6x7 frames at 2000 dpi). It would be too long to wait this time again after every tweak of parameters to see the effect. The solution is to cut several selections from the image and to find the best parameters on them. Then I open the full image and apply the settings on it. It is important not to save the result as jpeg or you would notice terrible decrease of quality. The last version of IDRMyImage 2.0 can load and save 48bit images, too. However, if I run the program on one of my images in 48 bit depth, it crashes, so I have to reduce the bit depth to 24 bit. I've found to be most appropriate to apply Smart contrast as the last adjustment after curves, saturation, denoising, etc., because it seems to add artificial gradients to the image. Every image needs a special setup of the smart contrast parameters. Some images put up with strong adjustments and the result looks fine, but there are also others which are made rather worse than before by the program. If you use the program, be aware not to make artificially looking images.
M.
|
|
3 registered and 3 anonymous users are browsing this forum.
Moderator: THEPLOUGH
Print Thread
|
Forum Permissions
You cannot start new topics
You cannot reply to topics
HTML is disabled
UBBCode is enabled
|
Thread views: 2326
|
|
|
|
|
|
|