Pixinsight - lite tips
Posted: 2012-11-20 07:24:14
God Morgon denna gråa tisdag!
Jag skrev till en herre på ett engelsk forum om hur man gör en riktigt snabb runda i Pixinsight med en färgbild. Den blev så uppskattad att jag tänkte att man kan väl publicera den här för er som vill komma igång med denna fantastiska programvara. Min egen uppfattning om PI är att den slår allt men saknar lagerhantering, något som gör Photoshop till ett väl så viktigt verktyg i slutändan. Man kan också se PI som görande Photoshop Elements till ett användbart verktyg för astrofoto eftersom man klarar sig med åttabitarsfärg om man har gjort den svåra och dynamikkrävande delen av jobbet i PI.
Nåväl, ni klarar väl engelska så jag lägger in den som den är och hoppas på forumkompatibilitet!
/////////////////////////
Paul,
Pixinsight has two gradient removal tools, one automatic and one manual. The former does the job in 95% of the cases and it is called Automatic Background Extraction (ABE). You start by just applying it with default values and it will give you an image that constitutes its perception of the background. Autostretch that and you see it clearly. If it correlates to what your brain/eye combination deems to be the background, then you chose "subtract" as operation, check "discard background image" and check "replace target image". Throw away the first version of the background image and re-apply the ABE. Presto!
Here is an example of some experimental M26C frames of M33, fresh out of image integration. Note that a few of the screens dumps are wide and show a dialog box to the right!
Stack, auto-stretched with screen transfer function (last button in the first group of buttons at the top):

ABE applied with default settings generates this background (again auto-stretched with screen transfer function):

ABE set up to actually do something about the problem:

And here is the auto-stretched result of ABE:

There is some gradient left (which we could have fixed with the settings (too much work, eh?
), so we'll try ABE one more time resulting in this background:

Applied, the image now looks like this:

Crop it with Dynamic Crop:

Next, a trick that I have learned. Fiddle with the screen transfer function (STF) starting with the A-button for autostretch. Reduce the stretch somewhat, move the black-point up just a notch until you like what you see. Note you can zoom into the sliders of STF by pressing the + button and then clicking in the slider area. Then click the arrow to make the adjustments. After that, take the STF triangle button and drop it in the bottom of a reset Histogram Transformation window. That will move the auto-stretch parameters adjusted by you to the histogram, like this:

Aplly the histogram, reset the STF, and here's a decent image:

Not many steps and not a finished image, but it is good enough for data judgement and decisions as to what can be done with it.
The STF has a magic auto-stretch that makes very strange adjustments to the curve, something that you cannot do on your own. Therefore, the auto-stretch is a good place to start your histogram stretch and get it done in just two steps: auto, then fiddle. The fact that the apply-curve can be moved to Histogram Transformation and actually applied to the image is great. I assume you know that STF only affects what you see, not the image itself.
/per
Jag skrev till en herre på ett engelsk forum om hur man gör en riktigt snabb runda i Pixinsight med en färgbild. Den blev så uppskattad att jag tänkte att man kan väl publicera den här för er som vill komma igång med denna fantastiska programvara. Min egen uppfattning om PI är att den slår allt men saknar lagerhantering, något som gör Photoshop till ett väl så viktigt verktyg i slutändan. Man kan också se PI som görande Photoshop Elements till ett användbart verktyg för astrofoto eftersom man klarar sig med åttabitarsfärg om man har gjort den svåra och dynamikkrävande delen av jobbet i PI.
Nåväl, ni klarar väl engelska så jag lägger in den som den är och hoppas på forumkompatibilitet!
/////////////////////////
Paul,
Pixinsight has two gradient removal tools, one automatic and one manual. The former does the job in 95% of the cases and it is called Automatic Background Extraction (ABE). You start by just applying it with default values and it will give you an image that constitutes its perception of the background. Autostretch that and you see it clearly. If it correlates to what your brain/eye combination deems to be the background, then you chose "subtract" as operation, check "discard background image" and check "replace target image". Throw away the first version of the background image and re-apply the ABE. Presto!
Here is an example of some experimental M26C frames of M33, fresh out of image integration. Note that a few of the screens dumps are wide and show a dialog box to the right!
Stack, auto-stretched with screen transfer function (last button in the first group of buttons at the top):
ABE applied with default settings generates this background (again auto-stretched with screen transfer function):
ABE set up to actually do something about the problem:
And here is the auto-stretched result of ABE:
There is some gradient left (which we could have fixed with the settings (too much work, eh?

Applied, the image now looks like this:
Crop it with Dynamic Crop:
Next, a trick that I have learned. Fiddle with the screen transfer function (STF) starting with the A-button for autostretch. Reduce the stretch somewhat, move the black-point up just a notch until you like what you see. Note you can zoom into the sliders of STF by pressing the + button and then clicking in the slider area. Then click the arrow to make the adjustments. After that, take the STF triangle button and drop it in the bottom of a reset Histogram Transformation window. That will move the auto-stretch parameters adjusted by you to the histogram, like this:
Aplly the histogram, reset the STF, and here's a decent image:
Not many steps and not a finished image, but it is good enough for data judgement and decisions as to what can be done with it.
The STF has a magic auto-stretch that makes very strange adjustments to the curve, something that you cannot do on your own. Therefore, the auto-stretch is a good place to start your histogram stretch and get it done in just two steps: auto, then fiddle. The fact that the apply-curve can be moved to Histogram Transformation and actually applied to the image is great. I assume you know that STF only affects what you see, not the image itself.
/per