was bewirkt sie genau ? Geht das mit allen Digiknipsen und benoetigt man dazu eine spezielle Software ?
Gruss Heiko
Wie stellt man eine "darkframe subtraction" her und
-
Markus Pfarr
Re: Wie stellt man eine "darkframe subtraction" her und
> was bewirkt sie genau ? Geht das mit allen Digiknipsen und
> benoetigt man dazu eine spezielle Software ?
> Gruss Heiko
Hallo Heiko.
Bei der Darkframe-Subtraction wird der Dunkelstrom oder auch das Rauschen des CCD-Chips von dem eigentlichen Bild enfernt.
Man macht also ein Bild z.B. Belichtungszeit 30 sec. und sofort darauf macht man das so genannte Darkframe. Dabei muß das Objektiv
absolut dicht verschlossen werden und die Belichtungszeit muß gleich
mit der des vorher gemachten Bildes sein, hier z.B. 30 sec.
Dafür gibt es spezielle Astrosoftware wie z.B. Astroart oder Quantum Image. Theoretisch geht das mit allen Digiknipsen, wenn die Belichtungszeit manuell einstellbar ist.
Gruß
Markus http://www.sternwarte-moembris.de
> benoetigt man dazu eine spezielle Software ?
> Gruss Heiko
Hallo Heiko.
Bei der Darkframe-Subtraction wird der Dunkelstrom oder auch das Rauschen des CCD-Chips von dem eigentlichen Bild enfernt.
Man macht also ein Bild z.B. Belichtungszeit 30 sec. und sofort darauf macht man das so genannte Darkframe. Dabei muß das Objektiv
absolut dicht verschlossen werden und die Belichtungszeit muß gleich
mit der des vorher gemachten Bildes sein, hier z.B. 30 sec.
Dafür gibt es spezielle Astrosoftware wie z.B. Astroart oder Quantum Image. Theoretisch geht das mit allen Digiknipsen, wenn die Belichtungszeit manuell einstellbar ist.
Gruß
Markus http://www.sternwarte-moembris.de
-
Bernd
Re: Wie stellt man eine "darkframe subtraction" her und
Hi Heiko,
anbei ein Artikel, den ich mir mal zum Thema Rauschen irgendwo aus dem Netz gezogen habe. Darin steht auch, wie man das Rauschen ohne spezielle Anti-Rausch-Software, sondern ganz einfach mit Adobe Photoshop los wird.
Grüße,
Bernd
Night Spots
By Peter iNova
One of the great things about film is that you can open the lens and soak in all those photons until the cows come home (assuming that they've been out all night) and build up that image over time. In fact Niépce (say "neeps") made the first permanent photo in 1827 with an 8-hour exposure out his bedroom window. All day long. But it worked! So time exposures have been around longer than short exposures. Er, if you get my drift.
How about these chip-cameras? No such easy path. The longer you keep that shutter open, the longer those sensors have to reveal their flaws. Noise, which looks like film grain, sort of, will build up. The picture gets covered with stars. You can hardly tell the image from all the speckles. Ugh!
This is what engineers call "fixed pattern noise" and if you want a readout on that go to Google.com.
***hier fehlt eine komplizierte Formel, die ich nicht in den Originaltext einfügen kann, da sie grieschiche Buchstaben enthält***
Standard deviation of fixed pattern noise. No, really.
Noise that stands still has an interesting quality. It stands still. Not like the snow you see on TV from between-channel noise-and some digital cameras like the Canon G1 and Sony DSC-S75 have a special routine inside that cancels it out. But did you know that you could use ANY digital camera and eliminate most of this fixed pattern noise? All you have to do is take a picture of nothin'.
Some of my most useful shots are of Nothing At All. Here's how it works: When you shoot an image with a long exposure-usually from 1/2 sec or longer-put the lens cap back on the camera and shoot another exposure with the same shutter speed. You might have to be in a manual mode to do that, or perhaps you can lock the shutter speed in your camera for purposes of shooting identical exposures for panoramics. Whatever it takes to match shutter speeds when no light is available.
This black frame gets a new name: Call it a Flaw Frame. Now you need a program like Photoshop 5 or higher or the new Photoshop Elements. Both have the tools you need.
Now you have a picture covered in speckles and a blank frame covered in the same pattern of speckles. Fixed pattern noise isn't permanent. You will have to gather that Flaw Frame very close to the time you made the real picture. But save them. Someday you may need to try to recycle one just to see if it fixes something, but I digress.
***hier fehlt das Bild von einer stark verrauschten Nachtszene***
Small slice of a night shot complete with flaws.
Nikon 990. 2 sec exposure.
***hier fehlt das Bild, das nur das Rauschen vor schwarzem Hintergrund zeigt***
Flaw Frame captured within a minute of the previous image. Now you can see the flaws much clearer without the building spoiling the shot.
The concept of how these two images work together is this: The Flaw Frame acts as a trigger to steer color into the blown pixels and noise of the color image. The colors are the ones that surround the blown pixels. How you do that is the trick.
Lay the Flaw Frame over the original image as a sandwich of layers. They have to be the same size and orientation. They can¹t be out of register, but if they were both gathered within moments of each other, switching the flaw frame on and off will confirm that the blown pixels are there in the same place on the photo.
Use the tools that turn the Flaw Frame into a selection based on brightness. Copy the Flaw Frame layer and paste it into a new Channel on the Channels Palette. Now Command-click (Macintosh) or Ctrl-click (Windows) on that fourth channel. Photoshop assumes you are making an "alpha" channel selection and any pixel that isn¹t zero brightness is proportionally selected.
Photoshop will let you do all this, but PSE needs a different approach. (This also works in the full Photoshop.)
With Photoshop Elements no Channels Palette is available. So you do this: Using the Hue/Saturation control, drain all color out of the Flaw Frame layer by sliding the Saturation control all the way to the left. Then using the Levels Control, slide the right-most triangle, the one that determines which value equals pure white on the image, to about the middle of the histogram. You have just increased the contrast of the image without lifting the black values. Handy, that.
Now you use the selection tool (looks like a magic wand) to pick one of the black pixels of the Flaw Frame. Almost all of the image is selected. With the Invert Selection command (Shift-Command-I = Macintosh; Shift-Ctrl-I = Windows) the black pixels are ignored but the selection now goes to the brighter pixels, the ones that flawed the original image.
Sounds complicated, but it¹s really fast. You will see the "marching ants" effect on only the brightest flaws. Go back to the Layers Palette if you aren¹t already there.
While this selection is active, click the visibility of the layer off. The marching ants seem now to be on the picture layer. Make the picture layer active then go to the Filters menu and select Noise > Dust and Scratches. Make the Radius 1 to 3 pixels and the Threshold zero. The preview window can be scrolled over your image so you can see the effect at work. In Photoshop you can even make the marching ants invisible with a Command or Control H.
***hier fehlt das Bild, das die gleiche Szene wie oben unverrauscht zeigt***
Where did the noise go? And why didn't that one pixel in the sky disappear? See text.
Hey, where did the noise go? All the flaws were just filled in with color that equals the average of the surrounding 8 or 24 pixels. For nearly every one of the flaws, this simply makes them go away and stay away. Only the strongest, toughest flaws remain. And even those are greatly diminished.
What about images that had things in them like real specks? I shot this building at night with a 2-second exposure and you can see its flaws clearly. But after running this procedure on the image, one prominent flaw wouldn't go away. But there was a good reason. It was a star. It would have disappeared if-and only if-it had been covered over with a flaw in the original shot.
The cameras that do this automatically do it by gathering a Flaw Frame right after the main exposure. You watch, you can't shoot another picture until they are through operating on the original shot. Very noticeable if it was an 8-second exposure.
I wrote a bunch of Photoshop Actions that are in the iNovaFX PS Action Filter set in my Nikon eBook which do this very thing in over a half-dozen different ways. Some fix the image in alternate ways and repair things that can show up in odd circumstances. With a choice of methods you can finesse a shot and select the method that works best for it.
But, that's a different story.
-Peter iNova
Creative Director, Metavision
anbei ein Artikel, den ich mir mal zum Thema Rauschen irgendwo aus dem Netz gezogen habe. Darin steht auch, wie man das Rauschen ohne spezielle Anti-Rausch-Software, sondern ganz einfach mit Adobe Photoshop los wird.
Grüße,
Bernd
Night Spots
By Peter iNova
One of the great things about film is that you can open the lens and soak in all those photons until the cows come home (assuming that they've been out all night) and build up that image over time. In fact Niépce (say "neeps") made the first permanent photo in 1827 with an 8-hour exposure out his bedroom window. All day long. But it worked! So time exposures have been around longer than short exposures. Er, if you get my drift.
How about these chip-cameras? No such easy path. The longer you keep that shutter open, the longer those sensors have to reveal their flaws. Noise, which looks like film grain, sort of, will build up. The picture gets covered with stars. You can hardly tell the image from all the speckles. Ugh!
This is what engineers call "fixed pattern noise" and if you want a readout on that go to Google.com.
***hier fehlt eine komplizierte Formel, die ich nicht in den Originaltext einfügen kann, da sie grieschiche Buchstaben enthält***
Standard deviation of fixed pattern noise. No, really.
Noise that stands still has an interesting quality. It stands still. Not like the snow you see on TV from between-channel noise-and some digital cameras like the Canon G1 and Sony DSC-S75 have a special routine inside that cancels it out. But did you know that you could use ANY digital camera and eliminate most of this fixed pattern noise? All you have to do is take a picture of nothin'.
Some of my most useful shots are of Nothing At All. Here's how it works: When you shoot an image with a long exposure-usually from 1/2 sec or longer-put the lens cap back on the camera and shoot another exposure with the same shutter speed. You might have to be in a manual mode to do that, or perhaps you can lock the shutter speed in your camera for purposes of shooting identical exposures for panoramics. Whatever it takes to match shutter speeds when no light is available.
This black frame gets a new name: Call it a Flaw Frame. Now you need a program like Photoshop 5 or higher or the new Photoshop Elements. Both have the tools you need.
Now you have a picture covered in speckles and a blank frame covered in the same pattern of speckles. Fixed pattern noise isn't permanent. You will have to gather that Flaw Frame very close to the time you made the real picture. But save them. Someday you may need to try to recycle one just to see if it fixes something, but I digress.
***hier fehlt das Bild von einer stark verrauschten Nachtszene***
Small slice of a night shot complete with flaws.
Nikon 990. 2 sec exposure.
***hier fehlt das Bild, das nur das Rauschen vor schwarzem Hintergrund zeigt***
Flaw Frame captured within a minute of the previous image. Now you can see the flaws much clearer without the building spoiling the shot.
The concept of how these two images work together is this: The Flaw Frame acts as a trigger to steer color into the blown pixels and noise of the color image. The colors are the ones that surround the blown pixels. How you do that is the trick.
Lay the Flaw Frame over the original image as a sandwich of layers. They have to be the same size and orientation. They can¹t be out of register, but if they were both gathered within moments of each other, switching the flaw frame on and off will confirm that the blown pixels are there in the same place on the photo.
Use the tools that turn the Flaw Frame into a selection based on brightness. Copy the Flaw Frame layer and paste it into a new Channel on the Channels Palette. Now Command-click (Macintosh) or Ctrl-click (Windows) on that fourth channel. Photoshop assumes you are making an "alpha" channel selection and any pixel that isn¹t zero brightness is proportionally selected.
Photoshop will let you do all this, but PSE needs a different approach. (This also works in the full Photoshop.)
With Photoshop Elements no Channels Palette is available. So you do this: Using the Hue/Saturation control, drain all color out of the Flaw Frame layer by sliding the Saturation control all the way to the left. Then using the Levels Control, slide the right-most triangle, the one that determines which value equals pure white on the image, to about the middle of the histogram. You have just increased the contrast of the image without lifting the black values. Handy, that.
Now you use the selection tool (looks like a magic wand) to pick one of the black pixels of the Flaw Frame. Almost all of the image is selected. With the Invert Selection command (Shift-Command-I = Macintosh; Shift-Ctrl-I = Windows) the black pixels are ignored but the selection now goes to the brighter pixels, the ones that flawed the original image.
Sounds complicated, but it¹s really fast. You will see the "marching ants" effect on only the brightest flaws. Go back to the Layers Palette if you aren¹t already there.
While this selection is active, click the visibility of the layer off. The marching ants seem now to be on the picture layer. Make the picture layer active then go to the Filters menu and select Noise > Dust and Scratches. Make the Radius 1 to 3 pixels and the Threshold zero. The preview window can be scrolled over your image so you can see the effect at work. In Photoshop you can even make the marching ants invisible with a Command or Control H.
***hier fehlt das Bild, das die gleiche Szene wie oben unverrauscht zeigt***
Where did the noise go? And why didn't that one pixel in the sky disappear? See text.
Hey, where did the noise go? All the flaws were just filled in with color that equals the average of the surrounding 8 or 24 pixels. For nearly every one of the flaws, this simply makes them go away and stay away. Only the strongest, toughest flaws remain. And even those are greatly diminished.
What about images that had things in them like real specks? I shot this building at night with a 2-second exposure and you can see its flaws clearly. But after running this procedure on the image, one prominent flaw wouldn't go away. But there was a good reason. It was a star. It would have disappeared if-and only if-it had been covered over with a flaw in the original shot.
The cameras that do this automatically do it by gathering a Flaw Frame right after the main exposure. You watch, you can't shoot another picture until they are through operating on the original shot. Very noticeable if it was an 8-second exposure.
I wrote a bunch of Photoshop Actions that are in the iNovaFX PS Action Filter set in my Nikon eBook which do this very thing in over a half-dozen different ways. Some fix the image in alternate ways and repair things that can show up in odd circumstances. With a choice of methods you can finesse a shot and select the method that works best for it.
But, that's a different story.
-Peter iNova
Creative Director, Metavision
-
Klaus Bagschik
Ergänzung (reine Korinthenk.....ei von mir ;-)
> This is what engineers call "fixed pattern noise" and
> if you want a readout on that go to Google.com.
Mit diesem Begriff (und seiner Definition) braucht Ihr Euch hier garnicht zu belasten -- ist vielleicht nett das mal gehört zu haben, darf aber (in diesem Zusammenhang) gerne wieder vergessen werden. Der lästige Dunkelstrom der Digiknipsbilder nach kurzer Zeit mit vielen bunten Sternchen überzieht ist (leider !!!) _kein_ "fixed pattern noise".
Wenn dem so wäre bräuchte man im Laufe einer längeren Belichtungsserie nicht so oft neue Dunkelbilder zu machen. Immerhin ändert sich das thermische Rauschen aber so langsam, dass es sich zwischen "richtigem" Bild und Dunkelbild noch nicht sehr stark geändert hat.
Z.B. hat nach etwa einer Stunde pausenloser 5-sec Belichterei das CCD meiner Knipse (Oly E-100RS) wohl (fast) seine Endtemperatur erreicht -- _dann_ kann ich es sogar wagen nur noch alle 10-15 min ein neues Dunkelbild zu machen anstatt alle 1-2 min wie am Anfang ...
ciao
Klaus
> if you want a readout on that go to Google.com.
Mit diesem Begriff (und seiner Definition) braucht Ihr Euch hier garnicht zu belasten -- ist vielleicht nett das mal gehört zu haben, darf aber (in diesem Zusammenhang) gerne wieder vergessen werden. Der lästige Dunkelstrom der Digiknipsbilder nach kurzer Zeit mit vielen bunten Sternchen überzieht ist (leider !!!) _kein_ "fixed pattern noise".
Wenn dem so wäre bräuchte man im Laufe einer längeren Belichtungsserie nicht so oft neue Dunkelbilder zu machen. Immerhin ändert sich das thermische Rauschen aber so langsam, dass es sich zwischen "richtigem" Bild und Dunkelbild noch nicht sehr stark geändert hat.
Z.B. hat nach etwa einer Stunde pausenloser 5-sec Belichterei das CCD meiner Knipse (Oly E-100RS) wohl (fast) seine Endtemperatur erreicht -- _dann_ kann ich es sogar wagen nur noch alle 10-15 min ein neues Dunkelbild zu machen anstatt alle 1-2 min wie am Anfang ...
ciao
Klaus
Wer ist online?
Mitglieder in diesem Forum: 0 Mitglieder und 2 Gäste