Login  Register

Re: chromatic aberattion

Posted by Tomáš Procházka-2 on Jun 14, 2008; 5:38pm
URL: http://imagej.273.s1.nabble.com/chromatic-aberattion-tp3695678p3695681.html

Hi,

G> I wonder if scaling the image x2 and using interpolation, align and scale back
G> to 0.5) would make any difference (as it would allow to position the planes
G> at 1/2 pixel locations of the original image). Not sure if this would be
G> useful or just blur the original.

Thanks you, this is good idea.

>> But now I need detect edge, smooth, create mask from image and remove
>> fringe under mask.

G> I do not understand what you mean here.

Look at this pictures:

http://www.atomsoft.cz/public/archive_public/foto_100fs/aberration_remove/step1_original.jpg
http://www.atomsoft.cz/public/archive_public/foto_100fs/aberration_remove/step2_LightRoom.jpg

The same result as last image I get from ImageJ. In image still exist purple/violet fringe. I need localize it and remove. I think, that best way is found edges (dark-white crossing) and desaturate image under edges or remove only purple/violet fringe.

Here is example how DxO Optixs Pro do it:

http://www.atomsoft.cz/public/archive_public/foto_100fs/aberration_remove/step3_DxO.jpg

   
-------------------------- Original message --------------------------
       From: Gabriel Landini <[hidden email]>
    Subject: chromatic aberattion
       Date: Saturday, June 14, 2008, 12:17:34 PM
Attachments: <none>
      msgid:[hidden email]

G> On Friday 13 June 2008, Tomáš Procházka wrote:
>> I maybe solved first part of problem.

>> with this

>> run("RGB Stack");
>> run("Scale...", "x=.99919 y=.99919 title=Untitled");
>> setSlice(3);
>> run("Scale...", "x=0.999 y=0.999 title=Untitled");
>> run("RGB Color");

G> Hi,
G> Yes this may work nicely, but note that there is a limit of the accuracy of
G> the Scale command. I do not think that you will be able to scale the plane by
G> less than 1 pixel steps (i.e. target plane size/current plane size will be
G> stepped will be limited to the corresponding nearest 1 pixel values).

G> There will be a also limit to what one can correct as not all the colours will
G> be present in a unique rgb plane.

G> I wonder if scaling the image x2 and using interpolation, align and scale back
G> to 0.5) would make any difference (as it would allow to position the planes
G> at 1/2 pixel locations of the original image). Not sure if this would be
G> useful or just blur the original.


>> But now I need detect edge, smooth, create mask from image and remove
>> fringe under mask.

G> I do not understand what you mean here.

G> Regards,
G> G.


----------------------- End of original message ----------------------

--
 Tomas Prochazka