Login  Register

Re: Import Image Sequence Contrast Bug (DM3)

Posted by Michael Schmid on Sep 29, 2014; 5:27pm
URL: http://imagej.273.s1.nabble.com/Import-Image-Sequence-Contrast-Bug-DM3-tp5009816p5009817.html

Hi Thorsten,

if you convert from 16 to 8 bits you can select in Edit>Options>Conversions:

"Scaled when converting"=off means that the 16-bit value will be transferred 1:1 to the 8-bit value, and everything above 255 will become 255 in the 8-bit image. Typically, this will lead to a lot of saturated pixels, so one would divide the 16-bit image by some number before conversion to 8 bits in that case, to get a reasonable 8-bit display.

"Scaled when converting"=on means that the current display range (as set in the Image>Adjust>Brightness&Contrast panel) determines the span from 0 to 255 of the 8-bit image.

If you have "Scaled when converting"=on and different B&C settings when converting, you get different results.

See the manual:

http://rsb.info.nih.gov/ij/docs/guide/146-28.html#toc-Subsection-28.1
http://rsb.info.nih.gov/ij/docs/guide/146-27.html#toc-Subsubsection-27.13.11


Michael
________________________________________________________________
On Sep 29, 2014, at 18:42, Thorsten Wagner wrote:

> Hi,
>
> I currently working with TEM image data. I imported a series of dm3
> images via the Import -> Image Sequence command.
>
> If I open a single image of that image series with Open -> Image it
> seems to have another contrast:
> https://dl.dropboxusercontent.com/u/560426/contrast_bug_16bit.png
>
> But the pixel values are the same in left (sequence) and the right
> (single) image, so there is not a problem. However, the images are 16bit
> images and I have to convert them to 8 bit. I do that with  "Image ->
> Type -> 8bit". But now the pixel values are DIFFERENT:
> https://dl.dropboxusercontent.com/u/560426/contrast_bug_8bit.png
>
> The marked pixel has in the left image a value of 22 and in the right
> image a value 8. It think this must be a bug because this is not the
> expected behaviour, isn't it?
>
> Cheers,
> Thorsten

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html