Re: Problem loading 16bit images using imread()

[Carnë Draug]

On 8 Aug 2013 08:29:57 -0700,  bmgamm <address@hidden> wrote:
> Carnë
>
> I seemed to figure out a solution/work-around. I originally photographed the
> images in RAW and converted them to 16-bit tiffs in Nikon ViewNX 2. When I
> tried to read those images into Octave as 16-bit, they were down-sampled to
> uint8. However, I re-saved the images in Adobe CameraRaw as 16-bit TIFFs.
> Those images did load into Octave as uint16. I'm not sure why one file
> loaded as uint16 and the other loaded as uint8. Nevertheless, I am now able
> to work with the images. I attached a Dropbox link below containing two
> seemingly identical images. 'Baseline_No' loads as 8bit. 'Baseline_Yes'
> loads as 16bit. I can email you directly if the link doesn't work.
>
> https://www.dropbox.com/sh/iw98a8tadjpg5uc/kiGpEjMo0f
>
> Brian

I can reproduce this issue with Octave 3.6.4. However, it has already
been fixed in the development series.

I am unsure which of the many things that have changed fixed it but my
guess is that it was the change from modulusDepth() to depth(). With
the former, GraphicsMagick would report the minimum number of bits to
store your image. Indeed, after loading your image it reports having
exactly 256 different colors. Why this changed after you saved the
image with CameraRaw I don't know.

Bottom line, this was a bug in Octave and is now fixed.

Carnë