Re: Change in log calculation behavior

[Gregory Reneff]

Take 2 because take one was rejected for zip file attachment.

On 4/27/16, Greg Reneff <[hidden email]> wrote:
> I'm getting different results when I run the attached script against
> the attached image between two different versions of ImageJ
>
> version 1.50b gives me an image as I might expect
> version 1.50i gives me an image with only two values in it.  Almost
> like it's been thresholded.
> --Greg
>

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

[Michael Schmid]

Hi Greg,

yes, I can confirm this, and I looked a bit into the reason for the issue:

Before ImageJ 1.50d11 ImageJ has used calibrated values for the
Process>Math>Macro calculations (but uncalibrated values for 8-bit
images and all other Process>Math operations). It seems that this
inconsistency was straightened out in 1.50d11.

You have a signed 16-bit image; for ImageJ this is an unsigned image
with a calibration that converts unsigned to signed values (that's why
you see the unsigned value in parentheses in the status field when you
have the cursor above the image). The other Process>Math operations take
care of the special case 'signed 16-bit image' [technically, inside the
ShortProcessor class], so it would make sense to separately handle
unsigned images also for Process>Math>Macro.


Michael
________________________________________________________________
On 2016-04-28 02:21, Greg Reneff wrote:

> Take 2 because take one was rejected for zip file attachment.
>
> On 4/27/16, Greg Reneff <[hidden email]> wrote:
>> I'm getting different results when I run the attached script against
>> the attached image between two different versions of ImageJ
>>
>> version 1.50b gives me an image as I might expect
>> version 1.50i gives me an image with only two values in it.  Almost
>> like it's been thresholded.
>> --Greg
>>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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

[Rasband, Wayne (NIH/NIMH) [E]]

> On Apr 27, 2016, at 8:21 PM, Greg Reneff <[hidden email]> wrote:
>
> Take 2 because take one was rejected for zip file attachment.
>
> On 4/27/16, Greg Reneff <[hidden email]> wrote:
>> I'm getting different results when I run the attached script against
>> the attached image between two different versions of ImageJ
>>
>> version 1.50b gives me an image as I might expect
>> version 1.50i gives me an image with only two values in it.  Almost
>> like it's been thresholded.

The latest ImageJ daily build (1.51a16) fixes a 1.50d regression that caused the Process>Math>Macro command to not work as expected with signed 16-bit images.

-wayne

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

[Gregory Reneff]

Thanks for the quick response.  Looks like it is working with my contrived demo, and with another image that originally flagged the issue.
--Greg

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