git HEAD bug: register can't display mask file after it has been edited by Display

[gdevenyi]

Before:

> register file.mnc mask.mnc

Can see mask and file, standard register operation

Steps to reproduce

> Display file.mnc -label mask.mnc
*Edit mask in some way*
Save label as mask2.mnc

> register mask2.mnc file.mnc

Register shows a black space where the mask should be, no adjustment of intensity can make it show.

[gdevenyi]

Also present in earlier versions.

[rdvincent]

@gdevenyi I have investigated this a little - with the mask file I am using, here's what happens. The original mask has the real range 0-1, but when Display saves a label image it sets the real range to the total number of labels possible in that image, so the range is changed to 0-255. What I see is that if I then load the mask file in register, it is possible to see the mask but only if I turn the uppermost intensity down to something close to one.

Arguably, Display should save labels using only the true range, so that a 0-1 image would be saved as a 0-1 image.

[zijdenbos]

Ah - this has been annoying for a few decades :-) People create a binary mask in Display, load it in register, and think something has eaten their labels. I'd definitely say that Display should save with a correct real range (as all other MINC tools do, no)?

…

-- A

On Tue, Feb 21, 2017 at 8:53 PM, Robert D Vincent ***@***.***> wrote: @gdevenyi <https://github.com/gdevenyi> I have investigated this a little - with the mask file I am using, here's what happens. The original mask has the real range 0-1, but when Display saves a label image it sets the real range to the total number of labels possible in that image, so the range is changed to 0-255. What I see is that if I then load the mask file in register, it is possible to see the mask but only if I turn the uppermost intensity down to something close to one. Arguably, Display should save labels using only the true range, so that a 0-1 image would be saved as a 0-1 image. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#20 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA_r-JuwK_vfzTlo7Ah7ECP6Cfpa3sKmks5re5UBgaJpZM4MCF9h> .

[rdvincent]

Yes, I agree that this should be fixed. I'm looking into it.

[rdvincent]

@zijdenbos @gdevenyi See develop branch of Display, which should now resolve this issue.