[MINC-users] register and [0-1] float volumes

[Vladimir S. FONOV]

Hello,


ok, I included all the latest versions of packages in the develop branch 
of minc-toolkit-v2, at least register doesn't crash right now.



On 15-11-23 04:34 PM, Robert D. Vincent wrote:
> Ok, I had hastily made a fix for Register to match what I had done for
> Display, but hadn't properly handled the merged volume. It should be fixed
> by the latest commit to develop.
>
> There will remain some situations in which the colour mapping isn't ideal
> for float volumes, but the latest commit should at least do something
> reasonable (and not crash).
>
>      -bert
>
> On Mon, Nov 23, 2015 at 3:17 PM, Vladimir S. FONOV <vladimir.fonov at gmail.com
>> wrote:
>
>> Hello,
>>
>>
>> it looks like new patches are not quote ready for general use. I updated
>> all the subprojects in minc-toolkit-v2 to the latest develop heads and
>> rebuilt register. And I am getting following results:
>>
>>
>> 1. running register on a float volume with range 0-8.57 - i have old
>> behavior (i.e all the values between 0 and 1.0 are mapped to black)
>>
>> 2. running minccalc -express 'A[0]/10' test1.mnc test2.mnc (test2.mnc now
>> have range 0 - 0.857) , followed by register test2.mnc - works properly .
>>
>> 3. running register test2.mnc test2.mnc results in segfault, register
>> test1.mnc test1.mnc still works.
>>
>> 4. running minccalc -express 'A[0]*10' test2.mnc test3.mnc,
>> register test3.mnc - works, but with old behavior
>> register test3.mnc test3.mnc segfaults.
>>
>> You can download files to try it on your system:
>>
>> https://www.dropbox.com/sh/7vtg2tszawnlqyi/AAD8mN60UttvAdLxLYKtDcMha?dl=0
>>
>>
>>
>>
>>
>>
>>
>> On 15-11-23 02:18 PM, Robert D. Vincent wrote:
>>
>>> Hi Simon,
>>>
>>> I plan to merge many of my changes into the master branch for bicpl,
>>> bicgl,
>>> Display, and Register soon. However, the minc-toolkit and minc-toolkit-v2
>>> packages pick specific commits from the submodules, so they don't
>>> automatically track what is in master for a given project. In fact,
>>> minc-toolkit-v2 uses commits from late June or thereabouts, so it's
>>> neither
>>> develop nor master.
>>>
>>>       -bert
>>>
>>> On Mon, Nov 23, 2015 at 7:38 AM, Simon Eskildsen <eskild at gmail.com>
>>> wrote:
>>>
>>> Hi Bert,
>>>>
>>>> Would it be possible to merge the fix into the master branch? At least it
>>>> would be good if minc-toolkit-v2 included a version of Register that has
>>>> the fix. I'm using the master branch of minc-toolkit-v2.
>>>>
>>>> Thanks,
>>>> Simon
>>>>
>>>> On Mon, Oct 12, 2015 at 5:43 PM, Robert D. Vincent <
>>>> robert.d.vincent at mcgill.ca> wrote:
>>>>
>>>> Hi Alex,
>>>>>
>>>>> There is a recent commit that should fix this issue. In a nutshell, I
>>>>>
>>>> made
>>>>
>>>>> the code recognize volumes that cannot use an integral colour table, and
>>>>> use a full colour mapping instead.
>>>>> On Oct 12, 2015 11:37 AM, "Alex Zijdenbos" <zijdenbos at gmail.com> wrote:
>>>>>
>>>>> Hi all,
>>>>>>
>>>>>> In the flurry of recent improvements to Display/register (thank you
>>>>>>
>>>>> Bert
>>>>
>>>>> et
>>>>>
>>>>>> al.!), I was wondering if the problem of register not being able to
>>>>>>
>>>>> display
>>>>>
>>>>>> float volumes with a range [0,1] has been addressed? AFAIC that is the
>>>>>> single most annoying issue with register (I am keeping a register
>>>>>>
>>>>> 1.3.6.
>>>>
>>>>> build from 2009 around that still works for such volumes; or I use a
>>>>>> wrapper that converts to byte/short on the fly).
>>>>>>


-- 
Best regards,

  Vladimir S. FONOV ~ vladimir.fonov <at> gmail.com