Closed glostis closed 5 years ago
This one is easy! I never implemented 16 directions! But I see the source of the confusion in the command line help announcing it. Thanks!
Ok, makes sense.
Out of interest, do you know why the code still runs sometimes successfully when given a parameter that is not handled?
That's a mistery, in my computer it failed all the time. The loop over the directions traverses a vector containing parameters that has only 8 entries, one per direction (mgm_core.cc line 885). Line 893 shows that the intention of having 16 directions was there... ;)
just for fun, can you show one of the results you obtained?
On Wed, Jun 5, 2019 at 9:54 AM Guillaume Lostis notifications@github.com wrote:
Ok, makes sense.
Out of interest, do you know why the code still runs sometimes successfully when given a parameter that is not handled?
— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/gfacciol/mgm/issues/6?email_source=notifications&email_token=ABGFI2TILVBNGDSXI3TDN3LPY5WLLA5CNFSM4HS7F4MKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODW645BI#issuecomment-498978437, or mute the thread https://github.com/notifications/unsubscribe-auth/ABGFI2VYCAKKDSNPYLU4SYDPY5WLLANCNFSM4HS7F4MA .
I am encountering a sporadic seg fault while running
mgm_multi
on a pair of images: sometimes the command succeeds, and sometimes it stops with a seg fault.The command is the following:
I have encountered this error while playing with different values for the flag
-O
. It seems that the bug only occurs with-O 16
.On a side note, when the command succeeds with
-O 16
, the output image has a lot of ugly horizontal stripes, which it does not have with-O 8
. I can open a separate issue on the subject if you think it is necessary.I am including the input images
u.tif
andv.tif
in a zip.mgm_bug_report.zip