You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
So I have a jpeg image that is 1932x2576 in resolution.
Not sure how it got those dimensions, perhaps by being cropped. I used your library to rotate/normalize the orientation, as it should be landscape. Exif says orientation: 6.
The incorrect result was that ~10 pixels of the left of the image were copied to the right edge of the photo and it wasn't rotated.
I restored from backup, opened it in gthumb instead, and chose "rotate physically." This worked correctly, so believe it is something to do with this library or module.
This library worked fine on my other images, and so far this is the only image that showed the problem.
If I had to guess it has something to do with the fact that the width is not evenly divisible by 8/16.
I believe this is a fundamental limitation of most lossless jpeg operations.
Quoting jpegtran man page:
other transformations operate rather oddly if the image
dimensions are not a multiple of the iMCU size (usually 8 or 16 pix‐
els), because they can only transform complete blocks of DCT coeffi‐
cient data in the desired way.
Feature request: it would be nice if this library warned us about this, and offered an option to trim like the jpegtran command line utility.
So I have a jpeg image that is 1932x2576 in resolution.
Not sure how it got those dimensions, perhaps by being cropped. I used your library to rotate/normalize the orientation, as it should be landscape. Exif says orientation: 6.
The incorrect result was that ~10 pixels of the left of the image were copied to the right edge of the photo and it wasn't rotated.
I restored from backup, opened it in gthumb instead, and chose "rotate physically." This worked correctly, so believe it is something to do with this library or module.
This library worked fine on my other images, and so far this is the only image that showed the problem.
If I had to guess it has something to do with the fact that the width is not evenly divisible by 8/16.
Thanks for your help.
The text was updated successfully, but these errors were encountered: