Images with bit depth > 8 (depth sensor or medical images)

[dibenedetto]

grayscale images with bit depth > 8 (think about depth sensor or medical images) are automatically converted to 8 bit instead of keeping the most adequate representation:

if color_mode == 'grayscale':
if img.mode != 'L':
img = img.convert('L')`

PIL images with mode == 'I' (capital i) are actually 32bit integer images, and should not be shrinked down to 8 bit, thus keeping all the information.

[rragundez]

I believe you have a point here. Can you provide me with one of these images so I can perform some tests? thanks!

[rragundez]

wow! how has this not been addressed before :s apologies

I'll make some more tests and make the PR promptly. Thanks @dibenedetto

[dibenedetto]

thank you for your time spent! m.

…

On Wed, Sep 18, 2019 at 4:00 PM Rodrigo Agundez ***@***.***> wrote: wow! how has this not been addressed before :s apologies [image: Selection_029] <https://user-images.githubusercontent.com/16643700/65155290-54a51800-da2d-11e9-8b69-8d09568d8b0c.png> I'll make some more tests and make the PR promptly. Thanks @dibenedetto <https://github.com/dibenedetto> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#239?email_source=notifications&email_token=AAFNHVICYOCHKHZ6MXHSEGDQKIYBPA5CNFSM4IU4EWZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7AFH3I#issuecomment-532698093>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAFNHVPOOWSFDOMURVYGQOTQKIYBPANCNFSM4IU4EWZA> .

-- Marco Di Benedetto, Ph.D. A.I. Researcher at NeMIS Laboratory - National Research Council (CNR) E-Mail: marco.dibenedetto@isti.cnr.it <marco@transformandlighting.com> Co-Founder at Transform and Lighting S.r.l. E-Mail: marco@transformandlighting.com Web: http://www.transformandlighting.com <http://transformandlighting.com/>

[rragundez]

@dibenedetto please have a look at #242. Any feedback from a user perspective would be appreciated. You can also try the code out from that branch.

[dibenedetto]

thanks rodrigo, i just replied on github. m.

…

On Mon, Sep 23, 2019 at 3:42 PM Rodrigo Agundez ***@***.***> wrote: @dibenedetto <https://github.com/dibenedetto> please have a look at #242 <#242>. Any feedback from a user perspective would be appreciated. You can also try the code out from that branch. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#239?email_source=notifications&email_token=AAFNHVJNGCOUPAWNRFAVPU3QLDBUZA5CNFSM4IU4EWZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7K4TDI#issuecomment-534104461>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAFNHVPHUGIRFC5GG7YGPIDQLDBUZANCNFSM4IU4EWZA> .

-- Marco Di Benedetto, Ph.D. A.I. Researcher at NeMIS Laboratory - National Research Council (CNR) E-Mail: marco.dibenedetto@isti.cnr.it <marco@transformandlighting.com> Co-Founder at Transform and Lighting S.r.l. E-Mail: marco@transformandlighting.com Web: http://www.transformandlighting.com <http://transformandlighting.com/>