[IPOL discuss] 16 bits IPOL demo available

Miguel Colom Miguel.Colom at cmla.ens-cachan.fr
Thu Mar 8 14:21:21 CET 2012

I've written some modifications to the IPOL code in order to support 16
bit images.

Unfortunately, the PIL library IPOL is using right now doesn't allow to
write 16 bit PNG and therefore it's not possible to select subimages in
this mode. If the input image is not 16 bits, the demo works as usual.

The new code is running right now in my noise estimation demo:

The JPEG images are 8 bits, but those labelled with "WB+TC" are 16 bits.
You can test both types and also upload new images.

The modification affects to these files:
- lib/image.py    : http://dev.ipol.im/~colom/base_app.py
- lib/base_app.py : http://dev.ipol.im/~colom/base_app.py

And the app.py of the demos need some minor modifications too. If they
aren't modified, they'll work as usual.

Of course, the modifications I'm talking about have only been done in my
fuchsia space.

Anyway, this is only a workaround to work with 16 bit images. However, in
my opinion we shouldn't use at all the PIL library and use other
alternatives (the png module is able to handle 16 bits PNG files


More information about the discuss mailing list