John Cupitt
I think this is a meson issue when linking the introspection code. @eli-schwartz , sorry to ping you, have you come across this? Should I report this upstream?
... I'll close, I don't think this is a libvips issue.
I experimented with git master libheif and it seems greyscale images are not well supported. I did: ``` $ vips colourspace mono.jpg rgb.jpg srgb $ vipsheader mono.jpg rgb.jpg mono.jpg: 3232x2071...
Hi @remicollet, thanks for testing the rc. I don't think we want to maintain two build systems, so I'm keen to remove autotools as soon as possible. In my mind,...
Hi @kaspergrubbe, I agree, this doesn't look right. Let's try to fix this in the 8.14 cycle.
OK, fixed in master. Thanks for pointing this out!
Hi @evanmcallaway, libvips doesn't support 8-bit LAB directly. It has three LAB modes: float, short and labq. There's some [discussion in the docs](https://www.libvips.org/API/current/libvips-colour.html#libvips-colour.description). You can rework your code like this:...
You can make the libvips tiff writer write an 8-bit LAB image like this: ```ruby result.colourspace(:labq).write_to_file(file_name) ``` You probably know. You'll find it won't hit the numbers exactly because of...
Yes, ASSOCALPHA means pre-multipled alpha, and you can only have one of them. libvips is not premultiplied, so you must use UNASSALPHA eveywhere.
I think UNSPECIFIED is for unrelated information, so I would use UNASSALPHA everywhere (this is what libvips does). ASSOCALPHA channels mean you must unpremultiply the pixel data using that channel...