John Cupitt
I guess the `squash` option could squash lab16 down to lab8. You could also load lab8 + alpha as float lab. That might be simpler.
I'm not sure, this seems like quite a large behaviour change :-( It's likely to break a lot of downstream code. How about adding a new operation with the new...
Yes, it would be odd :( Really we'd want to replace all the `vips_thumbnail*()` functions with your proposed new API, but we'd need to do it under a new name...
Yes, `conva` isn't really production quality, sadly :( I'll have a look.
This is a bug that needs fixing, so this one should stay open, I think.
Hi @BadgerCat1066, sorry, I'm not sure what's considered best practice. You could make a little factory class to wrap the vips constructors you need, would that work? All your static...
Expose the GIF / PNG / etc. quantizer as an operator.
Let's move any discussion to the PR.
The official way to handle this is to default the unspecified dimension to infinity, so I think I'd write it as something like: ```ruby def infer_dimensions((width, height), (current_width, current_height)) raise...
Could you try an image with an odd number of columns?