John Cupitt

That's very nice Kleis. I'm sorry your patch hasn't been merged, I've been so distracted :( Do you think we should merge for 8.9?

Ah, now I check on a bigger screen I can see the extra moire you were talking about more clearly. It seems you have the vector path enabled, so it's...

I wonder if it could be related to the extra vector path in reducev? Anyway, let's defer this to 8.10 and perhaps try to improve `shrink` round-to-nearest as well.

I got confused, so here are those timings in a table to make comparisons a little easier. I converted times to milliseconds, and called the three operations box (for a...

Oh, found it, 8x reductions will hit the vector path, 9.4x is a little too large and it falls back to C.

Yes, libvips copies edge pixels outwards, so you would expect the red to be a little stronger. Thank you for the notes on pillow precision. Orc only has 16 x...

Oh that's very nice, you've done pillow-7 SIMD already? To defend libvips a little, its implementations of box filter and kernel filter downsample have extra complexity (and hence lower speed)...

Huh weird, I see it on two ubuntu 19.10 systems. You're right, it's not pillow7. I tried running in gdb and butting a break on printf, but the string didn't...

Hi @tdmorello, yes, that's how OME-TIFFs are stored, they are always single plane images. You'll need to chop and recombine if you want to make a regular RGB. Could you...

I'd do it like this: ```python #!/usr/bin/python3 import sys import pyvips # ome images load as a tall, thin strip, with page-height indicating the breaks image = pyvips.Image.new_from_file(sys.argv[1], n=-1) page_height...