Hello John,
I have some questions regarding the 'tiffsave' command.
I am working with images ('.png' format), which are labels with colors containing red (255, 0, 0), blue (0, 0, 255), white (255, 255, 255).
To be able to work with those images, I first convert the large '.png' images into tiled pyramidic '.tif' files, using following vips (8.9.1) command:

vips tiffsave label.png --compression deflate --pyramid --tile --region-shrink mode --vips-progress ../labels/label.tif

Here I use the --region-shrink mode, which in my understanding should preserve the label colorings, by picking for each 2x2 pixels, the color, which occurs most often?

Afterwards, I open up the image using openslide-python:

import openslide
slide = openslide.OpenSlide('label.tif')
image = slide.read_region((0, 0), 0, (2000, 2000))
array = np.array(image)
np.unique(array) => [255, 0] # works for level 0

image = slide.read_region((0, 0), 2, (2000, 2000)) # any other level of the pyramid
array = np.array(image)
np.unique(array) => [0, ..., 255] # pixel values seem interpolated (red/blue are mixed at the edges)

Where is this interpolation happening? Is it from openslide, or is my tif conversion faulty?
Thanks a lot!