If an xarray is chunked, an indexing operation only loads the necessary chunks into memory, correct?

[GrowlingM1ke]

I have an xarray of a video file where I'm masking small parts of the video (20 by 20 pixel sizes approximately) to calculate the mean of those areas across frames. Currently it is chunked across frames with the height and width intact but I'm wondering if I rechunked the height and width dimensions and left the frames alone, would xarray be smart enough to only grab the chunks in the dimensions I've indexed and thus loading in at most 10% of the data necessary for the calculation.

Thank you in advance.