Chris Havlin

> Arbitrarily choosing to have the linthresh set to the maximum / 1000 seems problematic, particularly for cosmological data. yes! very much agreed. In playing around, just changing that to...

pinging @neutrinoceros and @chummels as the most interested parties, but all opinions are welcome!

> In the slightly longer term, I'll try to see if we can actually use this new keyword argument ... but I'm not sure how/if we can use it directly...

Note that I experimented with the new imshow kwarg (`interpolation_stage="rgba"`) but still got some rounding error crashes related to the normalization objects (`SymLogNorm`, `LogNorm`) and so gave up on that...

So the issue with using `interpolation_stage` is that it is only an `imshow` kwarg. When using the `Norm` objects directly, it's pretty easy to get it to error with the...

Ok, so I rewrote all of `_guess_linthresh`. It felt clearer to avoid the whole abs(min,max) vs min(abs) confusion and treat the positive and negative ranges sequentially. It's maybe a little...

interesting... that failed test passed locally. Will have to dig in further...

Thanks for taking care of the rebase, @neutrinoceros ! I'll look through your comments ASAP.

Latest batch updates the docs, bumps the answers and incorporates the remaining suggestions. I'll go through and close out obviously outdated threads (done)...

This example requires #1990 (hopefully soon to be merged!) and uses one of the test files added there. This data uses NaNs to mask (correctly) missing data so the following:...