Alexey Kuleshevich

@Twinside is right, it is possible to use JuicyPixels reading facilities directly and then use an explicit cast to HIP `Image` fromat with helper functions from the [conversion section](https://hackage.haskell.org/package/hip-1.5.6.0/docs/Graphics-Image-IO-Formats.html#g:11). FYI...

@Twinside I agree, that bundling this together with a bigger breaking change is a good idea.

@SwiftsNamesake I think I have a solution you are suggesting: [Color](https://github.com/lehins/Color) library > What about using newtypes to encode colour spaces? In fact, I also implemented an interface that sits...

You are absolutely right about this. We need a new concept of min max values on per channel basis, which has to be at the `ColorModel` level in order to...

@Magalame I definitely do have a strong desire to add SIMD support to massiv. I have tried a few things before and have been pondering on some ideas on how...

It is really good news that ghc will finally be able generate SIMD natively. I had a pretty good day in that direction as well. Was experimenting with implementing matrix...

I started a separate repo for now for the [`massiv-simd`](https://github.com/lehins/massiv-simd) package that will eventually find its home in this repository. For now it only contains a handful of functions that...

@Magalame It is a very interesting read, highly recommend it too. Unfortunately `vector` still doesn't use SIMD, despite what the paper says. It does give a very good insight into...

@klapaucius That's where that branch went, I was looking all over for it, I even asked Roman about it. :D (Thank you for posting a link to it!) Despite that...

@klapaucius your comments are always so concise! :smile: Spoke too soon about the SIMD support in ghc. But's ok, I am sure Ben Gamari will figure it out.