Difference between ReadPixels and ImportPixels (and reading pixel data in general)

[Athari]

What is the difference between ReadPixels and ImportPixels? They seem to be doing the same thing and accept roughly the same parameters.

Context

What I'm trying to do is to use FFMediaToolkit library (a wrapper for FFmpeg libraries) to read video frames into MagickImages (I need more control over what and how is being read from a video file). FFMediaToolkit can either produce a Span<byte> or write into a provided Span<byte> or an IntPtr address (the latter being more optimal as unnecessary copying is avoided). It supports various pixel formats (bgra32, rgb24, gray16 etc.; 8 non-yuv formats in total). I need a generic solution which handles all these pixel types (and preferably all quantum types).

Questions

1. What would be the most straightforward/fast/correct way to load pixel data into a MagickImage? Is new MagickImage() and image.ReadPixels(span, settings) enough?
2. Or should I use ImportPixels(span, settings)?
3. Do I need to set any properties for grayscale and transparent images, or do ReadPixels/ImportPixels methods handle this when reading/importing?

P.S. #1237 mentions ImportPixels, but after reading it, I'm more confused than before.

[dlemstra]

You would use ImportPixels when you want to import pixels into an area of the image. And yes that could also be the whole image. But in your situation I would advise you to use ReadPixels after you have created an empty MagickImage. And in the settings you will need to specify the mapping and the storage type of the pixels.

[Athari]

What about (3) — do I need to perform any extra operations for grayscale or transparent images, or will ReadPixels handle everything for me? (Grayscale is imported with "R" as the mapping, is that correct?)

[dlemstra]

I did not understand that part of the question. But yes you only need to use "R" as the mapping for grayscale.