Francis' BBC Micro Image Convertor has proved inspirational in these parts - I decided to try doing something similar (but nowhere near as good - or as fast!) as a plug-in for The GIMP:

http://kecskebak.blogspot.com/2010/11/bbc-on-bbc.html

Nice work on that, although I'm surprised that you didn't include the Atkinson dithering method (one of my favourites). You can get info on it here: http://www.tinrocket.com/projects/progr ... ics/00158/

Kind regards,

Francis

The absence of Atkinson is entirely down to laziness on my part - it would be very easy to add, but at the moment I get my divisor by summing the matrix values instead of storing it separately, which I'd need to do for Atkinson.

You're right about Atkinson being good - I've used it a lot in the BBC Micro Image Converter.

For Image2BBC I just store the error diffusion matrices as fractions so no need to sum the numbers up and work out the divisor.

The only other thing I will say is that because the Atkinson dithering method doesn't diffuse the full error you may hit a problem with your dithering strength algorithm depending on how you have implemented it. If my memory serves I had to do a special exception for Atkinson dithering.

Kind regards,

Francis

Many thanks for your advice Francis. Now you've pointed it out to me it's obvious that I should store fractions instead of integers for matrix values