Cool. Fork on github and send any pull requests. The issues right now are that it only works with ff3 and chrome. I'd have thought safari would work, but it doesn't (didn't for me). Some integrated browser detection to offer transparent reduction for images on compatible browsers would great.
Let me know how it works out for you :)
I like how it changes the dimension. But converting to PNG is massively increasing the file size?
"Original file was: 137724 bytes
Transmitted size was: 844176 bytes (due to base64)
New file is: 633131 bytes"
It might be - not sure it's just a PNG issue, or if it's a canvas flaw, or what. The system could ideally check the before/after sizes, and only send the smaller one. Good catch (I think I'd seen that once in testing too).