Recently, I was surprised to be told that we could not have colour figures in our paper (http://www.russet.org.uk/blog/2170), even though it was online only. Our assumption is that this is an enormous legacy issue; the publisher in question, OUP, still produces a tree-based version of the journal, Bioinformatics. The distinction between colour and monochrome is important here.
Of course, it is easy to criticise others for being trapped in a legacy situation. The reality is, though, it can to happen anyone; it is not possible to always take a step back, to reassess standard procedure, to think whether it still makes sense. The paper based publication process still affects all of our ways of thinking and this includes myself.
The paper in question contains graphs showing a time course, in this case showing an analysis of all the versions of Uniprot/Swissprot from the first archived version. For our print version (now available at arxiv (1208.2175)), we used a four panel graph (yes, in colour), showing the first version, the last, and two in the middle.
However, while writing the talk for ECCB12 we realised that what we really needed was movies; animated images showing the change over time. It actually takes us much less screen space that a four panel display and displays the results very clearly. He has also published the same images online (http://homepages.cs.ncl.ac.uk/m.j.bell1/blog/?p=610). In this process, he has managed to achieve with a simple web page what conventional scientific publishing cannot. Both colour and motion. He has also shown us how bound we are by the paper based legacy of scientific publishing; this is how we should have presented the data all along, and it should not have required a talk for us to realise this.