I guess there is no way to hide from the comparison, though what I hope is that people who are more familiar with Solr than myself will play around with ElasticSearch and an unbiased review on it.
What I can tell though is that the main different is the distributed model. ElasticSearch is built from the ground up to be distributed, which Solr solution is lacking somewhat :). Also, there is multi tenancy (multi indices, multi types), (near) real time search, atomic operations (no need to commit), write behind (so indices can be stored in memory and still be stored in a long term persistence, very cloud friendly, think storing the index in Amazon S3).
Of course, another feature is the comprehensive support ElasticSearch has for JSON based *domain* driven date which is schemaless at its nature. I think Solr does not really have what ElasticSearch has.
Regarding actual search features, Solr is probably ahead in the game since I focused mainly on getting the core features into ElasticSearch. In any case, the search features that I will add will be driven from what I have been doing for some time in Compass, not sure how exactly they relate to Solr.
I will probably have to write something about it somewhere (blog or something). But for now, I hope it covers it.