Have you considered a scenario where seeding relays with illegal images could quickly undermine the entire network? 

I actually think centralized services are great at handling media because they have more tools at their disposal to moderate for illegal content. I thought that having media live on relays could potentially  put them in legal hot water so to speak. But with text only, there is a much lesser risk of being taken offline. That and it keeps relays leaner and easier to spin up. Or were you thinking only some relays handling media?