* Text should not be centered.
* The Unique Selling Proposition (USP) needs to be more prominent. The quote box draws in the eye. There are no faces or other images to attract the eye to where you want it to be.
* The About/Goodwill/etc. tabs break basic usability guidelines. The user is unlikely to realize that the blue means 'selected.' This non-intuitive relative to scrolling.
* USP is too heavy on features and not enough on benefits. Say what's in it for the user and less on what the thing is.
* The table with the users is OK but you would get more persuasive lift from attributed testimonials like the one you have above it from the HN guy.
* The supporting copy needs a red marker through 90% of it.
* "We make sharing stuff suck less." -- sounds cool in an affable internet cool guy kind of way, but does not actually communicate in a way that drives the reader to action.
The old version is actually a little bit better, although the animated GIF where it is will keep the visitors' eyes bouncing to it repeatedly as it loops, which impedes comprehension. The old version also uses too many h2 subheadings and would be more appropriate for an FAQ. h2 is too big for such short snippets of text.
Also, I like the GIF -> HTML5 conversion bit in the original, because it shows a benefit of your service. Seeing the video play bar on mouseover was also a pleasant surprise. I would explain why your prospective user might want to do that (make your pages load faster to make your readers happier and more loyal) without using jargon. 'Kilobyte' is jargon, as is KB.
Apart from that the navigation on the website isn't clear. The About page also doesn't drive me to sign up -- your call to action at the bottom of the first tab is to check out your subreddit, IRC channel, and twitter account. The general convention is to ask for the signup and then to provide the option to learn more if they're not sold yet.
Some more general suggestions:
1. Excise all the technical terms (yes, even references to file types -- say what the file is) from the portions not targeted to developers.
2. Weight the benefits heavier than the features (tell me what it will do for me and then explain how it will do it rather than telling me how it'll do it and expecting me to figure out why I, the selfish and lazy user, should care).
3. I would work on refining the 'Mediacrush is a fast, open-source media hosting service' tagline.
Hope this was helpful -- good luck!