I don't know, I am still on the opposite end of this camp. Since the filter reduces the optical resolution to acomodate for this downsampling, why would it help the slightly downsampled image sensor stream which is in a different resolution that 1080p?
1080p is also recorded simultaniously to the live-view (AFAIK), we're not getting the same thing. If we were, the RAW output would too be limited to the 1080p or 720p depending on the shooting mode chosen.
That being said, the RAW video is much sharper than the 1080p, even at much lower resolutions. It is still being resized, but we're capturing this raw sensor data prior to the actual 1080p downscale. And since in this case we're using the full sensor here, the optical reduction would mean reduction in sharpness back to 1080p quality as the filter is acomodated for that specific resolution (again, NOT for stills because of REDUCTION IN OPTICAL RESOLUTION as mentioned in the page).
So... I think it is easy to pin-point why I believe in the uselessness of the filter for RAW imaging purposes.
HOWEVER, if someone actually owns the filter and would like to test it out and prove me wrong - I'd be happy to see the results. Until that moment I stand in this side of the YES/NO camp.
In any case, I will probably buy this filter for H.264 use as RAW is not practical a lot of the time (in terms of workflow, reliability and storage), but not now as I've shelled out my cash on a new piece of L glass and I'm moving houses this week
