Cool - I'm guessing it's done in-camera? That's a cool feature.

It is very hard to tell that this is in reality.
My understanding is that this is similar to my idea recently published on blog.
They get 720p60, and change ISO for each frame, or just processing.
It could be not easy to edit this stuff.

It's too bad it isn't supported on the 5D from what I can tell.

No they mention the framerate hack, has to be 48 frames.

Woho this is some amazing stuff here!

Way cool. I think the only issue would be not to record fast motion with or go handheld with a high shutter speed or you will see wired motion artifacts. You can see it a little with lady at the terminal in the beging if you look close at the pen moving across the screen. But whatever......

HDR video on a DSLR is huge news! I could totally use that.

This seems like a very technologically doable feature. No special processor chips etc. Very helpful because it both increases dynamic range. A problem with GH-2 and all HDSLRS, and very well may be a workaround the 8 bit video. Combining two video streams, even with just a 1 stop exposure difference, could be rendered into a higher bit depth working file for grading. Could eliminate banding.

They once shouted of the death of HDSLRs for video, as AF100, Scarlet, various Sony camcorders came out. But with these new feature discoveries, Our little cameras have extended usefulness.

I bet an even closer difference in exposure could be useful in noise canceling in the image. Even if it is not possible on a GH-2. I hope our favorite Panasonic Engineers start to feel competitive with Canon's feature in time for the GH-3. Very cool!@

this is hdrx for a bargain price. this is huge, if its editable! No DR Problems anymore? DSLRs final problem solved, using vitaliys idea! I reeeally hope this is doable on GH2, in theory. I won't use a 550D. :)

Problem is: cant work on fast motion. there are never 2 of the same frame.

Don't mess with the users. Where there's a will, there's a way. Great stuff!

wow, sounds great!
(not to mention other features of the ML...)

i downloaded the clip.
and did play with it in sony vegas.
but sorry, i cant find any advantage over what i already can do in post editing.

i hope i do something very wrong....

Exactly how the blend is implemented, I think, should make a massive difference when it comes to time differences and end result. Now, it would make sense to use some of the color information, but not all - even with motion estimation, which in itself is arduous - there will be a leap in missing movement information. So basically you are shooting in the blind for a particular scene.

I'm guessing that a "base" frame of sufficient low level information and an "extended" frame of some 50 levels of highlights would yield pretty reliable results, coupled with curves that mix the frames well. It could be done vice versa as well, depending on which end of the spectrum one prefers for moving objects.

@oedipax
It's ok, I didn't want hdr video anyway...
*sniff sniff*

If this guys make different processing for same RAW frame, rather than different frames, it'll be much better.