High Dynamic Range Video on two 5D's

One lens via prisma to to alligned 5Ds? Jusr my guess...
Very interesting!

Any idea how to set this rig up in real life?

I'd love to try it using two T2i bodies. Basically, how would you set up the beam splitter behind the lens to get proper focus, and then how do you get the image signal from the splitter to the bodies?
 
Why couldn't this be done with 1 camera, by putting the footage on 2 different tracks in the NLE and simulating the under and over exposure with plugins and blending the 2?

Possibly the point of view needs to be very slightly different or the timing of the takes needs to be very slightly different?
So pan the second one by a very small amount or time slide it by a few samples?
 
Why couldn't this be done with 1 camera, by putting the footage on 2 different tracks in the NLE and simulating the under and over exposure with plugins and blending the 2?

Because the 5D (in this example) does not have the dynamic range to extract the range of information this method can pull from the shadows and the highlights.


Possibly the point of view needs to be very slightly different or the timing of the takes needs to be very slightly different?

The point of view needs to be the same (ideally identical). The timing needs to be the same (ideally identical).


So pan the second one by a very small amount or time slide it by a few samples?

Video footage cannot be (temporally) slid by a few samples, the finest resolution you can move the footage by is one frame.
 
Understood on the range of information.

Vegas 7 will allow sliding video by samples if you choose to disable "quantize to frames"
in the Options menu, but it probably wouldn't be noticeable until you got past a few frames or difference.
 
Vegas 7 will allow sliding video by samples if you choose to disable "quantize to frames"
in the Options menu . . . .

Trust me, if you have two tracks of video - you cannot offset them by anything less than one whole frame.

If you were to have two video tracks and push the second forward by 4,800 samples (let's assume 48kHz as our sample rate - so 1/10th of second) what do you think you might see when these two video images are mixed together ?


but it probably wouldn't be noticeable until you got past a few frames or difference.

Not sure what this means ?
 
You could easily stop, finding a place where frame 100 was at the cursor on the top track and frame 99 was at the cursor on the bottom track, but only for part of a frame. It would be offset by how far it was slid.

I agree ... it's not an issue.
 
You could easily stop, finding a place where frame 100 was at the cursor on the top track and frame 99 was at the cursor on the bottom track, but only for part of a frame. It would be offset by how far it was slid.

. . . . and what might this hybrid frame look like ?
 
Hmm can't say I'ma big fan of these HDR images. I almost always prefer the under-exposed stuff. If the HDR is used in a more subtle way to increase the highlight detail better (and therefore to look more film-like) then that'd be awesome. But most of the time I just see un-natural shadows and midtones which make skin look awfully weird...
 
Yea... I know ...
I guess it's better to say, video can be slid by sample level but it doesn't matter.

I wouldn't even really say video can be moved/slid at the sample level - unless we are both getting confused about what we are talking about here !! It wouldn't be the first time I have got confused - not even today !! :)

If we have two tracks of video . . . track A and track B.

. . . and we slide track B backwards, that is: move it earlier in the timeline by - for example - 10% (1/250th of a second @ 25fps - or 4,800 samples @ 48kHz) . . .

. . . then because video is quantised (in this example to 1/25th second) the only option for the application is to move track B backwards by 1 whole frame and then reduce it's opacity to 10% to simulate this subframe movement.

I hope that makes sense !!
 
And all I was saying is that Vegas allows sliding video by samples, if you disable quantize to frames.
 
Last edited:
Back
Top