"Hard matte" for wide aspect ratio & maybe improved compression?

Peter J. DeCrescenzo

Peter J. DeCrescenzo

Veteran
I know the GH1's video quality can be pretty good when it's handled carefully, but I'm just wondering: Given the GH1's relatively low bits-per-frame data rate, would using a hard matte (for example, a black cut-out shaped in 2.35:1 aspect ratio) in front of the lens or sensor help image quality?

Theory being that the solid black matte's "image" should be easy for the GH1 to compress in-cam leaving more bits available for the motion video in the cropped frame.

Even if the resulting optical matte has a soft or slightly blurred edge, that could be hard-cropped in post w/o additional image degradation if the video was going to be recompressed in post anyway (I think).

Has anyone experimented with this on their GH1? Or, maybe the idea that this might improve video quality is completely crackers? :smile:

Meanwhile, there's an interesting and indirectly related thread here:
http://reduser.net/forum/showthread.php?t=31783&highlight=hard+matte
 
Last edited:
I think the variable bitrate will just adapt and drop down a bit if you block of a part of the image with black. It certainly drops if you shoot a even surface.
 
Good point, Martti.

I wonder if a benefit of a hard matte might be in reducing the VBR algorithm's compression workload at the point just as the content of a scene begins to change/move a lot, such as at the beginning of a pan, etc.?

In other words, the "even surface" (in this case the hard matte) is always a constant in the frame, so maybe it would give the GH1 some more "headroom" or "slack" for compressing the pixels which _are_ changing? Hmm ... seems worth experimenting with ...
 
Last edited:
Peter J. DeCrescenzo said:
Good point, Martti.

I wonder if a benefit of a hard matte might be in reducing the VBR algorithm's compression workload at the point just as the content of a scene begins to change/move a lot, such as at the beginning of a pan, etc.?

In other words, the "even surface" (in this case the hard matte) is always a constant in the frame, so maybe it would give the GH1 some more "headroom" or "slack" for compressing the pixels which _are_ changing? Hmm ... seems worth experimenting with ...
Click to expand...
It may or may not. My experience with encoding tells me the edges are going to go to hell.
 
Does it not matter that the edge is soft as long as you can cut a "hard edge" out in post?? The I'm curious about this point as well related to extracting 720 out of 1080 using wide 16mm lenses with lots o vig:
720in1080-vi.jpg


does it or does it not help the VBR codec by giving it more 'overhead'??
 
In theory if you could mask off the sensor 2.39:1 @ the filter stack level, you could do it. However, without easy access to the camera during manufacturing, this sounds like a real pain to me. I would not attempt anything like this in front of the lens or at the rear of the lens, it would not work well.

Furthermore, I personally wouldn't feel comfortable making such a mod to a GH1. While you'd then be working with less vertical real estate on the sensor, I'm not really sure how much this buys you from the compression codec. Getting exact 2.39 placement from your matte would be trial an error the first time. To get the matte to stay in the correct place you would definitely run a risk of damaging the sensor.
 
Hi Illya: Thank you very much for replying!

So if you used something similar to what these folks offer (for astronomy photography), it would produce a fixed hard matte that doesn't vary when the lens focal length changes, but the edges of the matte would appear blurred?
http://www.sciencecenter.net/hutech/filters/ffilter.htm

Obviously the inside shape of the their filter holder would be need to be changed to ~2.39:1 instead of the full squarish shape of the sensor.

If a hard matte was done this way -- behind-the-lens, but not on the surface of the sensor itself -- the resulting blurry matte edges could be cropped in post. You'd maybe -- maybe -- gain up to 25%-30% worth of in-camera compression quality, and only loose a certain amount of image quality if using a compressed edit environment?

Of course the "best" way to do this would be via firmware: Panasonic would let you choose an aspect ratio "wider" (less tall) than 16:9, which would turn off or ignore photosites on the sensor, yielding a clean sharp frame edge just like when it cuts 16:9 out of the full sensor now.

Hmm ...
 
Last edited:
You must log in or register to reply here.
Back
Top