I'm here to make the case again to Vitaliy to figure out how to control the GH2's cropping and downscaling features, and allow us to change them through ptool. (If it's not obvious, this post concerns video, not stills.)

There's a universe of old C-mount lenses out there, in 1/3, 1/2, 2/3, 1 inch, and 16mm, each of which makes a different sized image circle. To use each of these lenses to their fullest, you want your camera's sensor to be slightly smaller than that image circle. With the GH2 we have a camera with the capability to have any sensor size you want, up to 4/3. The "universal-format camera" is in our hands, we just need some awesome hardware hacker to unlock it!

The "proof" that the GH2 is capable of this is found in the facts that 1) the GH2 is currently able (has the processing power) to take the entire 4/3-inch, 4976x2800, 16:9 portion of the sensor, and downscale that to 1920x1080 or 1280x720 (did I see that some of you had got the GH2 to use other sizes?); and 2) the GH2 is also capable of using only part of the frame, with no downsampling (as demonstrated in ETC mode). So it's not much of a stretch to believe that the GH2 is capable of taking any 16:9 subset of the sensor's pixels AND downscaling that to any 16:9 pixel size! (Indeed, by de-centering the crop region, you could correct for the inevitable misalignment between lens and sensor.)

So the challenge to you, Vitaliy, is to figure out how to choose that subset of pixels and add that item to ptool. Then, for example, I can take the 4160x2340 subset of the GH2 sensor's 4976x2800 pixels* that my cheap, fast, good 1-inch format C-mount lens can cover, and get great video. Right now, the only way to use this lens is to crop the video in post, which kills resolution (up-sampling, yecch) and wastes bandwidth encoding parts of the frame that are going to be cropped away.

(*yes I know it's more complicated than that, since it's a Bayer sensor, and the full sensor is more like 4976x3456, but you know what I mean.)

Further, as balazer pointed out in the posting "GH2 ETC-mode noise penalty" ( www.personal-view.com/talks/discussion/comment/27293 ) the GH2's downsampling achieves 95% of theoretical efficiency at noise reduction by binning. So in my example above, where I'd be using about 70% of the area of the whole 16:9 portion of the sensor, my lens would have only a half-stop noise penalty compared to some way overpriced lens that could light the whole 4/3 frame--if such a lens even exists in this focal length. And spatial resolution (as in lines per picture height) would still be excellent, since the downscaling to 1920x1080 would still be a reduction to around 45% of the raw image. So the result would look far better than the existing ETC mode.

Suddenly, all that old C-mount glass, as ridiculously overpriced as some of it has become, will finally be worth using. No more dark corners. No more frustration with not knowing what image format a given C-mount lens that you find on eBay was designed for. If it's a C-mount lens, your GH2 would be able to use it to full advantage. As a side effect, this would encourage more reasonable pricing of Noktons, Noktors, and the rare cine lens that happens to fully cover 4/3.

Here's hoping that with Panasonic's release of the v1.1 firmware, this level of control will finally be possible! Perhaps this feature can be implemented as a modification to ETC mode, which would provide a handy way of turning this feature on and off through the camera's menu in the field.

Thanks again, Vitaliy, for all you've done.