Discussion on Newsshooter 13 comments
Photokina 2016: We ask Panasonic all the questions about the GH5
Will ; a reply to Mark Ellsworth 4 days ago
"Right, but the micro43 standard is rather open about sensor size and aspect ratio. Crucially, the standard requires that the imaging area of the sensor be a specified diagonal measure. All Micro43 cameras but two have featured identically sized 4:3 ratio sensors. The GH1 and GH2 featured larger sensors that covered that diagonal in 4:3, but also 16:9 (and necessarily, 3:2 as well). Because lenses produce a circle, standardizing the diagonal ensures that every lens will still cover.
Understand, this is not the same as generating 16:9 or 3:2 imagery from a 4:3 sensor, where you just throw away some of the pixels from the top and bottom of the 4:3 imaging area. This involves using fewer pixels on the top and bottom, but also additional pixels on the sides. So your 16:9 image is both shorter AND wider than a 4:3 image taken with the same camera.
As regards video, this means that the effective sensor size is actually bigger, and the "crop factor" is thus less. The "crop factor" of the GH4's full sensor is 2, but the 4k video in the GH4 uses only some of the full width of the sensor, so the crop factor ends up being close to 2.3. In the GH1 and GH2, the larger 16:9 imaging area resulted in horizontal crops closer to 1.8 or 1.9, if I recall correctly.
So the GH5 may be a return of the multi aspect sensor. And if it is, this would explain a few things. Namely the 6K photo mode. Nominal 16x9 6K (not full DCI spec) is 5760 x 3240, much like we call 3840 x 2160 4K. A 4:3 sensor with 5760 pixels wide would be 4320 pixels tall and thus a 24.8 MP sensor, quite different than any yet seen in Micro43. But if that width is only in 16:9, the 4:3 imager would be 20.9 MP... slightly more than half a megapixel more than in the GX8's sensor. So with a minimal reduction in pixel size, this theoretical multi-aspect sensor is possible.
In addition to the benefit of 6K photo mode, there's more magic in a 5760x3240 16:9 imaging area. That size is exactly 1.5 times wider and 1.5 times taller than 3840. 1.5 is a kind of magic number in this regard. Oversampling by a ratio of 1.5 allows for computationally easy downconversion with improved color accuracy at each output pixel. Don't take my word for it either: you can look at the specs for every different version of the Arri Alexa, which generates its 1920x1080 from a sensor with 2880 x 1620 pixels (which it now allows you to record all of, a compromise in the original design that attempts to eke more resolution out of the sensor at the expense of color accuracy).
In summary, it doesn't take much of a leap (basically, just Panasonic returning to multi-aspect sensor, and developing a new sensor design with pixels only marginally smaller than their current generation) to arrive at a possible GH5 with internal 10-bit 4:2:2, a horizontal crop factor of 1.8 or 1.9, vastly improved color accuracy in video (in addition to the greater color sampling rate), and a true 16:9 6K photo mode with no interpolation or computational trickery".
Pretty interesting post by Will on Newshooter. He might have sort of a answer on sensor size of the GH5. Food for thought..
