EspenB

Sorry I fucked up on that picture. Ignore it.

Finally here is the proof. I took the E-M1 Mk II sensor and superimposed it on top of the GH5. Marked the Oly sensor with red line for extra clarity. GH5 sensor is 1,42:1 ratio. Same as the old GH2. However, it's only utilized in C4K mode according to Driftwood.

It was a reference to the GH5 sensor. If DCI 4K crop is 1,89 is MAR as well.

If the crop is less than 2 it must be MAR sensor...

Driftwood: "Heres a little bit of trivia... Bit of trivia... Panny GH5 in 4k DCI has a 1.89x crop factor."

Something fishy are going on! Even in the presentation video the sensor seems wider than the 4/3 GH4 sensor. Also, measurements seems to indicate a 1,4ish aspect rato. (Suggesting a multi aspect ratio sensor like the GH1/GH2). Yet, the pixel dimensons given for the various picture aspects are all the same pixel width. (And same height for 1:1 and 4:3 aspects). Strange! UPDATE: The EM-1 Mark II sensor (20ish MP) clearly is in 4/3 aspect. The GH5 sensor seems wider. Could be that they chopped off 1 MP at each side to improve read out speed. Or could be because of IBIS.

"Its now confirmed, the GH5 will offer maximum bit rate of 400Mbps in 4K video! The 400Mbps will come as a firmware update on the second half of 2017. Right now, the the camera will be launched with 150Mbps." 43rumors

There is a working Gh5 sample at Photokina thats for sure...

It can take 60 pcs of 4K frames and then you can select which one captured the moment best.

Yep, its the rumored multi aspect sensor from may. It has the same sensor dimensions as the previous GH2.

Yes, current 20 megapixel sensor is 27 fps at 12 ch readout. The new sensor is 24 channels read out and multi aspect (18.9mm*13mm). Compared to a regular 4/3rds sensor (17.3 x 13 mm) that new sensor seems to have an aspect of 1.45 (vs 1.33 for standard). EDIT: The GH2 sensor is also approx 19 mm wide in video mode. (DPReview says the GH2 sensor is only 18.89mm on the long dimension)

re: B-frames Neither P og B frames is worse than I frames quality wise if the quantization level is identical. All prediction in the encoder is done on the allready compressed frames, so there is no accumulation of the prediction error (which is what is stored for P and B frames).