Luke Mason

It's called colour-aware pixel binning, which Canon used on C300 to produce 2K 444 image. It's different from simply reading out a group of pixels.

Canon uses large group pixel binning (3x3 or even bigger) to get HD image, the RAW bayer might only have 500-700 lines of resolution.

Again, it's impossible to get the highlight roll-off from real C-Log by using picture profile.

The global shutter option "C700 GS PL" has very clear details, it's just that you missed it. Advertised DR is 14 stops, one stop less than regular C700. Sensitivity on C700 GS seems normal: 4K 60fps or lower/2K crop 120fps or lower: [*200 - 320 - 25600 - *102400] 4K over 60fps/2K crop over 120fps: [*400 - 640 - 25600 - *102400] There might be a slight drop in base sensitivity, similar to F55.

C-Log is not merely a gamma curve, it maps colour differently (within BT709/BT601 color space in the case of 1DX II/1DC)

The title clearly states that "Canon 5D Mark IV brings dramatic dynamic range improvements to the 5D line” They were comparing it mostly with 5D Mark III.

You said it yourself, H264 is the basis of XAVC and XF-AVC, there were a lot of changes made to them and they are distinctively different from consumer H.264 designed for exhibition.

HEVC/H.265 was not suitable for and was not designed as an acquisition format, it's too computationally intensive for both encoding and decoding.

4K 60p

23.976 and 24.00 do not matter in image acquisition, they can be converted easily and safely from each other, for distribution you need both, one for TV/web, one for theatrical release. F5 and A7 series cameras are not capable of shooting 24.00, while C300 II, GH4, 1DX II and 5D IV can. For creative choice the aspect ratio can be anything imaginable, DCI 2.39:1 and 1.85:1 are just containers, they can be padded to contain arbitrary aspect ratios, for example, a lot of cinematographers started framing 2:1 (see Jurassic World).

Hope one day the fan noise would drop as well...

If you are doing comping and blending operations, it's better to use linear working space with 32bit floating point.

It's not highlight aliasing in this case, highlight aliasing on Sony cameras usually comes with magenta coloured egdes. Just compression artifacts, can be mitigated by recording externally.

The sigma ones are the best, and they work perfectly with DPAF.

I always calculate crop using horizontal dimension (whether it's mm or pixel count), in the case of a standard M4/3 sensor which is 17.3mm wide, the crop factor is 36/17.3= 2.08x, which we simply say as 2x. Diagonal size is too complex and yields the same results. The 1.06x thing you mentioned is also pointless, not because nobody says that, it's just an incorrect way of calculating.

haha what? we all "do like that", when talking about video modes that do not use the full sensor area. M4/3 is always 2x crop (I'm not aware of any camera that do not use full M4/3 area) , 2.08 is just a more accurate number.

Yes that math only works assuming the aspect ratio is consistent, usually crop sensor is the same aspect ratio as FF sensor, all 3:2. Video crops differently. 1.64 is actually extremely close to a Super 35mm size, I guess Canon took this into account when designing this 30.4mp sensor.

All 1DC and 1DX II 4K films were shot with MJPEG. "Knock Knock" starrng Keenu Reeves were shot with 1DC.

Hurt only in terms of storage space, MJPEG codec @ 500Mbps and 4:2:2 is actually a very high quality codec, the best codec among DSLRs and the only 422 codec used apart from cinema cameras.

Sensor is 3:2, 4K is 17:9, you need to calculate using a 17:9 area of the sensor which is 6720x3454 sqrt of 6720^2 + 3454^2 = ~7597 7597/4631 = ~1.64

For pure sensor size comparison, yes. But hey man, we are talking about video mode with different aspect ratios.

It's a perfect Super 35mm crop, same size as many other cinema cameras. Some people don't like it and prefer FF look, but it's definitely not useless.

I'm afraid your understanding of crop factor is completely wrong, M4/3 sensor is 17.3mm wide, FF is 36mm wide, 36/17.3 = ~2.1x crop. You always use horizontal dimension to calculate crop factor.

You don't even need to adjust the aspect ratio, just use the horizontal pixel 6720/4096 = ~1.64. I was trying to demostrate to Andrew that both work and give the same result, seems like someone skipped a lot of math class lol.

Measuring horizontally and diagonally gives you exactly the same results, 1.64 on 5D Mark IV. HD mode would be 1.0x crop (6720/6720=1) because the camera is using/pixel-binning from the entire 16:9 area of the sensor. Does 5D Mark IV shoot 3:2 video? No. So you measure the corresponding area using video aspect ratio when you calculate crop factor, in this case, 17:9 and 16:9. See my edited post for more examples.