XF100 - 422 vs 420 chroma key

[Pavel Sedlak]

I did a test of the XF100, 422 vs 420 with chroma key (Avid MC) at 1920x1080/50i (without any moving, there is no interlace issue).

There is a noticable difference, the "422" looks ok (i know that it is computed by de-bayering from 420 BAYER, from 1920x540 green, 960x540 red and 960x540 blue, but it is ok).

[Bram Corstjens]

So it's practically - in real life - , a non isue whether 422 or 420 is used for chroma keys? Interesting would be how this example compares to the EX1 series from Sony with their 35mbps VBR 420 codec.

The higher bitrate on the XF (50 mbps) would give a headstart, but the 50mbps is needed for a) intraframe compression and b) 422 colour space where the Sony's have the more efficient interframe compression which is variable too and have 'only' 420 colour space to encode.

My guess is that with very rapid movement of very specific scene's the 35 mbps VBR codec can break down a bit, but that in 99% of 'normal' shots it's very adequate - even for chroma keying.

[Peter Moretti]

I believe that the XF codec is Long GOP (interframe).

[Harm Millaard]

Both XDCAM and XF are long GOP MPEG2 formats. The only difference is 35 Mb VBR 420 versus 50 Mb CBR 422.

[Pavel Sedlak]

Originally Posted by Bram Corstjens

So it's practically - in real life - , a non isue whether 422 or 420 is used for chroma keys? Interesting would be how this example compares to the EX1 series from Sony with their 35mbps VBR 420 codec.

The higher bitrate on the XF (50 mbps) would give a headstart, but the 50mbps is needed for a) intraframe compression and b) 422 colour space where the Sony's have the more efficient interframe compression which is variable too and have 'only' 420 colour space to encode.

My guess is that with very rapid movement of very specific scene's the 35 mbps VBR codec can break down a bit, but that in 99% of 'normal' shots it's very adequate - even for chroma keying.

XF100 is not the king in chromakey, XF300 is much better. I did this test for more information about "debayering" from CMOS with 2mil pixels. I want know if the 422 colors are only "upscalled" from 420 (or if the interpolation durig the debayering gives 422 samples).

I think that this interpolation gives 422 precomputed samples, after this easy a chroma key test. The 422 colors are not only "upscalled" during the output compresion .

[Derek Green]

Originally Posted by Pavel Sedlak

XF100 is not the king in chromakey, XF300 is much better.

How much better really is the XF300 for greenscreen vs the XF100? In your opinion is it worth the extra $thousands? Do you have any shot samples you can share? I can't seem to find any comparisons on the web anywhere.

[Pavel Sedlak]

I don't have the xf300, but it has much more better lens and the three cmos chips, it gives true 422 color sampling. The lens of the XF100 is much worse and 422 colors are interpolated by deBayering.

[Derek Green]

Thanks for the info Pavel. Any chance you would be interested in shooting a person in front of a green screen with the XF100 for me? I would like to pull a key from this cam myself with real world footaage. I could compensate you for your time. Let me know thanks.

[Pavel Sedlak]

This needs some big lights, I have only camlights (the small TV studio will be fine). Sorry, I can't do it.
The green chromakey may be better than a blue (or red) because green samples are made from 1920x540 pixels, red and blue samples are made from 960x540 pixels by the deBayer interpolation.