HD-SDI 4:2:0 or true 4:2:2? at DVinfo.net

March 7th, 2006, 06:04 AM

Regarding the new Sony's PDW-F330 XDCAM HD, Michael Devlin info us from this here:

Quote:

Originally Posted by Michael Devlin

Sony has told me that the CCD/DSP image is only capturing 4:2:0. The 4:2:0 signal is then interpolated to 4:2:2 to produce the SDI output.

Is it the same thing at the Canon's XL-H1 or not?

John Jay · March 7th, 2006, 06:44 AM

Ask someone to shot some red, blue, green, magenta, cyan and yellow triangle shapes and then we will know for sure

Pete Tomov · March 7th, 2006, 07:25 AM

I don't see a reason for that since all the ccds have the same resolution.

Graeme Nattress · March 7th, 2006, 07:38 AM

There's no reason other than:

to save money on a cheaper DSP,
to differentiate between low end and high end products.

Graeme

Michael Devlin · March 7th, 2006, 09:18 AM

In the post referenced above I think I mentioned that we were comparing the HDSDI output from the Sony HDC-X300 (same image processing as the XDCAM HD) to a Canon XL H1. My impression is that the Canon is full 4:2:2. There are several tests we did which would support this. The first involved chroma keying and wire removal shots. We captured HD SDI from both cameras into Avid Adrenaline Media Composer HD using DNxHD 220 10bit. For Chroma Keying we tried both the Avid Spectramatte effect and the Boris Continuum Keying effect.

We found that the Canon pulled a very clean matte with almost no work. In fact, sometimes we could let Boris pull the matte with no manual cleanup. The resulting images were extremely clean, with very good definition on the edges, and only a few places where we had to define a manual matte or use other tricks. In contrast, it was very difficult to pull a clean matte with the X300, and we could never eliminate all of the edge crawl from most shots. As always, proper lighting was absolutely critical. Small changes in lighting can make it very difficult to pull a clean matte.

We had similar results with wire removal, where the clean edges of the Canon image allowed us to define a narrower wire region and disturb the background less.

We also monitored the HD SDI output in a Leader LV5750 HD waveform monitor / vectorscope which has a digital data dump feature. While trolling through the data is a bit tough, certainly my impression is that edges are transitioning in 1-2 pixels with the Canon, versus 2-4 pixels with the X300.

I should mention that on other tests (lattitude, vertical smear, etc.) the X300 greatly outperformed the Canon. If there is interest I can post a more complete comparison.

While the X300 is the "same" CCD and image processing as the XDCAM HD, I have an early X300 and I believe Sony has made some improvements since then. Playing with the PDW-F350 a few weeks ago the images looked pretty good. That is why we ordered one. Obviously we will run a more complete set of tests once the F350 arrives (end of this month hopefully).

March 7th, 2006, 06:44 AM	#2
John Jay Major Player Join Date: Sep 2002 Location: US & THEM Posts: 827	Ask someone to shot some red, blue, green, magenta, cyan and yellow triangle shapes and then we will know for sure __________________ John Jay Beware *PLUGGER-BYTES*

March 7th, 2006, 07:38 AM	#4
Graeme Nattress RED Problem Solver Join Date: Sep 2003 Location: Ottawa, Canada Posts: 1,365	There's no reason other than: to save money on a cheaper DSP, to differentiate between low end and high end products. Graeme __________________ www.nattress.com - filters for FCP

March 7th, 2006, 09:18 AM	#5
Michael Devlin Regular Crew Join Date: Mar 2006 Location: Zephyr Cove, NV & Anchorage, AK Posts: 82	Canon XL H1 vs XDCAM HD HDSDI output In the post referenced above I think I mentioned that we were comparing the HDSDI output from the Sony HDC-X300 (same image processing as the XDCAM HD) to a Canon XL H1. My impression is that the Canon is full 4:2:2. There are several tests we did which would support this. The first involved chroma keying and wire removal shots. We captured HD SDI from both cameras into Avid Adrenaline Media Composer HD using DNxHD 220 10bit. For Chroma Keying we tried both the Avid Spectramatte effect and the Boris Continuum Keying effect. We found that the Canon pulled a very clean matte with almost no work. In fact, sometimes we could let Boris pull the matte with no manual cleanup. The resulting images were extremely clean, with very good definition on the edges, and only a few places where we had to define a manual matte or use other tricks. In contrast, it was very difficult to pull a clean matte with the X300, and we could never eliminate all of the edge crawl from most shots. As always, proper lighting was absolutely critical. Small changes in lighting can make it very difficult to pull a clean matte. We had similar results with wire removal, where the clean edges of the Canon image allowed us to define a narrower wire region and disturb the background less. We also monitored the HD SDI output in a Leader LV5750 HD waveform monitor / vectorscope which has a digital data dump feature. While trolling through the data is a bit tough, certainly my impression is that edges are transitioning in 1-2 pixels with the Canon, versus 2-4 pixels with the X300. I should mention that on other tests (lattitude, vertical smear, etc.) the X300 greatly outperformed the Canon. If there is interest I can post a more complete comparison. While the X300 is the "same" CCD and image processing as the XDCAM HD, I have an early X300 and I believe Sony has made some improvements since then. Playing with the PDW-F350 a few weeks ago the images looked pretty good. That is why we ordered one. Obviously we will run a more complete set of tests once the F350 arrives (end of this month hopefully).

March 7th, 2006, 07:25 AM	#3
Pete Tomov Regular Crew Join Date: Sep 2005 Posts: 106	I don't see a reason for that since all the ccds have the same resolution.