Canon MPEG-2 Full HD Vs Sony AVCHD
 

I would like to know if the new MPEG-2 Full HD (4:2:2) recording codec recently adopted by Canon is easier to edit than the new Sony AVCHD. Also are the two codecs similar as far as picture quality?

http://www.simplydv.co.uk/2010/04/ca...hd-camcorders/

Stelios

Not an easy group of ?'s. I highly doubt anyone can answer them yet being that the Canon camera has not been released yet.

Yes. MPEG2 is easier to edit than AVC-HD. The Canon may not have been released, but the codec is effectively the same as on higher end Sony cameras.
You can't directly compare two codecs for quality - the quality depends partly on such as bitrate etc, but partly on the actual coder being used. So two coders producing AVC-HD at the same bitrate, from the same source pictures can give different qualities.

But the Canon 50Mbs codec is likely to outperform AVC-HD, period. For a start it's 4:2:2 - all flavours of AVC-HD are 4:2:0. It's a fully approved broadcast codec like AVC-Intra 100.

What AVC-HD will do is give pretty good quality at a fairly low bitrate (average 21Mbs) - likely better than HDV quality at a lower bitrate. The 50Mbs codec as used by the Canon is better, easier to edit - but takes up 250% more space on memory cards. There's always a trade-off.

Also, the AVC-HD codec is tending to be associated with lower range cameras. In Sonys case, the NX5 is 1/3", 1megapixel chips versus 1/2", 2megapixel chips of the EX. A more interesting comparison may be to see how the new Canon stacks up against the EX - which makes the most difference, the bigger chips of the EX, or the full 50Mbs codec of the Canon?

I use both MXF 50Mbs 4:2:2 files and AVCHD files on my computer. The MXF files don’t come from a new Canon XF camera, as along with everyone else I don’t own one yet. My MXF files come from AVCHD files as I sometimes use the MXF 50Mbs 4:2:2 files as intermediates.

I’m running a 4 core 2.83 MHz processor with 4Gb of ram, 2 Sata hard drivers and twin 22” monitors. If I place play an MXF file from the timeline in Vegas 9c then I see about 18% on the processor performance display. A 1 second dissolve between clips pushes the processor up to around 48% during the dissolve. The dissolves are judder free and all the playback looks smooth and judder free even with the preview set to best. The hard drive light flashes briefly at just under one second intervals and stays like this even with FX applied.

If I now place the same clip as a AVCHD file the processor uses around 55% and performing a 1 second dissolve pushes the processor up to around 88%. Most dissolves are judder free and the playback is smooth and judder free even with the preview set to best. The hard drive light flashes about the same as when playing back an MXF file.

If I apply a colour correction FX setting on a track insert then the MXF file playback pushes the processor to around 42% and 74% whilst doing a dissolve. The dissolves are still judder free and all the playback looks smooth and judder free even with the preview set to best. The same track insert colour correction setting whilst playing an AVCHD file will push the processor to 76% and 97% whilst doing a dissolve. Most dissolves are judder free and all playback is smooth and judder free.

Adding a second track FX of a brightness and contrast pushes the processor to 59% for straight playback and 92% with a dissolve whilst playing an MXF file. The dissolves are judder free and all the playback looks smooth and judder free even with the preview set to best. An AVCHD file with the same track FX settings pushes the processor to 71% and 99% during a dissolve. Only a few dissolves are judder free, but the rest of the play back is fine even with the preview set to best.

It’s hard to say if the MXF or AVCHD is the better looking codec as my MXF files are renders from the original AVCHD files from my camera. If I compare the original AVCHD file to a rendered MXF file playing back at a resolution of 1920 x 1080, then the AVCHD is much sharper. If I render to Canopus HQ or Cineform codecs then I can’t tell any difference from the original to the render, even at 1920 x 1080. But the MXF when viewed at 1920 x 1080 does lose something. At 1280 x 720 there is not much between the original and an MXF and at 720 x 576 I can’t see any difference.

The only other difference is the file size. The original AVCHD is around 27.5Mb for a 10 second clip whereas the same clip as an MXF is 63.6Mb. I find the MXF codec very fast for rendering to. A 10 second clip usually takes about 12 seconds to render to an MXF from an AVCHD original if it’s a straight render with no FX.

Rog, Thanks so much for the very in-depth reply ;). Awsome.

David Heath

"The 50Mbs codec as used by the Canon is better, easier to edit - but takes up 250% more space on memory cards. There's always a trade-off."

True, but of course the canon shoots to cheaper CF cards which will give people like me who've stayed away from P2 and SxS a good reason to go Solid State.

Hang on... It's just occurred to me that AVCHD shoots to SDHC cards that are cheaper than CF, but i find those cards a bit flimsy, while CF cards are much more robust IMO.

"A more interesting comparison may be to see how the new Canon stacks up against the EX - which makes the most difference, the bigger chips of the EX, or the full 50Mbs codec of the Canon?"

Agreed. I rented an EX1 last weekend, and while i thought it shot nice pictures, i'm more interested in the new Canon, even though it has smaller sensors. I prefer the ergonomics amongst other things. The glass looks fantastic too.


What i would say, is i was hugely impressed by the latitude of the EX1 at the grading stage compared to HDV. If the Canon is even better, then the Canon may well be my next camera. Unfortunately, in the Sonycentric UK it's tough finding a rental house for Canon cameras to try out, but with its professional codec, that may well change with the XF range.

Thank you guys for all the answers and especially Rog with the detail report.

Stelios

Dom, don't forget that the Canon comes w/ knee and shoulder settings that allow you to add lattitude to your shots. They change how the shadows and highlights are mapped from the sensor. (Kind of like a fake Log recording that you don't have to use a viewing LUT to see properly.)

I imagine very few people use these settings, but they can be quite useful if you need extra dynamic range.