Canon EOS C100

[Philip Lipetz]

Some people,such as Andrew Reid, suggest converting AVCHD to ProRes prior to grading to stop secondary losses, and that then AVCHD acquisition works in long delivery chains. What do you think?

[Joe Lawry]

If you HAVE to work with AVCHD. I would re encode it to an I frame codec straight away.

I've just spent a week in Da Vinci grading a show that was shot Sony FS700. They used an KiPro Mini recording ProRes422 but there was also a lot of offspeed stuff (200fps etc) which of course ends up AVCHD.

I was able to push the ProRes422 MUCH further than the AVCHD even once I had re encoded it into ProRes as well. I managed to break the slowmo footage codec as soon as I started to dive into the mids and blacks.

Straight out of the box, the AVCHD & ProRes did look nearly identical. I checked.

The HDMI connector is the only thing putting me off the C100.. There's no way i'd be using it without a recorder of some kind.

[Dom Stevenson]

Originally Posted by David Heath

In terms of the first generation material, then yes, it may be hard to tell the difference.

But apart from grading/green screen (as you mention) there are other things which can show issues with AVC-HD.

Firstly is processing power required for post work - codecs like XDCAM422 are more computer friendly than AVC-HD.

Secondly are the implications of cascading codecs in a production or broadcast chain. There have been lots of tests which show a "falling off a cliff" effect as images get successively decoded/recoded. For a few generations there doesn't appear much degradation - then suddenly quality goes rapidly downhill.

If you know the entire production chain - such as you are producing the final Blu-Ray etc - then this may not be an issue. If the final product is acceptable - fine. But if the work is being passed on to someone else (such as a broadcaster for final compression for transmission) it's wisest not to take the risk - use a codec that's better than AVC-HD. Even if the edited master seems OK, the final compresion could be the one that sends it off the cliff.

Yes i'm well aware of all of the above David. My point is, that at this price range there will be lots of interest in this camera from people who are not going to broadcast (or heavy grading) and the camera's inbuilt recorder will be fine. Those who want more can stick a box on the back. A win win for everyone.

[Philip Lipetz]

Originally Posted by Joe Lawry

.

The HDMI connector is the only thing putting me off the C100.. There's no way i'd be using it without a recorder of some kind.

At least it is a locking HDMI system.

[Nigel Barker]

The Canon XA10 uses the same sensor as the XF100 but instead of the 50Mbps MXF format it uses 24Mbps AVCHD. However MXF is basically MPEG2 whereas AVCHD is MPEG4 which is 2-4x more efficient in compressing an image for video. Therefore the video straight off the memory card for the XA10 & the XF100 is indistinguishable & I suspect that it will be the same with C100 & C300. MPEG2 is much easier to edit on older computers but recent systems don't have a problem with editing native AVCHD files in Premiere or FCP X (probably Edius & Vegas too but I don't use either of those).

[Philip Lipetz]

Here is a five minute hands on review of the C100. Canon C100 Preview on Vimeo

It is in French. My French is rusty, but the only new information I got from the video is concern about using the EVF while wearing a hat. The reviewer uses many words of praise but it seems that he hasn't used the camera, only held it.

[David Heath]

Originally Posted by Nigel Barker

However MXF is basically MPEG2 whereas AVCHD is MPEG4 which is 2-4x more efficient in compressing an image for video.

The normal rule of thumb quoted is that MPEG4 can be *UP TO* 2x more efficient than MPEG2 - I've never heard it claimed that it can ever be anything like 4x as efficient?

And all that can really said is that "it is more efficient". How much depends on a number of factors, and not least are the individual coder (some don't give much improvement over MPEG2 at all) and the actual bitrates used.

In general, the lower the bitrate, the higher the improvement factor. So compared to 50Mbs MPEG2, using MPEG4 doesn't offer much benefit (certainly nowhere near 2x). Talk about the bitrates used for broadcast transmission (19Mbs MPEG2 or less) and the use of MPEG4 with very expensive encoders pays real dividends and it's here that you may indeed get improvements of 2x, maybe even a bit more.

It's worth thinking how it achieves these improvements, which are (put simply) along the lines that it uses extra tricks to try to mask the imperfections that would be seen with simple MPEG2 - such as varying the block size. The point is that the better the MPEG2 encode is, the less there is to try to disguise, so the less point to the MPEG4 tricks.

[David Heath]

Originally Posted by Philip Lipetz

Some people,such as Andrew Reid, suggest converting AVCHD to ProRes prior to grading to stop secondary losses, and that then AVCHD acquisition works in long delivery chains. What do you think?

I think it will help, but is not as good as recording to XDCAM422, AVC-Intra 100 or ProRes or whatever from the start.

Following on from above, AVC-HD (and MPEG4) improve on MPEG2 coding by using tricks to help mask the flaws. One consequence of that is that the basic flaws are still there, and can still have a bad effect on subsequent codecs in the chain - even if the MPEG4 does a good job at the first generation.

Transcoding at an early stage will certainly improve the situation compared to working with AVC-HD all through grading etc, but far better to use a better codec at acquisition.

That is not to say AVC-HD is "bad". It's not, and I fully agree with what Dom says above - "My point is, that at this price range there will be lots of interest in this camera from people who are not going to broadcast (or heavy grading) and the camera's inbuilt recorder will be fine."

Expect 24Mbs to give you better than HDV quality, and full raster, at a lower bitrate. Don't expect it to be as good as XDCAM 35Mbs, and certainly not XDCAM422. Horses for courses.

[Andy Wilkinson]

As always David, excellent insightful comments to the technical discussion we like having on here.

Thank you.

[Paul Cronin]

Good points David, and once again it proves you do get what you pay for in camera and codec. XDCAM 50Mb/s 422 is fantastic.

[David Heath]

It's worth thinking of why AVC-HD was developed in the first place - as a way of recording video on to consumer memory cards with limited speeds and capacities.

It's main implementation is at slower than 24Mbs speeds in consumer gear, the 24Mbs mode was intended to bridge the gap between the consumer and true pro solid state.

But ironically, the first products were barely released before it was proven that ordinary SD cards could record 35Mbs XDCAM via adaptors in the EX. And current SD cards are far better performance than a couple of years ago. The moment that was demonstrated, the real question to ask was what was the point of AVC-HD at 24Mbs?

It wasn't as high quality as XDCAM 35Mbs, it was far more difficult to process - the only real advantage was a bitrate saving for a given quality. But it's only about 30% less than XDCAM 35Mbs - for lower quality and more difficulty of processing. Personally, I'd rather just get a couple of extra cards, and accept the somewhat higher file sizes. At least at SDHC card prices per GB - it may be different if you were talking about SxS or P2.

If pro solid state video meant SxS/P2 costs, AVC-HD meant consumer media costs, you can can see the point to AVC-HD. But it's not the case, and that's emphasised with the Canon cameras that record XDCAM422 to Compact Flash.

Yes, all the above is academic to an extent. As users we can only use what manufacturers will make. But do bear in mind that at the end of the day it's down to marketing. An XDCAM422 coder shouldn't cost any more to put into a camera than an AVC-HD, and neither should there be issues of power etc - and nowadays either is perfectly happy with consumer grade solid state. All that holds true not just for Canon and the C100 but for Sony and Panasonic as well.

I suspect the bulk of the cost of a camera is in the R&D and inital design costs, which obviously have to be recouped - and hence the cost of something like the C300. In that case, a lot of the same costs apply for the C100, so we should be pretty grateful that we do seem to be getting most of the C300 good points at much lower cost. But at the risk of sounding churlish, I'm also only too well aware that the C100 could be so much better, at probably no extra production cost, with the XDCAM422 codec - it's marketing that's kept it out, not real costs.

[Mikko Topponen]

Originally Posted by Nigel Barker

The Canon XA10 uses the same sensor as the XF100 but instead of the 50Mbps MXF format it uses 24Mbps AVCHD. However MXF is basically MPEG2 whereas AVCHD is MPEG4 which is 2-4x more efficient in compressing an image for video. Therefore the video straight off the memory card for the XA10 & the XF100 is indistinguishable.

That is not true. The XF100 is compression wise considerably better than the FS700 which uses the AVCHD codec at 24mbps. The FS700 will have blocking and mushing in any moderate movement. The XF100 will not. You may not notice these in mids of action but I saw enough while I was editing the FS700 material to come to this conclusion.

The XF100 codec is excellent.

Also some odd suggestions here. Why do people do many generational transcodes? I always keep the material as original files, right through to the end. If send to grading, we will make a .dpx sequence. I don't see any need or wisdom in transcoding to prores (unless you are mac guy) or sending material to grading as another newly encoded prores or avchd file. That's just degrading the material further.

[Paul Cronin]

Mikko I agree to stay in Native Codec. Also I agree the XF100 is a great codec since Canon worked out a deal with Sony to use XDCAM422 50Mb/s on their XF100/300, C300. Just shows you how good that codec is. I have used it for years and it stands up, there is a reason it is used by so many broadcast companies.

[David Heath]

Originally Posted by Mikko Topponen

That is not true. The XF100 is compression wise considerably better than the FS700 which uses the AVCHD codec at 24mbps. The FS700 will have blocking and mushing in any moderate movement. The XF100 will not.

Which backs up well comments about the relative efficiencies of MPEG4 and MPEG2. It may be about 2x at much lower bitrates - it's not anything like at these sort of rates. If it was true, then 24Mbs AVC-HD should be comparable to XDCAM422 at 50Mbs - as Mikko and others have found, it's not, and that's only on first generation comparisons.

And that is NOT to say AVC-HD is "bad". It's good - but not a true professional codec.

Also some odd suggestions here. Why do people do many generational transcodes? I always keep the material as original files, right through to the end. If send to grading, ..........

Unfortunately, not doing that is not an option for some users, especially in such as broadcast. And some codecs are suitable for some purposes, others better for other uses. Long-GOP may be optimum for acquisition and such as satellite links, I-frame only may be better for video servers for example.

Just imagine a broadcast scenario where material is shot, sent FTP to a local studio, recorded, edited and mixed, then sent by satellite to the transmitter and finally compressed for transmission. That's likely to be 5 steps of compression/recompression at least and it would be easy enough to see occasions when it could be more.

It would be great to remain uncompressed throughout - but that's obviously impracticable. The problem is to devise guidelines that are viable, without compromising quality too much. It's far easier if your production route is short and defined (say shoot, edit, and make Blu-Ray) than in an example such as above.

Add in to all this the way aliasing can also act to screw up codecs within the chain, and you start to see why such as broadcasters have to try to set rules. "It looks OK" to the eye on the first generation just is not adequate.

[Nigel Barker]

I suspect that the majority of us on this forum will never be producing anything for broadcast & are generally shooting & editing for the Web or DVD/Blu-ray disc in which case AVCHD may well be good enough for our purposes just as HDV was good enough for our purposes & before that DV & before that VHS were also good enough.