Canon EOS C100

[Nigel Barker]

The ergonomics are the biggest differentiator for me. The C100/300/500 are a new class of camera like a grown-up DSLR. Small & self-contained. The C300 feels really comfortable for me to use. There is a vast difference between the simplicity of a C300 & the ugly fussiness of the FS100/700 covered with tiny little buttons. The new cameras just continue with Sony's own strange ideas about what constitutes good ergonomics on a camera.

[David Heath]

Originally Posted by Nigel Barker

The new cameras just continue with Sony's own strange ideas about what constitutes good ergonomics on a camera.

Well, I need to see one in the flesh, but on first impressions the new cameras seem to have answered many of the criticisms about cameras such as the FS100/700 ergonomically. Viewfinder further forward and to the side, sensible shouldersupport thought about, things like that.

[Andy Wilkinson]

OK I'm going to need some advice here.

I've just checked the Canon EU specification page for the C100 which now seems to have details on it - as I remember for a long time all I got was 'not available' for that section so I had to look at the Canon USA spec pages (which as I remember were a bit vague about anything other than NTSC specs at the time). Link to Canon EU C100 Specs here:

Canon EOS C100 - Cinema EOS Cameras - Canon Europe

Ever since I ditched HDV (which I did as soon as I went professional in 2008) and especially since my stuff rarely gets broadcast/almost always goes to web I've only used progressive for my work. As I live in PAL land typically this is at 1080p25 on the EX3 and Canon 7D (yes I know it's not really that resolution on the 7D!) and more recently 1080p50 AVCHD2 on a little Panasonic TM900 for some cutaway/slow motion shots.

Now, and I don't know if this is related to what Philip is concerned about, but I see 25FP is listed on the specs with a little note stating that this is "recorded as 50i". I DON'T like the idea of anything being recorded as interlaced, AT ALL....

I want and indeed fully expected to see 25p listed - this is not a cheap camera and is supposedly a professional one, right? But I do not. I have a feeling this is the 25FP method that Canon used in some of their consumer camcorders from several years back. I have no idea what it means technically and implications on picture quality but I am now concerned. I need 1080p25, heck 1080p50 AVCHD2 would have been nice too.

Also, what are the implications of this FP thing regarding an external recorder taking 4.2.2 prior to in camera AVCHD compression?

Can someone who knows this stuff inside out please enlighten me please - as I have a C100 on pre-order and now need to fully understand this quickly.

Thanking you in advance.

[Nigel Barker]

Just as the C300 uses the MXF recording back end from the XF series it appears that the C100 uses the AVCHD recording back end from the XA10 (reduced cost baby XF100). The XA10 records 25p (PAL) and 30p (NTSC) as PSF (Progressive Segmented Frame). Which means that each progressive frame is divided into 2 fields and recorded as 50i (or 60i). When the files are imported into your NLE it will understand that it must interpret the footage as 25p (or 30p). So basically you don't need to worry as Premiere Pro or FCP will do the right thing.

[Andy Wilkinson]

Phew, thanks Nigel. Just reading Wikipedia and other web pages now to try and understand Progressive Segmented Frames more.

EDIT: Well I've now read a bunch of stuff and the main thing that concerns me is a mention that PSF may not handle chroma information in the same way as true progressive - but that might just be me not fully understanding this yet. There is also an excellent 2011 article by the Pro Digital Group out there that discusses all the things you must know about PSF. That one suggests Adobe CS5.5 Production Premium (what I mainly use now) will attempt to handle the clips as interlaced (and that will reduce quality of course) - unless I specifically direct it not to. I don't know if that changed when I updated to CS 5.5.2. Still reading and trying to get to the bottom of this chroma subsampling thing causing artifacts on coloured objects as mentioned in the Wiki article and in a few other places.

I also link to a thread that discusses this PSF thing regarding the Canon XA10 here on DVinfo.

http://www.dvinfo.net/forum/canon-xa...plain-psf.html

EDIT 2: Also, just found this for us CS 5.5 users.

http://provideocoalition.com/index.p...e_pro_cs5.5.x/

[David Heath]

Originally Posted by Andy Wilkinson

Phew, thanks Nigel. Just reading Wikipedia and other web pages now to try and understand Progressive Segmented Frames more.

EDIT: Well I've now read a bunch of stuff and the main thing that concerns me is a mention that PSF may not handle chroma information in the same way as true progressive - ..........

With true interlace, odd lines are captured together (1,3,5,7 etc) then even lines are captured 1/50 sec later (2,4,6,8 etc). The odd lines are transmitted as one FIELD, then the even lines as the other field of a frame. Then time for the next frame. This why interlace is so named - odd and even lines are interlaced!

With progressive all the lines get captured together, all at the same time. Easiest thing is simply to store or transmit them sequentially.

But it can be useful to record/transmit or otherwise deal with the progressive signal with equipment designed for interlace. Easily enough done with a framestore - just store a frame, then read out the odd lines one after the other, then the even ones, then on to the next frame. The important point is that the operation is transparently reversible. The only difference between p and psf is the order in which the lines are presented, the information for each line doesn't change.

That's the theory. Unfortunately, practical systems may not always interpret correctly. They MAY wrongly recognise the psf signal as ordinary interlace and do a deinterlace - which will actually change the information within each line, not just do a simple reordering.

You are also correct about possible problems with the chroma - though not in the case of 4:2:2 systems where each line is distinct. It's really down to interlace that 4:2:2 is seen as desirable as it is. Have you ever wondered why it's considered desirable for the chroma res to be twice as high vertically as horizontally? It's down to these sort of reasons, and the (interlace) TV signal not being symmetrical horizontally and vertically. 4:2:2 becomes far less relevant in the progressive world.

Move to 4:2:0 and it all gets a lot more complicated. Some systems cope well, but others misinterpret.

[Andy Wilkinson]

Thank you David. I was hoping you'd chime in and help me sort this out in my head!

One other question. Using 4.2.2 uncompressed out of the HDMI (which we are told is "the same" as out of the C300s SDI) will bypass all of this PsF stuff - or not?

[Tom Roper]

Andy, if you use a good recorder like the Nanoflash it's going to take the 4:2:2 psf output from the C100 hdmi and save it as true progressive anyway. It does the work of taking the split frames and recombines them into whole progressive frames before your nle ever sees it.

The difference from the old days of 24f from the canon xl1/xha1 days is that the sensors themselves began life as interlaced natively, so there was a drop in resolution from row summation just as there is now with native interlace, and your nle had to combine the fields to make a progressive frame.

The C100 sensor is different in that it will be progressively read as a whole frame at one time, then split into two fields (25, 30 fps) and output from the hdmi as 50i/60i, but since it natively originates from a progressive read of the sensor and sampled at 4:2:2, row summation is not needed to smooth alias artifacts as with an interlaced sensor. Each field is exactly half the frame, and the recorder will take care of recombining the two fields perfectly stitched into the whole frame. By the time it gets dropped onto the timeline of the nle, the frames have been recombined perfectly and it will be seen and be as the true progressive it originated as. This is the same way it works with the Sony hdmi/sdi output as well (or the c300). The stream out the sdi interface is 25/30p contained within 50/60i.

[Andy Wilkinson]

OK, now I've got it - thank you! Sorry for the naive questions everyone but I've not needed to REALLY understand this PsF thing until now.

You know I'd sware the original Canon USA spec page (that I viewed just after the C100 was announced) listed 25p amonst the frame rates - but I might have remembered that wrong. I know they originally listed the HDMI out as 4.2.0!

[Tom Roper]

I think it said it "records" to 4:2:0 (to the card) which is not the same as saying what comes out the hdmi. They don't assume that everyone who is using the hdmi will be recording from it, could just be monitoring for example.

[Mike Marriage]

Originally Posted by Nigel Barker

The new cameras just continue with Sony's own strange ideas about what constitutes good ergonomics on a camera.

The ergonomics appear excellent to me, although I wasn't able to test the optional shoulder mount which should make a big difference. It is a truely modular camera - it can be stripped right back for tight spaces, Steadicam or 3D, or built up into a well balanced shoulder mount including accessories. Important details like the top handle are rock solid unlike the wobbly one on the C300.

IMO the ergonomics are absolutely first rate.

Although I dislike the C300 ergonomics as I find in is simply too heavy once you add good glass and accessories, the C100 may well sneak under that threshold so I would love to try one. It may make an excellent B/C cam or travel camera.

[David Heath]

Originally Posted by Andy Wilkinson

One other question. Using 4.2.2 uncompressed out of the HDMI (which we are told is "the same" as out of the C300s SDI) will bypass all of this PsF stuff - or not?

From a practical viewpoint, I suspect Tom has already answered it: "Andy, if you use a good recorder like the Nanoflash it's going to take the 4:2:2 psf output from the C100 hdmi and save it as true progressive anyway. It does the work of taking the split frames and recombines them into whole progressive frames before your nle ever sees it. "

Technically, and as background, what should happen is that when a progressive signal is sent as psf, a flag should be set in the stream to tell other equipment that the signal is psf. So a recorder may be set to take in the psf stream, do a line reorder and then record the true progressive signal reconstituted. All this assumes that the original device sets the flag correctly - and the accepting device is capable of recognising and acting on it. And - you've guessed it, this is not always the case. Hence many of the problems.

If you want a bit of background, I found this article: ProVideo Coalition.com: TecnoTur by Allan Tépper

On top of this there are the chroma issues with 4:2:0, which at least you don't have to even think about with 4:2:2.

The obvious question that may be in your mind is "why"? Why make it so complicated, with so much that can go wrong? And it's probably that the HDMI output can drive many different devices, including monitors which won't support a native progressive signal. Hence, if carried over interlace, (psf) such a device will just treat it as interlace and give a result, whilst (hopefully) other devices will just be able to properly resurrect the progressive signal.

[Tom Roper]

Forgetting about HDMI and the need to assure compatibility with monitors, SDI works the same way, the stream goes out as 1080/psf 23.97 or psf 29.97 over 1080/i59.94 in NTSC parlance. As David says, the recorder can recognize through flagging or cadence when psf 23.97 is contained within 1080/i59.94 and lay it down as p23.97 to a mxf file with no problem. The problem for the recorder is when psf 29.97 is contained within 1080/i59.94. To that point psf 29.97 appears the same as regular 1080/i59.94, the recorder being unable to distiguish the difference. To remedy that, the user sets a bit informing the recorder when the 1080/i59.94 being received contains psf 29.97 segmented frames. The user-set bit tells the recorder to store the file as progressive frames.

The main point to take from all the confusion, is that Canon should know how it should be output for external recorders, and there should be no problem capturing perfectly reconstructed, fully progressive frames.

[Philip Lipetz]

Atomos has indicated that the C100 passes the necessary signals to the Ninja 2, including start and stop as you record,

[Andy Wilkinson]

Tom, David. Thanks again for taking the time to explain all the nuances of this. Much appreciated.

Philip, any news from Canon about your potential issue with the C100 yet?