Canon adopts MPEG-2 Full HD (4:2:2) file-based recording codec for upcoming cam

[Robert M Wright]

Me thinks a lot of folks are going to turn blue, holding their breath for Canon to announce a low cost (like as if somehow under $10k) camcorder that is something akin to a handycam version of a RED ONE (not the low-end 2/3" Scarlet that so often gets referred to as evidence of the feasibility of cheap camcorders with full frame sensors - very apples to oranges thinking - and Scarlet is still vaporware anyways).

 

[Robert M Wright]

...Secondly, the hardware required to process that data would be much more expensive.

It's the R&D that costs an arm and a leg, not manufacturing the silicon wafers as a result (if it's on a mass production scale).

 

[studio1972]

More surprises (and potential arguments) in store?

http://www.hdwarrior.co.uk/2010/02/04/canon-to-announce-new-camcorder-at-bafta-in-london/

Really interesting rumour.

If they actually did a large sensor camera like this which was specifically for video they would need to solve the aliasing issue (either a new sensor, or perform a down conversion from the original sensor to 1080p, i.e. not simply sampling a small selection of the pixels) and also have a set of removable lenses (a 10x zoom would only be possible with a slow lens and that would defeat the purpose of the large sensor).

I think the lenses would be pricey, though it depends on how big they think the market is. I am assuming they would have autofocus and probably motorised zoom.

Maybe a 2/3 inch sensor would be more realistic as they already make video lenses for that size? That would be pretty cool IMHO and probably more usable for a wider range of users.

 

[Cranky]

It's the R&D that costs an arm and a leg, not manufacturing the silicon wafers as a result (if it's on a mass production scale).

Yep. And with MPEG-2 being aroung for at least 15 years all R&D is already done and all kinks are ironed out. MPEG-2 @ 50 Mbit/s will use no more processing power than AVCHD, that is I am sure of. Cannot say for sure as I don't have access to a Sony XDCAM 422HD camera.

It is another question whether consumers want it or not, but technically it is completely viable.

 

[studio1972]

It's the R&D that costs an arm and a leg, not manufacturing the silicon wafers as a result (if it's on a mass production scale).

I know, but a consumer cam with a pro codec wouldn't sell much as consumers want something they can input to iMovie etc or plug the SD card straight into the TV. So the volumes would be low and the price high. The hardware for current consumer cameras is already available and cheap as there are lots of them all using AVCHD for example.

Plus, bigger more complex chips cost more than smaller simple ones in any case. Then you've got battery consumption, memory requirements etc.

 

[Spartacus]

It's the R&D that costs an arm and a leg, not manufacturing the silicon wafers as a result (if it's on a mass production scale).

well, that´s why platform production is all over the place in a lot of industries.
the codec part for this camera should be more or less proven and tested already, the same goes for the audio segment a true video camera needs, all there already.
Recording video to SD cards, well that´s what Canons photo dept. does already.
A larger sensor? yep, also can be borrowed from the photo dept.
A new lens? um, we´re talking Canon here, so...

R&D isn´t all about pioneer work for a camera like this, it´s just looking on what´s already in the company and recombining it.

 

[Cranky]

I know, but a consumer cam with a pro codec wouldn't sell much as consumers want something they can input to iMovie etc or plug the SD card straight into the TV.

MPEG-2 is MPEG-2. If decoder is written correctly it should have no issues with either 25 Mbit/s HDV or 50 Mbit/s HD422. As to plugging cards, HDV is tape-based and is not "pluggable", still there were quite a few of HDV consumer camcorders sold.

 

[PDR]

MPEG-2 is MPEG-2. If decoder is written correctly it should have no issues with either 25 Mbit/s HDV or 50 Mbit/s HD422.

Usually not an issue for PC users, but Quicktime's MPEG2 decoder playback component doesn't work with 4:2:2. (but native AVCHD isn't so hot on MAC's either...)

 

[Huy Vu]

MPEG-2 is MPEG-2. If decoder is written correctly it should have no issues with either 25 Mbit/s HDV or 50 Mbit/s HD422. As to plugging cards, HDV is tape-based and is not "pluggable", still there were quite a few of HDV consumer camcorders sold.

Ok, but would the average consumer care? You're applying a professional mindset to a consumer product. It's the Playstation 3 versus the Wii. Faced with a choice between a technically "better" codec/system and a much lower price point, your average Joe will go with the cheaper price everytime since they won't be able to tell the difference.

 

[Cranky]

Ok, but would the average consumer care? You're applying a professional mindset to a consumer product. It's the Playstation 3 versus the Wii. Faced with a choice between a technically "better" codec/system and a much lower price point, your average Joe will go with the cheaper price everytime since they won't be able to tell the difference.

Having multiple codecs on camera is not a problem. Use it or not, your choice. JVC was offering both MPEG-2 HD and AVCHD last year in its models transitional from TOD. Once designed, putting it into firmware costs next to nothing.

BTW, I like Wii because it was probably the first gaming console that combined an electronic game with physical motion.

 

[Huy Vu]

Having multiple codecs on camera is not a problem. Use it or not, your choice. JVC was offering both MPEG-2 HD and AVCHD last year in its models transitional from TOD. Once designed, putting it into firmware costs next to nothing.

Having an extra codec option won't make the consumer more likely to buy the camera. It's not worth implementing a feature if it's not going to net you extra sales. Consumer cameras are stripped down to the bone feature wise so they can make as many as cheaply as possible.

I would expect Canon to do nothing less than implement this in an A1 or a newer L1, simply because they don't make cameras that are over $10k. Canon just isn't in that game historically and it seems more than a coincidence that they would show mockup of this new camera and then announce MPG-2 support so soon after.

 

[Dino]

well, that´s why platform production is all over the place in a lot of industries.
the codec part for this camera should be more or less proven and tested already, the same goes for the audio segment a true video camera needs, all there already.
Recording video to SD cards, well that´s what Canons photo dept. does already.
A larger sensor? yep, also can be borrowed from the photo dept.
A new lens? um, we´re talking Canon here, so...

R&D isn´t all about pioneer work for a camera like this, it´s just looking on what´s already in the company and recombining it.

I think I made this point previously on this thread, except about what Panasonic is showing at NAB in a large sensor camera. They will leverage their SD card knowledge, AVCCAM codec and CMOS sensor knowledge, along with their palmcorder know how to build an inexpensive prosumer camera to fight against VDSLR's, Scarlet and whatever else is coming from competitors at NAB.

The more I think about it, such as what lens mount would be used, the more I think the Panasonic offering is going to be 2/3"(maybe even CCD, not CMOS), which makes it useful to a larger segment of prosumers and keeps their lens partners happy. Obviously RED saw a need for a small form factor 2/3" sensor camera.

I think Canon felt the need to get into tapeless, but offer something above AVCHD, even XDCAM EX, and closer to AVC-Intra 100. This gives Canon two differentiators, codec and lenses, even if they stay with 1/3" sensors this year.

 

[Huy Vu]

I think Canon felt the need to get into tapeless, but offer something above AVCHD, even XDCAM EX, and closer to AVC-Intra 100. This gives Canon two differentiators, codec and lenses, even if they stay with 1/3" sensors this year.

Canon can always do what Panasonic did for the HVX200a by keeping the sensor class 1/3'', but increases the size of the chip slightly to accommodate more pixel or get better light efficiency.

 

[Dino]

Canon can always do what Panasonic did for the HVX200a by keeping the sensor class 1/3'', but increases the size of the chip slightly to accommodate more pixel or get better light efficiency.

Huy,

Canon will have to go to CMOS for full raster pixel count, lower noise and power consumption--much as Panasonic did with the HPX300. The question is, how much less will their new model cost than an HPX300? I prefer the AVC-Intra 100 codec and HPX300 form factor and ENG lens approach. I've never had much interest in Canon video cameras, but if they can do the new codec in a camera for under $5K, that will be a great differentiator in the prosumer space.

 

[Cranky]

CH vouched for CCD because, partially, "Canon doesn’t currently make a 1/3″ CMOS chip anyway". What about 1/2.7-inch chip used in the HV10/20/30/40, HG10 and DC50? Who makes it?

 

[Robert M Wright]

Canon doesn't make CCDs. Canon does make CMOS chips. The only reason Chris mentioned for Canon going with CCDs (that I am aware of) is that they have no experience designing a 3 chip imaging block with CMOS. I'm not sure why that would be any sort of serious issue though.

Frankly, I think it would be pretty foolish of Canon to go with CCD, when they are going to use a codec that could really take advantage of full raster progressive imaging chips.

 

[Robert M Wright]

The other reason I think it would be wise of Canon to go with full raster CMOS chips, is that if they want this cam to be seriously competitive for close to a half decade (like the A1/G1/H1 has been), low res chips would likely be pretty weak for remaining competitive that long. How likely is it that new cams being introduced by Sony and Panasonic a couple years from now will only resolve something like 600-700 lines of detail? The NX5U already exceeds that considerably, and I'd be mildly surprised if we don't see a successor to the HMC150 with a considerably higher res imaging block in the next year or two. Yeah, yeah, I know resolution isn't everything. It's hardly a trivial detail though. These are at least alleged to be "high definition" camcorders, aren't they?

A camera that reasonably approaches resolving 1000 lines of detail and records 4:2:2 color at 50Mbps MPEG-2? Yeah, that could still be selling pretty well in 2015.

 

[John_in_NY]

4:2:2 50Mbps long GOP MPEG-2 should pretty much smoke AVC-I (and AVCHD). At the same bitrate, using interfame compression should fairly easily cover the gap in efficiency (image quality per bit wise) between AVC and MPEG-2 (for most typical footage).

It's not like intraframe only compression offers much advantage for editing either, compared to long GOP MPEG-2. Modern mainstream desktop computers are plenty potent enough to handle long GOP MPEG-2 quite easily nowadays.

I really wish Canon had gone with 32Mbps AVC (at 4:2:2) though.

I am especially surprised that it is MPEG-2, which I thought was no longer being further developed.

 

[Robert M Wright]

I'm not entirely sure what you mean by no longer being further developed. MPEG-2 technology has certainly matured, but is still a very viable choice for an acquisition codec, offering distinct advantages (and disadvantages).

 

[Finnegan]

I'd be pretty surprised if a Scarlet with anything bigger than a 2/3" imager costs less than $10k.

Well, be surprised because RED lists the SCARLET S35 and FF35 brains at $7500 and $9000, respectively. Course, with the full peripherals, media, lenses, and other such necessities it will be well over your $10k mark.