Canon introduces XF105 and XF100

[Glen Vandermolen]

Cool! Please try a low light test. I'm curious as to how the single chip will perform.

[Jim Martin]

Yep, thats why its here....we're doing a low light test for 2 of our clients (big reality producers) who are very interested in the low light/ IR capabilities to use on their many shows.

Jim Martin
FilmTools.com

[Andy Wilkinson]

Excellent news Jim. Look forward to hearing (and hopefully seeing) your test results.

Also, for those that follow Mac Video there are two new Canon interview films where the XF100/105, 300/305 and 60D, other DSLRs (and a brand new DSLR lens) get discussed by Canon sales staff. More a sales pitch than anything - but it's good to see the key points about the XF100/105 spelled out as you watch it sitting next to it's brother and sister. Link below:

MacVideo - Camera Technology - Interviews - Canon XF 100/105 and XF300/305; plus DSLR update with EOS 60D and 14 - 15mm fisheye L series lens

Also, here is a snap my mate took with his 7D showing the fateful moment when I got a XF100 (or was it 105, can't remember now) in my hand at the recent Canon Pro Photo event in London. Note my concerned look as I realise that the next few minutes might become another "expensive moment" for my business purchasing decisions in early 2011. He's got a video clip of me handling it too - but I look really ugly when drooling...

[Jim Martin]

Okay....We spent about 2 hours playing with the 105 and here are some thoughts.....

1) as the ladies here said..."It's so cute" 3.3lbs, compact, and really feels solid.

2) we looked at the picture via HDMI thru a KiPro, changed to HD SDI and onto a TV Logic 50" LCD.....and we were very surprised on how sharp the picture was....it handled all the most minute details very well, the corners were fine with no CA.....in fact it was so sharp, it made me look older!

3) we next played with the gain with the lights out (some ambient coming in from other rooms)...the noise at +33 was somewhat as expected but still quite usable. When we moved down 1 step (I think it was +25), the noise appeared to drop quite a bit....quite remarkable for that much gain.

4) then we went to use the IR.......in the green vs. white debate, everyone in the room liked the white better and when we used the IR illuminator, the white seemed to work better. As for the Illuminator itself (located on the front of the mic), the 1st 6ft we quite strong and was able to produce distinct shadows when my hand was put in front of my body on a black shirt. From 6 to 10ft or so, the illuminator still did a good job with the hand shadow more diffused. We then went into a completely dark, large closet and, just using the LCD, it was very easy to see, focus,etc......our clients gave it a big tumbs up.

All in all, very impressed...but still no solid word on the pricing (I expect that we'll here soon on that).

As for the size, we made a quick, fun video comparing the 105's size to a large avocado and it should be up on our youtube site later today.

Jim Martin
FilmTools.com

[Andy Wilkinson]

Hi Jim,

Excellent post. Thank you. I also had a quick look on YouTube but could not find your much anticipated Canon XF105 footage. Is it up yet?

[Ivan Pin]

Night video:

Canon XF100, de nuit on Vimeo

1080p25, 50mbps, 4:2:2, Shutter 1 / 50 (except for the bumper cars at 1 / 25)
Gain between 0 and 6dB,
Shooting without a tripod in hand,
Autofocus "Instant AF".


Timelapse:

Canon XF100, Timelapse on Vimeo

1080p25, 50Mbps, 4:2:2, Shutter 1 / 50
2 frames all over 3h 4s.


Infrared function:

Canon XF100, Infrared on Vimeo

1080p25, 50Mbps, 4:2:2
Everything is automatic.

[Jim Martin]

Originally Posted by Andy Wilkinson

Hi Jim,

Excellent post. Thank you. I also had a quick look on YouTube but could not find your much anticipated Canon XF105 footage. Is it up yet?

We had to re-shoot the peice on Friday....its being edited now and should be up later today or tomorrow....without the avocado....and we did not add any actual footage because we had a very limited time with the camera....but the stuff above looks very good and seems to cover what people want to see.

Jim Martin
FilmTools.com

[Robin Davies-Rollinson]

I'm so impressed with the French material - even for web video it looks good.
I'm definitely going to look out for this little camera when it reaches these shores next year...

[Paulo Teixeira]

If only you can compare it to a couple of it's competitors, the JVC HM100 and the Panasonic HMC40.

[Robin Davies-Rollinson]

Maybe, but it's the 50Mbs capability that's the clincher for me since I need to shoot material for broadcast sometimes in the UK...

[Brian Drysdale]

Yes, the Canon XF 300s are the only 1/3" camera on the BBC's HD camera list. They'll need something to replace all those Z1s that they've been using in recent years.

Although, the sensor may be the stumbling block for this new camera being used for more than inserts.

[Andy Wilkinson]

Thanks for pointing us at those test videos of the XF100 Ivan. Looks pretty promising for such a small cam.

The night time shots show that this camera is certainly capable of some pretty decent stuff at 0 and +6dB gain. Sure, I don't expect it'll be quite as good as my Sony EX3 or Canon 7D at night/in low light due to sensor size etc. but it looked OK from a quick viewing just now (purely from a noise aspect, ignoring other factors).

I wonder what the JVC HM100 and Panasonic HMC41 would have made of those types of shots? (as I guess they are the most directly comparable competition in this size of cam/price point).

[Dave Haynie]

Originally Posted by Glen Vandermolen

It would make for an intersting b-cam. I like that codec.
But how will a single 1/3" CMOS perform? Is it superior to the competition's 1/4" 3 CMOS?

I'm really concerned about the 1/3" sensor. Not so much that it's a single 1/3" sensor, but that it's only 2Mpixel. They certainly did that for light concerns... with the Bayer filter, they're down to 1/3 the sensitivity of a 1/3" 3-chipper, and even a bit less than a 1/4" 3-chip camera, all things being equal.

But if you look at the $1000-ish market, the modern Canons hold their own, pretty much, against the Panasonics, despite the single chip. They do this largely by using an 8Mpixel sensor. Sure, that gives consumers a reasonable still photo mode (stills on my HMC40 and TM700 are questionable at best), but it also allows the camera to use pixel-binning -- same reason DSLRs and high-end cinema cameras can use a single sensor and not screw you with color errors.

With the single chip at 2Mpixel, they're going to have the same color issues as my old Sony HVR-A1 and Canon HV10. Also, not sure how you get any digital zoom of interest out of that chip without a corresponding loss of video quality. At least with a large sensor, you can do a modest digital zoom and not do much damage to the image (you lose the pixel bucketing).

Sure, the higher bitrate is a good thing... but even with that, there are issues. Ok, full 1920x1080 at 50Mb/s in 4:2:2, that's good. But it's actually slightly more compression than DV, since you need to double your bitrate for 4:2:2 vs. 4:2:0, but they're also going to 1920x1080 vs. 1440x1080. Probably still looks great, but we'll see. The MPEG-2 workflow is certainly more tried and true, but the AVC encoders are getting very good, with as much as twice the coding efficiency of MPEG-2.

[Dave Haynie]

Originally Posted by Thomas Smet

I find it interesting that nobody has compared this to the Panasonic HMC-40 which is one heck of a camera.

Yup... as an HMC40 owner, and as well a Sony HVR-A1 owner, this one is right up my alley. But I remain skeptical. The HMC40 was more of an improvement over the HVR-A1 than I expected. I was more or less happy with the A1's performance, at least in adequate light, and don't have the budget for a 1/3" 3-chipper (though a video DSLR definitely looms in my future).

Originally Posted by Thomas Smet

The big difference of course is one shoots AVCHD at 24mbps and the other mpeg2 at 50mbits.

Ideally, that's a pretty even match. Well encoded AVCHD delivers twice the coding efficiency of MPEG-2; that's largely why it's taking over. This hasn't even been much of a question.. the real question is just how close the hardware running in realtime and 3W or so of power can get to that ideal. Up until this year, there was a pretty big jump in AVC encoding quality, year to year... MPEG-2 was actually a mature technology when HDV started shipping. The Canon adds the additional detail of 4:2:2 encoding at 50Mb/s... that bitrate anyway is largely for the color, a little for the extra resolution, versus 1440x1080 at 25Mb/s.

Originally Posted by Thomas Smet

The Panasonic also has 3 cmos 1/4" chips so it will be interesting to see which is actually better. In recent years I think people have started to realize that single chip designs are not as bad anymore. They do tend to be slightly softer but in terms of color and noise rendition there is not a huge difference.

But there's a big reason for that: higher resolution sensors. Crack open any video DSLR or higher end consumer camcorder and you'll find many, many pixels. Canon's own S21 has an 8.5Mpixel sensor, the DSLRs generally about 18Mpixel these days. So while they still use Dr. Bayer's patten on those single sensors, they don't have to interpolate between pixels. You have a 2Mpixel frame, something more than 8Mpixels in the sensor, and you can simply bucket together your G/R-B/G set (if it's pure Bayer, which some of them aren't these days) as one single pixel... no need to look at the neighbors. Thus, none of that color distortion at color boundaries I used to get with the A1. In truth, it's not terribly obvious except when you're chroma keying... except when it is.

Originally Posted by Thomas Smet

With the way bayer works however I think the HMC40 would still win in terms of raw detail. Bayer needs to much interpolation to have a pixel perfect rendition. I do think the Canon would win in terms of low light however which is the one area that has hurt the HMC40.

Not so sure about the low-light thing. Here's my math. A standard 1/4" sensor has an area of 7.68mm^2, while a 1/3" sensor has an area of 17.3mm^2. If they have the same pixel efficiency, you're getting an aggregate 23.0mm^2 coverage with the 1/4" sensor... 33% more light. Ok, the HMC40 is actually using 1/4.1" sensors, so that's about 7.3mm^2 each, 21.9mm^2, 26.7% more light (the comparison is fair... every pixel site is getting 1/3 of the light entering the camera -- on the four chipper, it's split 3 ways, on the Bayer sensor, 2/3 of the light is just eaten).

They're close enough, of course, that "all else being equal' may not apply. If the Canon sensor is better at using light (more pixel coverage, better microlenses, just plain more recent CMOS tech... noise thresholds seem to shrink a little every year), the Canon could still emerge the winner here.

Of course, this is also why three 1/3" sensors is such a big win versus three 1/4" sensors in low light... that's a big improvement in actual imaging area. By comparison, a Canon 7D or 60D has a sensor (APS-C) that's 419mm^2, while the 5D's (full 35mm) is 864mm^2. Thus the advantage in low light for DSLRs, even though they're using high resolution sensors and Bayer patterns.

[Dave Haynie]

Originally Posted by Ken Hull

The Sony HVR-A1U comes to mind. Sure, it's getting a bit long in tooth, but it got very good reviews.

I've used an HVR-A1 since they came out. While it's a great little camera, it did suffer from a lack of low-light sensitivity and color errors. Both of those are inherent in going to a single HD-sized sensor versus three similar sensors... it's 1/3 the light per image pixel, some softening of the image due to interpolation, and color errors along color boundaries.

Certainly sensors have been rapidly improving, so these new cameras should be better in low light than the A1 was, but I'd be surprised if it was all that dramatic. And the color issues, errors and softening, have been solved in consumer and DSLR cameras by pixel bucketing -- more sensor pixels than image pixel, so you don't interpolate across image pixel boundaries. All of the top consumer models have higher resolution sensors for this: Sony, Canon, JVC, even Sanyo (Panasonic uses 3-chips on their higher end consumer models).

It's hard to believe Canon doesn't know what they're doing here, but this is kind of a shock. 2-3Mpixel single sensor camcorders have been second tier consumer models for the last year or so.