New HF11 review and video clip sample
 

Check it out at:

http://www.google.com/translate?u=ht...F8&sl=ja&tl=en

Sample of 24 Mbit/sec file in .mts format direct from camcorder also provided at this URL as well as slower 17 Mbit/sec encoding for comparison to present HF10/100 performance.

I think it answers the question I posed earlier. 24Mbps AVCHD will not playback on BD players other than the red-laser DVD unit Canon sells.

Also of note, accessory-wise: "This new accessory also appeared in the video light VL-5. Release September, tentative price 9,000 yen."

I authored a couple AVCHD disks with the 24 MB/sec HF11 content and was delighted to find that they played beautifully on the Sony PS3 as well as on my PC player software. The improvement in image quality was visible but not at all dramatic. In fact, slight sharpening of my 17 Mbit/sec clips with Vegas or several of the other AVCHD editing and authoring programs I have here produce the same edge enhanced detail, making HF100 video look indistinguishable. Not surprisingly, the HF100 video when slightly sharpened, has about the same bitrate as the new HF11 when played, roughly 21 to 24 MBit/sec VBR.

Here it is, powered by the accessory shoe:
http://cweb.canon.jp/ivis/lineup/hiv.../light-vl5.jpg

There will also be a higher capacity battery option, BP-827, at 2670mAh:
http://cweb.canon.jp/ivis/lineup/hiv...ory/bp-827.jpg

Having just purchased the current largest BP-819 battery, the new BP-827 sounds like another huge leap, essentially 50% more juice (2670 versus 1790 mA-hrs). Should run an HF100 for over 4 hours but I assume the newer HF11 will draw higher current with resulting time reduction.

This is kind of what I thought might happen. We've probably reached the point of diminishing returns with bitrates in the current crop of AVCHD units. If you take the SR12 and play with the highest bitrate vs. the next lowest bitrate, you won't see as dramatic change in PQ as the bitrates might suggest.

We'd discern a much higher increase in PQ with larger sensors IMO.

I agree that sensor improvements could be the "next frontier" for additional IQ/PQ in the 24 MB/sec AVCHD camcorders.

I base this statement on the very easy experiments I performed using 8 megapixel super crisp still photographs taken with the best of my Nikon cameras and lenses, then rendered in all of the current AVCHD editing/authoring NLEs as slide shows. It is very easy to then take a crop of the original frame from the original 8 megapixel still photo and directly compare it to the rendered, cropped, captured still frame in the AVCHD disk, and shows just how well the renderer performs at various bitrates.

No doubt there is still further fine detail to be seen in future AVCHD camcorders, presumably when their sensors and optics begin to better approximate those in consumer still cameras. The visual improvements will, however, not likely be especially dramatic based on my own tests, since the ultimate 1080p frame is, after all, 'only' a 2 megapixel image, and present camcorders / sensors/ optics are already resolving detail to at least this level in today's better AVCHD camcorders. My 8 MP still photo results show visible gains in detail, but nothing I would call dramatic.

Having played with this AVCHD/mpeg4/h.264 quite a bit now, I would personally hope that the engineering emphasis will next come in the area of motion artifacts. Some combination of sensor lag, A/D conversion errors, h.264 encoder errors, and / or overwhelmed processing demands placed on the camcorder's encoder still make AVCHD look far too 'busy' during pans, especially horizontal pans. Such artifacts do not appear to be nearly as noticeable in commercial releases using AVC / h.264 codecs, so I can comfortably put the blame on the present prosumer AVCHD camcorders, which seem to be on the verge of not handling motion very well. This is something I would hope and expect to improve dramatically in the next cycle or 2 of product refinements.

Just wondering what y'all think of those sample clips... is there a better way to visually test the differences in bit rate? How would you have done this?

I think the clip is much too short to make any definitive conlcusions Chris. For some reason they always seem to shoot those clips on a cloudy day. A bright sunny day might have hinted at a difference in dynamic range as well as color quality. It might also have been nice to see a horizontal pan as well as a vertical one. As it shows now, the difference is a bit subtle.

Thanks for the feedback, Ken -- how long do you think a proper sample clip should be... one that's distributed over the web, but still in its native file format? Guess I'm asking what is the right balance between duration and file size, if compression is not to be altered.

Chris, I think the clip length & type that was shot and posted here with the HF10 when it first came out, was much better. I don't recall who shot it, but it began as a wide shot over some water showing a bit of a cityscape and then zoomed in to a construction site. I seem to recall the clip being about 30 seconds in length. Not sure if that helps at all.

Yes that's a great help, thanks Ken. That shooter was Austin Meyers, and the location was Austin, Texas. I remember that clip very well.

For some certain points I agree with you. But ,,commercial releases" are shooting mainly/only in controled environments; they know not to make fast panning, use insuficient light and so on. That's why I can't compare my SR11 even with the lowest quality 720p trailer (even in quicktime)... the quality difference is dramatic.
I was very pleased to shoot many close-up to sea in motion, water flowing from a pipe, etc, and found not a single compression artifact from motion!
Of course, my main grip with SR11 is the sharpness. I don't know if it's noise reduction to blame or the focus, or my HDTV...

How does the sharpness of the HF11/HG11 in MXP mode compare to an HV20?

My Canon HF100 may not be quite as good in this regard as your SR11 but I do see some motion artifacts, occasionally, and most often in horizontal pans. It is not dramatic, but visibly more noticeable on my camcorder clips than I see in commercial encodings.

The fact that commercial encodings are not encoded in a hurry while the video is being captured helps a lot. They can take as many hours as they need to get the software encodeer to make the very best looking AVC /h.264. The camcorder designer faces a different problem.....the need to do real-time encoding, running off a small battery and small computer. This is why I think this area is likely to get some attention in future designs.

I own both the HV20 and the HF100, and they are very similar in sharpness. The new HF11 may be a tad sharper than the HV20 but we are taling about pretty small differences.

Can't say I see any issue with sharpness on my 60" 1080p Kuro plasma. So perhaps it's your HDTV if you sense that focus is not an issue. If you're accustomed to lots of edge enhancement in camcorders, you won't find it in the SR11/12. That could also account for you feeling it's lacking 'sharpness'.

No Ken, I don't really think it's my TV fault (Philips 42PFL9632), because like Larry I compare with HD trailers and it's a long way to clarity/sharpness. I will look to HF11 and report, for me HF100 appears to have the same lack of sharpness as SR11.
Or maybe I want too much from a prosumer camcorder.

that would actually be perfect - finally an undeniable replacement for tapes.

It's not your monitor nor your eyes. You CANNOT get HDCAM visual detail from ANY (pro-) consumer camcorder. Which is why I did not keep the V1 Sony sent me. As one zoomed wide -- fine detail just went away!

Same with the HD7 and SR12. The only solution is to very slightly increase mid-frequency response -- which is NOT inherently adding edge enhancement -- by using the Sharpness control. (See BBC tests of Sony camcorders.) Not possible with the Sony.

All consumer camcorders that either UNDER sample (3-chip) or use a Bayer filter look soft compared to the HDCAM video you see on HDTV. The Bayer filter cuts rez in half in each direction, so to get 2MP you need an 8MP chip.

The EX1 captures (three 2MP chips) and records 1000 TVLph while CineAlta HDCAM captures and records over 1100 TVLph. (A must better lens offers higher MTF so more rez with less chip pixels.)

All consumer camcorders -- except the Canon pro-HDV units -- capture and record around 650-TVLph. This is not opinion, it is measured. And, as you have found is very visible.

The SR12 will look very slightly softer -- and measures confirm this -- than the Canon because it uses an 8-cell Bayer filter rather than the traditional 4-cell Bayer filter. (The more pixels in a cell, the less precisely defined will be each interpolated luma.)

Moreover, Sony only supports 8x8 macroblocks (BASE profile) while the Canon uses the more powerful HIGH profile that can switch to 4x4 macroblocks on fine details. (BD uses HIGH.)

There are only 2 solutions: buy an EX1 or wait for the 24Mbps Canon or Pana. Why?

The lower the recorded data-rate -- the more fine detail must be filtered-out to avoid wasting compression on noise. When the data-increases -- the less noise (and the less detail) that must be filtered-out.

PS1: Sanyo is using a 3D noise filter that is equivilent to increasing the data-rate by 10-15%. Very clever, since 3D filters nicely eliminate RANDOM noise but not detail. Which points-out that the PRE-FILTERING circuit is as critical as the encoder.

PS2: the cheaper the camera, the lower its MTF. So that is another reason why cheap offers less. And, why the Pana will offer better rez than the Canon. It should, as it has an MSRP of $4000!

Thank you very much Steve, I really enjoy your critical and technical approach and I can't agree more. However, the more I look at my SR11 clips I think that AF is one problem, even in good light; I will try to shoot at widest settings with manual focus (set at infinite)/AF and see if there is an improvement (for now I don't trust very much AF metering, especially in HD).
To be ontopic, I look at the HF11clip ezsm021.mts. It definitely looks sharper than my SR11, but vertical panning was very bad. I change refresh rate on my HTPC, 24/25/30/50/60Hz to minimize judder, but even without judder I don't like the loss in detail.
I don't have a PS3 (as a hardware player), could someone tell me at what refresh is this file?

I shot many hours with my sample V1 and often thought it must be a focus problem. But, eventually I decide it only happened at WIDE of shots with tiny details like leaves.

I'm not dissing the consumer camcorders -- I think for $1000 they offer as good an image as the V1 which costs about $4000. I'm only explaining WHY we don't get video that looks like Discovery. And, why IF YOU HAVE THE MONEY the EX1 will give you these images. But, the HD7 and SR12 are far more comefortable to shoot with. At my age I would not try to balance the off-balance EX1.

Mircea, if I'm correct, your Phillips is a 768p (usually referred to as 720p) plasma. You'll see more sharpness/detail (the two are different) with a 1080p plasma and either the SR11/12 or the Canon HF10/100. But you are probably correct in that if you are expecting to see the level of detail you see in pro cameras, you won't.

However, to be honest, I don't think you'll see the jump in sharpness you are hoping for with the HF11. It might be a bit better, but it still won't approach the level of detail a pro camera will.

Ken, my Philips is a 42" 1080p LCD. Of course it's not near Kuro plasma as black level, but as processing is in the upper league (except from some damn bugs). I can correct with sharpness in TV but edges doesn't look very good (especially looking at grass). Heck, I must buy a PS3 to be sure it's not my HTPC...

I already post my impression about the HF11 clip, could you please comment about vertical scroll? For me it's not good at all.

Mircea, on my computer the vertical pan is very smooth.