Canon HF20 - any thoughts?

[Bob]

You can also get cineform neoscene at the video guys for $30 cheaper: http://www.videoguys.com/Item/CineForm+Neo+Scene+PC/54E4543435F454E4.aspx

[peggig]

How does the Legarith codec compare to the Cineform codec?

Does the Cineform codec come with the Vegas "Movie Studio" versions, or just with Vegas Pro? Also, how does Vegas Movie Studio Platinum Pro compare to PE7? (I wish I hadn't just bought PE7...)

How much bigger is a Cineform AVI video, converted from AVCHD, than an HDV video of the same duration?

Thanks!

[Paul LS]

I dont think Cineform comes with Vegas Studio... infact Sony have dropped the codec from their latest Vegas Pro release (which got me mad)... so for both now you would need to buy Cineform NeoScene. Regarding the size of Cineform files... 1920x1080 Cineform files will be about 60GB per hour, 1440x1080 about 40GB per hour. HDV MPEG2 which is only 1440x1080, because of the standard, is 13GB per hour.

I haven't used the Lagarith codec. but I hear folks use it with Vegas.. especially the 64 bit version as it was the only indermediate codec supported under 64 bits... until Cineform introduce 64 bit support.

[peggig]

Thanks, Paul. Wow. Those are some huge file sizes. I don't understand why they have to be so much larger than HDV files. I realize HDV files are compressed, but apparently not so compressed they can't be edited in real time. Now I'm undecided again... I just bought a terabyte hard drive a few months ago, and it's already half full. -- And I've just been shooting in standard definition.

Does HDV take longer to render and export than standard definition? I've been focusing on the differences between HDV and AVCHD, and I haven't considered how much longer HDV might take to edit than what I'm used to.

Has anyone on this forum had experience with the Legarith codec (32 bit version)?

[Paul LS]

Even DV-AVI is compressed... approx 5:1 compared to uncompressed standard definition AVI. If you consider the 1920x1080 frame size is ~6x bigger than DV-AVI frame size you can see why the files are so large.

Rendering HDV is fairly quick... I haven't worked with standard definition DV-AVI for a few years so I can not compare speeds.

[peggig]

Thanks, Paul. But now I have a couple more questions. (Sorry about that.) How is 1920x1080 6x bigger than 1440x1080. Mathematically, it should be about 1/3 bigger.

Also, how is frame resolution related to sensor size and pixel count? For example, the HV40 has a significantly larger sensor, and fewer pixels, than the HF20, but the frame resolutions on the HV40 is 1440x1080, while the HF20 has 1920x1080 resolution. What is it that determines the frame resolution? I understand it's a different codec, but there's only so much information being captured by the sensor. The codec can't add more than is there to begin with.

Thanks for your infinite patience.

[Bob]

Paul said it was 6x bigger than dv-avi which is 720x480 for NTSC. (1920*1080)/(720*480) equals 6.

The sensor determines the maximum resolution by how many photosensitive regions were created on it. The size of the regions can be larger or smaller from one sensor to the other with the result that you can have small sensors that have as many or more regions as a larger sensor. Larger sensors generally have larger sized regions which generally leads to less noise in the image. Pixels are generated by a demosiacing process that combines the photosensitve regions. Often, the sensor is larger than the video frame that is captured. The canon HV series camcorders have a sensor that is 1920x1080 (using square pixels -- pixel aspect ratio 1.0) but records 1440x1080 (with rectangular pixels -- pixel aspect ratio 1.33) using the HDV codec when going to tape. You can capture the 1920x1080 frame if you bypass the tape and capture live.

[peggig]

Paul said it was 6x bigger than dv-avi which is 720x480 for NTSC. (1920*1080)/(720*480) equals 6.

My apologies. I thought he was comparing it to HDV, which I was thinking tops out at 1440x1080 but, as you point out, that's a limitation of the tape medium, not the codec.

What's the difference between gross pixel count and effective pixel count? For example, the HF10 has 3.31 megapixels (gross), 2.07 megapixels (effective - video), and 2.76 megapixesl (effective - still). The HF S100, claims to have 8.59 megapixels (gross), 6.01 megapixels (effective - video), and 8.0 megapixels (effective - still).

[jackfalbey]

To put it simply:

When talking about gross pixel count in consumer HD camcorder sensors, generally a higher pixel count means smaller receptor sites, which translates to better sharpness and detail at the expense of low-light sensitivity. Lower pixel count generally means larger receptor sites, therefore it has a softer picture but better low-light sensitivity. Of course there are always exceptions to this rule.

Peggig, before you get bogged down in the tedious task of number-crunching in comparing various camcorder specs, what is your intended use for the finished video? Family & home movies, indie films, for-hire work, TV broadcast, etc. This will help us advise you better on which camera and format best suits your specific needs.

[peggig]

I produce a monthly half-hour program for our local public access TV station, which I then break out into ten-minute segments and post on YouTube. I generally shoot about 3.5 hours of tape for each program, and edit it down to a half hour. Most of the programs consist of interviews that I do on-site at various locations. Whenever possible, I tape location footage to overlay on part of of the interview. I do about six interviews per program, with a seventh person doing the introduction and summary. I use fades between interviews and lots of cross dissolves during the interviews and in the video montages with the interview as a voice-over.

I frequently have to "tighten up" interviews by cutting out long pauses between words because of the time constraints (and the fact that a lot of the people I interview talk v-e-r-y s-l-o-w-l-y...). In the cases where I don't have visual footage to overlay on the interview, I try to match frames as closely as possible before and after the cut, and use a 2-frame dissolve to soften the transition. (I know that sounds ridiculous but it's often necessary to cut out just a few seconds to squeeze two interviews into the ten-minute limit for YouTube, and I don't want to lose any of the actual content.)

I spend a whole lot more time editing than I do shooting, so significantly increasing the editing time wouldn't be practical for me. On the other hand, If I were to shoot in 1920x1080 AVCHD, and convert it to a format that requires 60 GB of storage per hour of video footage, I'd be filling up a terabyte hard disk every four months. (I archive all the footage I shoot.)

It's starting to look like HDV is really the only practical option for me, though I would prefer a flash memory based camera for a number of reasons. I don't want to settle for a tape-based camera until I exhaust every possible option that might make AVCHD work for me.

[Bob]

What's the difference between gross pixel count and effective pixel count? For example, the HF10 has 3.31 megapixels (gross), 2.07 megapixels (effective - video), and 2.76 megapixesl (effective - still). The HF S100, claims to have 8.59 megapixels (gross), 6.01 megapixels (effective - video), and 8.0 megapixels (effective - still).

Gross pixel count is the total number of pixels on the sensor. That's not a very meaningful measure unless you also consider the sensor size and technology used to build the sensor. Jack's example is correct for two sensors of the same size and manufacturing technology. In general, if two sensors have the same gross pixel count, the larger physical sensor will have less noise and greater light sensitivity.

The effective pixel count is the number of pixels that are used to capture or record an image of a given aspect ratio and size. And, that will differ according to whether you are capturing video or stills and the image aspect ratio and size.

Take the HF10 with 3.31 gross megapixels. 1920x1080 equals 2073600 pixels or 2.07 megapixels. The video frame size is fixed at that by the spec, so that's the effective pixels for shooting HD video. Shooting stills, you can use a larger portion of the sensor but you are limited by the aspect ratio of the still image. The 2.76 megapixel effective size reflects that.

With the HF S100, the sensor has a gross pixel count of 8.59 megapixels which makes for a much higher resolution still photo. For stills, you can get an 8.0 megapixel image. That's much larger than needed for HD video which is limited to 2.07 megapixels. If the camcorder only recorded the central 1920x1080 pixels, you would not be utilizing the full angle of view. If you were to take a snapshot while recording video, the snapshot would look normal, the video would look like you had a telephoto lens. (Incidently, the Canon has a digital teleconvertor mode that takes the center 1920x1080 pixels and does exactly that.) Instead, they combine the sensor data in a different mathematical way taking 6.01 megapixels of the sensor data and combining them to create a down sampled 2.07 megapixel HD 1920x1080 image.

[peggig]

Thank you, Bob! I wondered how the HF S100 managed to use 6 megapixels for HD, when the spec is for 2.0736. I also wonder if the downsampling might actually degrade image quality. I'm not even considering the HF S* series because of the high pixel count (more noise), and poor low light performance.

Interestingly, the HV30 has a larger sensor and fewer pixels than the HF10, but shows more noise in low light. (Going by the tests on camcorderinfo.) They both use the same processor, so I assume the difference must be in the codec, but I could of course be mistaken about that.

I guess I'll probably get the HV40, unless someone can convince me that a 2.4GHz quad proc Vista 32 machine with 4 GB of RAM actually can handle AVCHD in PE7 without exponentially increasing my editing time. (I'm still hoping that someone will be able to convince me, but I'm starting to lose hope...)

[Chuck Engels]

without exponentially increasing my editing time.

That's the problem, even HDV will probably double your editing time over SD. But that's not as bad as AVCHD

[Bob D]

I've been reading this thread with some interest. The number crunching sometimes is beyond me, but I do find the conclusions useful.

One suggestion to peggig, although I have never done this,is to rent or borrow the camcorder you are interested in from somewhere and try a sample recording and editing to see if it will work with your computer and workflow.

Bob D

[jackfalbey]

Go to hv20.com and download sample video from each camera you're considering. Then try them with your workflow to see which is better for you.

My personal opinion is that the HV30/40 will be best for your needs: TV & YouTube. You said that most of your editing is cuts and transitions, so you'll be fine with HDV and a 4:2:0 color space. That's what I would buy if I were in your situation (assuming that a real 3-chip prosumer cam like the Sony FX1000/Z5U or the Panasonic HMC150 are out of your price range). The only real advantage you'll get from AVCHD is longer record times between SDHC card swaps.

As far as speed of editing, you would have to use an intermediate like Cineform with AVCHD to have a reasonable experience with your PC.