I understand it's probably better suited to digital cinema acquisition than the HMC150, but its AVCHD codec leaves something to be desired and it's unclear how well aliasing and rolling-shutter artifacts will be handled. Codec issues can be circumvented via HDMI or HD-SDI. I believe it also lacks variable frame rates.
Sony's answer to the new Panny...
- Thread starter ATL Media Group
- Start date
ATL Media Group
Veteran
What is the codec lacking? Plenty of films have been made with HDSLRs with codecs that aren't as good as the AVCCAM implementation of this one.
Ben_B
Veteran
They said there's no aliasing, the RED camera has rolling-shutter, and AVCHD has the same if not better performance as XDCAM-EX.
If there were no need for anything more than 8-bit color depth and 4:2:0 subsampling, then 4:4:4 HD-CAM SR, RedRAW, and Cineform are overkill and Panasonic is wasting its time with AVC-Intra. Green/bluescreen work works better with greater color subsampling.
We'll have to see how well Panasonic has tackled aliasing.
It's true that Red cameras have rolling shutters, but their read-reset times -- at least those of the latest generation, Mysterium X -- are among the fastest, purportedly rendering rolling-shutter artifacts at about the same level as you'd see with film.
I'm sure that AVCCAM looks at least as good as XDCAM-EX, but it doesn't look as good as the formats I mentioned above. The AF100 and the 2/3-inch Scarlet will probably come out within a few months of each other. They will be in roughly the same price range. But other that being capable of shallower DOF, which is of limited usefulness, the AF100 holds no real advantage over Scarlet, which was designed from scratch as a digital cinema camera.
I'm not saying the AF100 will be a bad camera. I'm certain it will fill many filmmakers' needs admirably.
Barry_Green
Moderator
It has variable frame rates in 720p mode.
Barry_Green
Moderator
Of course. Now please comment on the price of an HDCAM-SR unit, any RedRaw unit, or any Cineform recorder.
What ATL said was that AVCCAM is much better than the existing DSLR codecs (which it is), and even better than XDCAM-EX (which it is). And lots of folks seem content with those codecs, so why would a step UP be considered inadequate?
Anyway, if you need better than AVCCAM, you can of course use the HD-SDI to output to AVC-Intra, or NanoFlash, or Cineform.
Purportedly. But has anyone actually tested a movie camera against a Red, specifically for "rolling shutter" artifacts? Because I've never, ever, ever seen 35mm movie film "wobble". Point is, film does not work the way a rolling shutter CMOS sensor does. Film acts more like a global shutter, with a little bit of rolling at the beginning and end. You can change the read-reset time from here to purgatory, but as long as the system rolls line by line instead of exposing the whole frame, it's never going to match the imaging of a 35mm cine camera*
*except for the extremely rare case of using something like a 10-degree shutter; in a case like that, yes a rolling shutter CMOS camera would act very much like a 10-degree shutter on a rotating mirror movie camera.
Nor does it cost as much. And, as previously mentioned ad nauseum by XDCAM-EX supporters, you can always use an external recorder for those cases where you need more performance than the onboard recorder can deliver.
Saying "no real advantage" is not appropriate, because a) the 2/3" Scarlet that comes out will be a fixed-lens version, and I think a lot of folks would argue that an interchangeable lens is an advantage over a fixed lens. Secondly, does the Scarlet even offer timecode? Is it actually a video camera, as well as a digi-cine camera? The panasonic is being designed to do both jobs.
There are digi-cine factors where the Scarlet will have advantages, obviously. So maybe your statement should be "there's no real advantage over Scarlet in terms of digital cinema". Even then, the shallow-DOF argument is a compelling advantage. Look at the DSLRs -- they shoot artifacted, aliased, moire-filled video, but they have shallow-DOF. Almost the ONLY advantage they have over any camera is their shallow DOF, and yet that's been enough for them to become massively popular.
Last edited:
Ben_B
Veteran
Low light performance, reduced crop factor (wider lenses), and a short flange distance allowing for the mounting of just about any lens ever made ever. Not sure what the Scarlet will have on the last one but the crop factor issue is going to be huge for those of us wanting to use inexpensive 35mm still camera lenses.
Nektonic
Veteran
I'll believe it when the Scarlet actually starts shipping.
ATL Media Group
Veteran
Exactly!
Thanks for the elaboration. You have options with the new cam. Personally, I like options.. I like being able to record using the latest technology using inexpensive SDHC (and now SDXC) media. Isn't that what all of us shoestring film makers want?
Thank you.
Otis Grapsas
Carbonite Member
A bayer sensor is just the source. It's what you do with it. Each bayer pixel contains a large part of the spectrum of its neighbours and this is used by good algorithms in reconstruction.
The green channel of a bayer sensor can easily achieve maximum performance, even in mediocre demosaicking algorithms, so greenscreen will look great if the recording format is good (RAW to RGB, 4:4:4 or 4:2:2). A good bayer demosaicking algorithm will even do bluescreen with good results.
A 1080p bayer sensor contains 960x540 pixels with red filter and 960x540 pixels with blue filter, but that doesn't mean the output has any relation to 4:2:0. Green pixels also capture a part of the red and blue spectrum and this is useful in reconstructing the image. You will not see 4:2:0 color sampling artifacts on edges in a raw dSLR image developed to TIFF or other high quality formats.
But this advantage is lost in a bayer camera recording 4:2:0 due to 4:2:0 recording. The output will not be better than other 4:2:0 camcorders and DSLRs using the same format.
The green channel of a bayer sensor can easily achieve maximum performance, even in mediocre demosaicking algorithms, so greenscreen will look great if the recording format is good (RAW to RGB, 4:4:4 or 4:2:2). A good bayer demosaicking algorithm will even do bluescreen with good results.
A 1080p bayer sensor contains 960x540 pixels with red filter and 960x540 pixels with blue filter, but that doesn't mean the output has any relation to 4:2:0. Green pixels also capture a part of the red and blue spectrum and this is useful in reconstructing the image. You will not see 4:2:0 color sampling artifacts on edges in a raw dSLR image developed to TIFF or other high quality formats.
But this advantage is lost in a bayer camera recording 4:2:0 due to 4:2:0 recording. The output will not be better than other 4:2:0 camcorders and DSLRs using the same format.
combatentropy
Veteran
Hmm, that's not how I understood Graeme's postion. Pivotal to Graeme's favor of Bayer is counting all the pixels --- all of them.
If you have three chips of 1920 x 1080 each, that's not 2 million pixels, it's 6 million (1920 x 1080 x 3). With 6 million pixels, you could make a 3300 x 1800 Bayer sensor (laying them out in a 16:9 ratio). Then, it should have a final resolution (multiplying by 78%) of 2500 x 1400.
Or, you could shoot for the same final resolution and save pixels. A 2500 x 1400 Bayer sensor would yield a 1920 x 1080 final resolution, using about half as many pixels.
Last edited:
Otis Grapsas
Carbonite Member
Exactly. Film gets lots of global shuttering time in the normal shutter angle settings. There are other differences also. You can certainly whip pan a film camera and, yes, they do that sort of thing in Hollywood and many other things a rolling shutter cannot handle.
Price wasn't mentioned as a variable. The question was, what's wrong with AVCCAM with reference to its use in digital cinema, and I answered. And in an earlier post I mentioned using HDMI or HD-SDI out. But in terms of being ideal as a codec for a digital-cinema camera, which was the original context, it's not ideal.
As to people being content with vDSLR codecs, I think it's more accurate to say that because these large-sensor HD video capture devices are relatively cheap, people may not like but accept their limitations, particularly since they understand these cameras are still pretty much first-generation technology. Would they want them to be better -- no rolling-shutter Jell-o, no aliasing, etc.? Probably.
Red has slated an interchangeable-lens 2/3-inch Scarlet, as well.
Yes.
It's not really a video camera. But it's supposed to be able to output scaled (not line-skipped or pixel-binned) 1080p and 720p RGB in addition to RedCodeRAW.
I'm sorry if that wasn't clear, despite the fact that my original comment was in reference to the AF100's use for digital cinema: "I understand it's probably better suited to digital cinema acquisition than the HMC150, but its AVCHD codec leaves something to be desired." This was in response to Ben_B's statement, "This alone makes the HMC-150 make more sense for ENG. . . . The AF100 is best suited to digital cinema or other applications where a cinema look is desired."
Last edited:
Stephen Mick
Totally Usable
Hey, so how about that Sony camera, eh? Wasn't that what this thread was originally about? 
Yup. Sorry.Hey, so how about that Sony camera, eh? Wasn't that what this thread was originally about?
ATL Media Group
Veteran
You hardly ever hear of complains about the vDSLR codecs.. Actually the only one in recent memory was the GH1 and that was because it was low bitrate and 24P over 60i... This camera fixes those issues so it's definitely an upgrade.
Also:
Then you say: "HD video capture devices are relatively cheap"
So you say price isn't mentioned then you mention cost....
I see.
I hear it all the time -- 7D, 5D, 1D, GH1.
I mentioned low cost as one reason people accept the limitations of vDSLRs. That was an issue of motivation. When I was asked what was wrong with the AVCCAM codec relative to its application to digital cinema, I assumed that question involved only its technical characteristics, not whether it was a reasonable compromise given the cost. I said the codec wasn't ideal. That's all. But I can understand why people would accept its limitations rather than spend tens or hundreds of thousands of dollars on high-end equipment. After all, before people were using vDSLRs, they used consumer AVCHD camcorders, and before that, HDV, and before that, DV, etc. But why people do it doesn't have any bearing on the technical limitations of a format, except that they are willing to work around -- and sometimes even embrace -- those limitations.
Last edited:
combatentropy
Veteran
This reminds me of the backwardness of prices for codecs:
AVCHD is cheaper than AVC-Intra. That means
MPEG-4 at 24 Mbps is cheaper than MPEG-4 at 100 Mbps.
Do you get it? Let me put it another way:
Uncompressed high definition video is over 1 Gbps.
To compress it to 24 Mbps is to make it 1/50 its original size.
To compress it to 100 Mbps is to make it 1/10 its original size.
Which do you think takes more technology?
Last edited:
Barry_Green
Moderator
The best camera engineers at the top camera companies in the world, including Sony, Panasonic, Ikegami, Hitachi, Thomson, Canon, and JVC, have all put three-chip systems in their broadcast cameras. Why wouldn't they just go with a much-cheaper single-chip Bayer?
Bayer was relegated to single-chip cheap handicams, up until the recent craze for film-lens systems. Nobody -- NOBODY -- put Bayer in their top-end cameras. I stand by my assertion that a 3-chip system results in sharper images with better color reproduction than the same chip implemented as a single-chip Bayer system. And the only reason someone would useo with a single-chip system using a Bayer pattern is if they need to use film lenses, which don't allow for the prism.
Look around -- for top-end cameras that don't have to take film lenses, do any of them (any?) use a single-chip Bayer?
A single-chip Bayer will never outperform a three-chip system, of comparable chips. An oversized single-chip Bayer could outperform three smaller chips, of course. But like for like, the best performance comes from splitting the light with a prism and giving each primary color its own chip.
Okay, put another way -- if there were no restrictions placed on the designers, Bayer would most definitely not be the upper echelon of quality that engineers reached for. Bayer is a compromise that has recently become more acceptable specifically because of the need to accomplish the whole imaging block in a single chip. If you have to go with a single chip, then Bayer is probably the best single-chip solution. But if you don't HAVE to go with a single chip, you can achieve better results with a three-chip system.
Last edited: