Panasonic reveals more details of AG-AF100

No -- because they announced it *did* have variable frame rates. :thumbsup:

I should have been more clear. Im talking more than just 1080:59.94i/50i/29.97p/25p/23.98p or 720:29.94p/50p/29.97p/25p/23.98p. Or did they announce more and I missed it. Either way I guess the hvx has me spoiled. :laugh: I am however excited about the new "29.94" though hehehe :-Dum(DBG):
 
They announced it and you missed it. As I understand it it's going to have the same variable frame rate capability as the HVX/HPX170/HPX370/HPX500.
 
Honestly, if there's one thing that really bugs me about the design, it's how uncomfortable that handle looks.

edit: It also looks like the VF doesn't tilt?
Actually I disagree. There is going to be various size lenses used on this camera. It makes sense for the handle to be more forward biased and big enough to slide your hand to the center of gravity. I have a 35mm adapter on HVX and with those longer lenses, I sure wish I had a handle like this. Also, I guess most users will be using an off board monitor, hence the lack of tilting VF.

What I meant about the handle was that it looks uncomfortable (too thick?), not that it was too far forward.
 
Heh. There's an old saying that goes "If you have to ask the price, you can't afford it."

Ultra Primes are about $900 for a set of six.

To rent.

For ONE DAY. :)

I didn't find a buy price for Ultra Primes either, I could only find rental prices. Consider, though, that buying Compact Primes (their super-budget line of movie lenses) will run you about $5,000 per lens. And that the Cooke Panchro, their "budget" line, is about $7400 per lens.

The closest I could find is the Cooke 5/i 32mm, it costs $20,700. I would have to believe the Ultra Prime is going to be somewhere in that ballpark.



So... yeah, you don't buy Ultra Primes. Rental houses buy Ultra Primes, and you rent them from the rental house when you need 'em.

The lenses people are going to buy for the AF100 are going to be the m4/3 lenses (like the 20mm f/1.7 or the 14-140 f/4); those are more in the range of $300 to $1000.

Or, if you want inexpensive cinema-quality lenses, the Zeiss ZFs range from about $600 to $1500 each, and they're the exact same glass as the Compact Primes.

Ultra Primes

http://www.creativevideo.co.uk/public/show_category.php?categoryID=3&subcategoryID=336

Master Primes

http://www.creativevideo.co.uk/public/show_category.php?categoryID=3&subcategoryID=347

Dave
 
They announced it and you missed it. As I understand it it's going to have the same variable frame rate capability as the HVX/HPX170/HPX370/HPX500.
Wow, I missed that too!

"Multi HD Format and Multi Frame Rate" Those terms are often both used to describe multi HD formats, so it is easy to miss. I see now they describe slow motion and fast motion in 720p. I assume in this case it will mean it can record multiple frame rates into a 720p60, 720p30 , and/or 720p24 file (instead of making you do things in post). This can easily be done using the same technique the hmc40 uses for interval recording... which hopefully also gets carried over.

Using that same technique, I don't see why it couldn't also work for 1080p modes and allow for slower rates than 12fps. You could also theoretically jam 1080p60 into a 1080p30 file at twice the speed, or even > than 60fps (assuming the processing speed and read times were capable).

Just curious, does audio get dropped in other cameras when using non-native frame rates?
 
Okay I have a dumb question, lol
The af100 will have a SINGLE CMOS 4/3 chip.
The HPX 370 has 3mos chips (which sounds like 3 cmos chips).
So would I be right in thinking that although this new camera will give us the coveted shallow depth of field, that it's resolution will not be anywhere as good as the HPX 370 (Yeah I know it's half the price so save your typing fingers) I'm just trying to get a handle on what to expect.
What about the RED ONE camera? Is it 3 cmos chips or just one?

Remember when JVC first came out with that little 1 chip HDV camera? And nobody wanted it right? Shortly afterwards came the 3 chip cameras.
Are we gonna see the same thing happen here ya think?

Thanks for the lowdown
 
The HPX370 has three 1/3" chips, with 1920x1080 pixels on each of them. That makes it basically a 6.6-megapixel imaging block. It renders images at full raster resolution, and records them at 4:2:2.

The AF100, on the other hand, has a single 14-megapixel chip, and records to 4:2:0.

Will there be differences? Probably. Will the HPX370 be sharper? That's the big question. There's more than enough pixels on the AF100 to outdo it, and it is a much larger chip, but the big question is whether it'll be able to use all that extra power. My guess is that they will be of equivalent sharpness, the 370 will have better color recording, and the major advantage to the AF100 will be the shallow DOF.

The Red One is a single CMOS chip, about the same pixel count as the AF100, and just slightly larger than the AF100.

Remember when JVC first came out with that little 1 chip HDV camera? And nobody wanted it right? Shortly afterwards came the 3 chip cameras.
Are we gonna see the same thing happen here ya think?
Broadcast wants, prefers, and has always gone for three chips. Single-chips are mainly used when you want either a) large chip size, or b) low cost, or c) interchangeable film lenses.

Three chips gives better color accuracy and is the broadcast standard, but it involves a prism and as such, the lenses should be designed specifically for that three-chip system.

Film-style or film-emulation cameras use a single chip, which can then be a "replacement" for where the film would be. You can use film-style lenses, whether cinema or SLR, on a single-chip camera.
 
Ahh, that makes perfect sense. THanks for the explanation. Better color is what I would expect a 3 chip would do. Sounds like the key here with the single Cmos chip is the better interchangeable lense abilities it will posses then. IF this AF100 is as sharp as the Hpx370 I would be quite satisfied. But it also sounds like with this camera, if they wanted to, they could offer it with at least 2k resolution..right? (I know they wont at this price, but the ability seems to be there).
 
A 3 chip 1080p camera is not a 6 megapixel imager. It produces a 2.2 megapixel RGB color image and most cameras sub sample the red and blue channels at 1/2 the bandwidth of the green channel.
A single chip Bayer pattern sensor does exactly the same thing, it produces an interpolated pixel for pixel RGB image with red and blue channels sampled at half the bandwidth of the green channel. This also corresponds very closely to the way human vision sees color.
3-chip cameras wind up with effective resolution free of significant aliasing at about 85% of the total RGB pixel resolution at best, around 900 lines vertical for 1080p.
A single chip Bayer pattern sensor will wind up with effective alias free resolution of about 70% of total pixel count. But this is still significantly oversampled for 1080p with a 12 megapixel RGB image vs a 2.2 megapixel RGB image from a 3-chip camera. So the way the image is scaled and compressed from 12 megapixels to 2.2 megapixel HD video impacts the final image quality much more than whether it is a 3-chip or single chip camera.
There is no color compromise with a single Bayer pattern sensor over a 3-chip sensor system. The still image performance of DSLR's exceeds the color gamut limitations of encoded video by a significant margin.
 
Sounds like the key here with the single Cmos chip is the better interchangeable lense abilities it will posses then.
Mainly, yes -- but it's also a much bigger chip, making it more suitable for creating the shallow-depth-of-field look.

But it also sounds like with this camera, if they wanted to, they could offer it with at least 2k resolution..right? (I know they wont at this price, but the ability seems to be there).
It may be a stretch to get 1080p out of it. 2K is just barely more than 1080p (2048 vs. 1920). But no, you're not going to get "k" resolutions out of video cameras. The "k" notation is used in non-video digital cinema cameras.
 
A 3 chip 1080p camera is not a 6 megapixel imager.
It is a 6.6-megapixel imaging block. Three individual chips that each respond to a portion of the color spectrum and are read together to create an image.

It produces a 2.2 megapixel RGB color image
Out of 6.6 million pixels, yes, and can do so at 4:4:4 if so designed.

and most cameras sub sample the red and blue channels at 1/2 the bandwidth of the green channel.
Irrelevant to the chip block. The point is that it is a 6.6-million pixel imager that has 2.2 million red, 2.2 million blue, and 2.2 million green pixels. That class of imaging block could be paired with dual-link HD-SDI to output a 1080p 4:4:4 image.

As used in traditional video cameras, some of that color information does indeed get discarded in the name of compression efficiency and bandwidth-saving; 4:2:2 would discard half of the color info, 4:2:0 would discard 75% of it.

A single chip Bayer pattern sensor does exactly the same thing, it produces an interpolated pixel for pixel RGB image with red and blue channels sampled at half the bandwidth of the green channel.
A single chip Bayer pattern sensor does most definitely *not* do "exactly" the same thing. A bayer pattern sensor is a monochrome sensor, with none of its pixels co-sited and is inherently incapable of ever producing a 4:4:4 image, nor of producing an image with any color accuracy greater than 4:2:0 at native resolution. A 3-chip block could easily create 4:4:4.

But this is still significantly oversampled for 1080p with a 12 megapixel RGB image vs a 2.2 megapixel RGB image from a 3-chip camera. So the way the image is scaled and compressed from 12 megapixels to 2.2 megapixel HD video impacts the final image quality much more than whether it is a 3-chip or single chip camera.
With this, I agree entirely.

There is no color compromise with a single Bayer pattern sensor over a 3-chip sensor system.
With this, I disagree. You can argue that "within the context of a 4:2:0 recording system, there is no compromise" and I'd agree with that. But to empirically compare the two imager-block technologies, there is absolutely a compromise in color sampling accuracy from a single-chip Bayer system, versus a three-chip system. And it extends beyond the actual image-cositing and 4:4:4 potential; a Bayer system inherently allocates 25% of its pixels to blue and 25% to red and 50% to green. However, in the acknowledged and accepted video encoding systems, that's not how the percentages break down. In professional video/imaging, about 60% of the color of each pixel comes from green, about 29% from red, and about 11% from blue. Bayer can't match those percentages accurately, but 3 chips can.

The still image performance of DSLR's exceeds the color gamut limitations of encoded video by a significant margin.
Certainly, a 14+ megapixel sensor, downconverted to 1080p, will be substantially more accurate and sharper than a native 1080p image block. But try a 1920x1080 Bayer chip, and it'll be a whole different story. Bayer works well only in a scenario where it has substantial oversampling capability, to make up for the resolution loss due to the demosaic process, and the more oversampling the better for creating more accurate color sampling.

The three-chip system typically employs a slight pixel offset to increase resolution, which causes a loss in the ability to render true 4:4:4, which was an accepted engineering tradeoff in the context of maximum 4:2:2 recording systems. But it doesn't *have* to do that, it can be kept in perfect alignment and achieve 4:4:4.

Every technology involves compromise. For digital cinema cameras with interchangeable lenses, a single-chip Bayer is unquestionably the best approach (unless one wants to argue the merits of the RGB Stripe system). I would certainly rather have the AF100 with a single large chip rather than three small chips.

But for ultimate color accuracy and quality, every camcorder manufacturer manufactures every broadcast camcorder with three chips, for very good reasons. Bayer is not superior in imaging to a three-chip system, for those purposes that broadcasters value. But the compromises that three-chip systems demand, make them impractical choices for a film-emulating interchangeable-lens digital cinema camera. Bayer's advantages make it a much better choice, and while it may not offer some of the capabilities of the 3-chip system, its advantages far outweigh its disadvantages for this particular purpose.
 
Last edited:
Rumors say Panasonic scheduled a press announcement on July 28th to unveil a "revolutionary" camcorder, suggesting that final specs and images will be revealed.

who knows, but it would suit as a quick reply to Sony's NEX camcorder
 
It may be a stretch to get 1080p out of it. 2K is just barely more than 1080p (2048 vs. 1920). But no, you're not going to get "k" resolutions out of video cameras. The "k" notation is used in non-video digital cinema cameras.

Okay, wrong terminology I guess. But RED ONE boasts 4k resolution. It's still a video camera (as opposed to film). I'm just wondering with 14mp chip why Panasonic cannot get higher resolution that 1080. But looks like you did explain that above. So RED ONE must be alot more megapixels than 14 right?
 
Back
Top