Something new from Canon on Nov. 3rd...

[Robert Sanders]

In my opinion the Sony F3 is the target. When I look at Sony's camera I can definitely see Canon coming up with a product that competes with it. The downside, however, is that you're never going to see a sub-$10k camera with those types of robust features.

At this point, though, it's only a guessing game as to what Canon's intentions are. The good news is that we'll all know soon enough. LOL!

But speculation is what's so fun about forums like these -- a collective of happy hand-wringers Mwahahaha'ing the possibilities.

[David Heath]

Originally Posted by Lawrence Bansbach

If it's so much more complicated, why is every manufacturer using H.264? I seriously doubt Panasonic would announce new codecs if it couldn't reasonably implement them.

Such as H264 defines the DEcoder, not the coder. So H264 coders vary tremendously in complexity and performance - and price. Increase the complexity, increase the cost, possibily things like the difficulty of doing post work - the question has to be asked if it's worth it? Hence the analogy with the car. (Which wasn't intended too literally! ;-) )

Originally Posted by Lawrence Bansbach

Some people do it for the environment despite the added cost.

And the comparable situation may be when we're talking about broadcast transmission. Datarate may not be too significant when it just means using a bit more memory. But it's hugely important to a broadcaster with limited bandwidth - an increase in coding efficiency can mean several more channels in a multiplex! It may mean hugely expensive ( and efficient) coders which would not be viable if the benefit was just to save on memory costs - but a different story if it enables extra channel transmission.

Originally Posted by Lawrence Bansbach

Camera codecs and associated processors are the products of constant technological evolution. By the time Panasonic implements AVC-LongG, it'll probably be as cheap to do so as AVC-HD is today..

True - but memory costs are falling all the time as well, so the codecs, processors etc are chasing a moving target.

Originally Posted by Lawrence Bansbach

Possibly, but Panasonic claimed adding AVC-Intra to the AF100 would have added several thousand dollars to the cost.

It's difficult to see how that could be so. Look at the HPX250 and the AC160 - two recent models, fundamentally similar, but with the difference that the HPX250 has a few extra refinements and is P2/AVC-Intra to the AVC-HD of the AC160. With nowhere near "several thousand dollars" in cost between them - just over one thousand, but that includes more differences than just P2/AVC-Intra.

The AC160 is pretty much the same cost as the AF100. So how come the AC160 gets a P2/AVC-Intra big brother in the HPX250, but no equivalent for the AF100?

[Jon Fairhurst]

Originally Posted by David Heath

This is why 8 bits is enough for normal recording and viewing - it matches the range of the eye. It's easily proven by looking at greyscales and seeing how small the differences get before it merges into a continuous mass. (Corresponds to about 7 bits, or about 128 levels.)

Eight bits is enough when you...
1) Expose properly (that's the goal, but some extra headroom can save expensive, non-optimum footage.)
2) Get the S-curve and gamma just right, and
3) Don't grade.

But if you want to fix a non-ideally exposed image, mess with it's curve, and grade colors beyond reality while maintaining smooth gradients and natural texture, you need more bits.

[Charles W. Hull]

Originally Posted by Chris Barcellos

Canon already has the basic camera. How about the price of a Canon XF100, only a little bit of tweaking from there-- add a different sensor, rearrange the body a bit, and add an EOS mount. Everything else can remain close to same.

Oh Chris! I've ignored this whole thread until your post. I have an XF100 and I can just imagine using my Canon lenses with a large sensor on that camera; OMG that would be nice!!

[Brian Drysdale]

Originally Posted by Jon Fairhurst

Eight bits is enough when you...
1) Expose properly (that's the goal, but some extra headroom can save expensive, non-optimum footage.)
2) Get the S-curve and gamma just right, and
3) Don't grade.

But if you want to fix a non-ideally exposed image, mess with it's curve, and grade colors beyond reality while maintaining smooth gradients and natural texture, you need more bits.

A lot of landmark broadcast programmes have been shot using 8bit HDCAM, many of which have been graded in post. Exposing correctly is one of those skills DPs need to develop, although it's not as difficult with video as shooting reversal film. However, it's not always helped by the quality of the the LCD viewfinders found on the cheaper cameras, which are fine for framing but not really for the finer judgements you have to make in setting the exposre.

[Chris Barcellos]

Originally Posted by Charles W. Hull

Oh Chris! I've ignored this whole thread until your post. I have an XF100 and I can just imagine using my Canon lenses with a large sensor on that camera; OMG that would be nice!!

Hey Charles, how about this one. You remember 16mm and 8mm cameras with three lense rotation mounts. How about an internal rotation mount for two different sensor chips on the XF-100 type camera. Then you could switch between current XF 100 chip, and a larger sensor. Add interchangealbe EOS lens capability and you have a lot of needs covered in one camera.

[David Heath]

Originally Posted by Chris Barcellos

How about an internal rotation mount for two different sensor chips on the XF-100 type camera. Then you could switch between current XF 100 chip, and a larger sensor.

Well - I think that may be quite difficult to engineer. I can think of an easier way to sort of get what you want - what about the whole chip area being S35 and 4k sensor resolution. (Which I'll take to be the 16:9 version - 3920x2160.)

So either use the whole sensor, downscaling to 1080p or recording RAW - or just use the centre quarter area of the chip, which will be 1920x1080, and roughly equal to Super16 (?). (And with resolution equivalent to the XF100.)

One point that's worth mentioning is to do with the idea of two cameras sharing the same chip, a more expensive camera and a cheaper one, like the F3 and FS100. If we assume the S35 chip, and assume it to be a Bayer of 3920x2160, you could either read it in the normal way (to get such as 4k RAW) and then deBayer, OR treat it in a far simpler way.

It could be considered as a 1960x1080 matrix of blocks, each:

R G
G B

So treat each block as a single three colour pixel, and you very simply get full resolution 4:4:4 1080p. No need for de-Bayering, downconversion etc etc. So the more expensive camera gets full 4k ability (and possibly this mode as well), the cheaper one "just" gets the 4:4:4 1080p.

[Evan Donn]

No need for de-Bayering,

Maybe I'm confused about what you're suggesting, but it sounds like you're describing a standard bayer filter, which would of course require debayering. Plus, taking four adjacent samples and reducing them to a single pixel would also be a form of downconversion. The only difference I see in your two scenarios is that one performs the downconversion & debayering in post, while the other does it in camera.

[David Heath]

Originally Posted by Evan Donn

Maybe I'm confused about what you're suggesting, but it sounds like you're describing a standard bayer filter, which would of course require debayering.

Yes, a standard Bayer filter sensor. But a signal can be derived without a true deBayer process, which is exactly what happens with getting video from still camera sensors. The advantage of the system above is that the number of pixels (4x1960x1080, or about 8 megapixels) is optimum for the process. "DeBayering" tends to describe a fairly specific technique, and there's a good demo at HowStuffWorks "Demosaicing Algorithms: Color Filtering"

Digital cameras use specialized demosaicing algorithms to convert this mosaic into an equally sized mosaic of true colors. The key is that each colored pixel can be used more than once. The true color of a single pixel can be determined by averaging the values from the closest surrounding pixels.

Hence it's complicated. It will give a 3920x2160 output - though the actual resolution won't be as high as that, and the luminance resolution will be better than chrominance. Each output pixel will be formed by computation from values of surrounding pixels. It will give a lot better than 1920x1080 performance - as you may expect from a 4k sensor, but obviously it won't be as good as three 3920x2160 sensors.

Originally Posted by Evan Donn

The only difference I see in your two scenarios is that one performs the downconversion & debayering in post, while the other does it in camera.

No, either could happen in camera. But the former does need to do a true deBayer, and because the result is a 4k output, then needs to downconvert to get 1080p. A lot of computing to do it well.

The simpler option is equivalent to reading from three 1920x1080 chips, directly getting a R,G,B value for each 1080 output pixel. So no deBayering, and no downconversion - it gives 1080 directly. The only difference is that the R,G,B photosites are sitting side by side, not on three separate chips. (And there are two green photosites in the 2x2 block, but they could be averaged together.)

Originally Posted by Evan Donn

Plus, taking four adjacent samples and reducing them to a single pixel would also be a form of downconversion.

I suppose that depends on definition. I see what you mean, but "downconversion" tends to be used in relation to taking a higher definition video signal, and converting to that to one of lower definition. That's not the same as would be happening here, the lower definition signal is being formed directly on read out.

[Peter Moretti]

Originally Posted by Brian Drysdale

A lot of landmark broadcast programmes have been shot using 8bit HDCAM, many of which have been graded in post. Exposing correctly is one of those skills DPs need to develop, although it's not as difficult with video as shooting reversal film. However, it's not always helped by the quality of the the LCD viewfinders found on the cheaper cameras, which are fine for framing but not really for the finer judgements you have to make in setting the exposre.

Everything after the F900 has been unncecessary :) ;).

[Jack Zhang]

Originally Posted by David Heath

Ah, OK, found this - AVC-Intra - Wikipedia, the free encyclopedia . AVC-Ultra has been talked about for a while, but I'd only heard it referred to an I-frame only version at 200Mbs - presumably more intended to rival HDCAM-SR. They now seem to have widened the meaning, so the version I was aware of is now "AVC-Intra Class 200".

Please remember that this is Wikipedia. AVC-Ultra was Vaporware for the past 3 years. I don't see citations or references, so this could all be hokum.

[Brian Drysdale]

Originally Posted by Peter Moretti

Everything after the F900 has been unncecessary :) ;).

Not always the case, but it's more that 8 bit formats can probably manage the productions that most people are actaully shooting.

There are trade offs going on in a camera design and you may find that you can have other 10bit codec, but because of compromises in keeping the cost down, other aspects of the camera's design don't in reality make it that worthwhile. It may be better having lower compression 8bit rather than higher compression 10bit.

[Lawrence Bansbach]

Originally Posted by Jack Zhang

Please remember that this is Wikipedia. AVC-Ultra was Vaporware for the past 3 years. I don't see citations or references, so this could all be hokum.

It was announced at IBC (video).

[Chris Hurd]

An overview of AVC-Ultra (in English, with Spanish subtitles):

Panasonic AVC Ultra on Vimeo

[Robert Sanders]

The problem with using the word ULTRA is.... then what? What do you call the codec once you implement 1080p 4:4:4? AVC-MAX? And then what? What do you call the AVC flavor that supports 4K in the future? AVC-SUPERDUPER?