IBC: Sony announces HVR-V1e

[Steve Mullen]

Originally Posted by Emmanuel Plakiotis

The reason why the V1/FX7 and the previous HC1/A1, HC3 sport a 4:3 sensor instead of a 16:9 has to do with their ability to shoot high resolution still images and is not a indication of inferior quality.

If we ignore the exact V1 CMOS resolution until Sony makes their announcement -- the CMOS is about 1000x1000.

In the case of the Z1, its 960x1080 pixels are "wide" so they can be spread over a 16:9 area. I guess that means they are twice as wide as they are tall. If so, their pixel aspect-ratio is 2:1.

ClearVid has square pixels -- if the drawings are to be believed. Given the chips have about 1000x1000 pixels, that would seem to mean the CMOS chip itself is nearly square.

So, we have a nearly square sensor onto which a 16:9 rectangular image is projected.

Think of it as 1000 wide and 1000 high while an HDCAM chip is 2000 wide and 1000 high. The latter is 16:9 so the former must be 8:9. This not 4:3 either.

The published pix of the ClearVid sensor shows it to be 16:9, not 8:9 or 4:3. (Although this pix may not be of the V1 chips.)

Note, that I'm not talking about the fact the pixels are diamonds. However, as I understand it, the diamonds make it possible to squeeze pixels even closer together, thus likely making taller pixels.

[Simon Wyndham]

IMHO there hasn't been a great film made in the last decade.

Sounds like you need to look further afield than just American output!

[R Geoff Baker]

In truth, there haven't been many 'great' films made, ever, in any decade.

But to ignore the last decade seems a tad stubborn, as would ignoring indie films -- even reaching back to the pre1995 films, some of the best were independent productions.

I've watched thousands and thousands of films over the years, and the ratio of dreck to cream has always been in favour of dreck -- but there have been films as good as any in the last decade, and there have been some great indies. And World Cinema has never offered better opportunities ...

It's a big old world out there -- plenty of good stuff to go around!

GB

[Steve Mullen]

Originally Posted by R Geoff Baker

In truth, there haven't been many 'great' films made, ever, in any decade.

But to ignore the last decade seems a tad stubborn, as would ignoring indie films -- even reaching back to the pre1995 films, some of the best were independent productions.

I agree that some pre 1995 "indies" were great. Then "indie" became a marketing term as did "Shot on DV!" and is "World Cinema" now. There's nothing new about cinema from other countries. I've seen most every film that came into the USA from the late-50's -- long before it became "fashionable" to import "ethnic" films.

Unfortunately -- and this is really OT -- filmmakers world-wide have learned to copy (because via DVD they can see everything) what "sells" and so no matter the country of origin, they are copies of the last big success.

Most adults have stopped going to movies and with home theaters and HD -- even fewer will go in the future. That means no matter the country of origin, films have been and are being made for teens. Which in turn means those who are under 30 have primarily seen only "teen" movies -- so unless they have taken many film history classes -- they make copies of what they've seen. This is how both music and film have become worse over time.

It's the reverse of what happended after WW2, when a few great films from Italy led others in Europe, India, and Japan to make wonderful films. By the `60's and `70's the USA joined the ranks of great filmmakers -- and then the Chinese. Then it was over. Suddenly film and music became fast food for the eyes and ears.

Fast food isn't any better at 24p than it is at 60i.

[Emmanuel Plakiotis]

Originally Posted by Steve Mullen

In the case of the Z1, its 960x1080 pixels are "wide" so they can be spread over a 16:9 area. I guess that means they are twice as wide as they are tall. If so, their pixel aspect-ratio is 2:1.

ClearVid has square pixels -- if the drawings are to be believed. Given the chips have about 1000x1000 pixels, that would seem to mean the CMOS chip itself is nearly square.

Think of it as 1000 wide and 1000 high while an HDCAM chip is 2000 wide and 1000 high. The latter is 16:9 so the former must be 8:9. This not 4:3 either.

Note, that I'm not talking about the fact the pixels are diamonds. However, as I understand it, the diamonds make it possible to squeeze pixels even closer together, thus likely making taller pixels.

http://www.hdvinfo.net/articles/sony/firstlookv1u.php

If you take a look at the V1U announcement above you will see that it clearly states the pixel number as 1080X960 which means that you were right claiming the chip being 9:8 (I think thats the right way and not 8:9, since the latter calculates to a totally different number and traditionally the industry uses frame size ratios with the bigger side being on top).

My question, how are you certain that the pixels are square and not wide like the Z1?

[Marvin Emms]

"If you take a look at the V1U announcement above you will see that it clearly states the pixel number as 1080X960"

If you look at the Sony page the word they use is 'effective'.

This does not automatically bear any relation to the arangement of physical pixels.

The number to remeber is 1MPixel.

Similar to the pixel count one would expect in a 720p system.

Edit, Working out the aspect ratio is not useful without more information and without taking into account of the clearvid pixel arangement. A square grid as you have calculated would be 8:9 not 9:8. This is meaningless maths though and my money is on square pixels tilted 45 degrees in a 16:9 arangement for mathematical reasons allready posted.

[Boyd Ostroff]

I just attended the press conference where they announced the V1, and Hugo Giaggioni (Sony CTO) gave a technical presentation about the sensors. I really haven't been following all this, so maybe it's old news to you? But I've sent a lot of info to Chris which I'm sure he will eventually put online.

In the meantime, you might find the following to be of interest:

[Marvin Emms]

Thanks Boyd, you are a saving my sanity.

That is how I guessed they were counting them, and the problem is, its not self consistant. If you use the method in one direction, you get 960x1080, if you use it in the other direction, you get 1920x540 for the full sensor.

Either interpretation is true, and neither is indicative of the performance, its just a convenient method to make people think they can compare the performance directly with the FX1 and its siblings.

Edited, I had X and Y the wrong way round, thanks Heath!

[Boyd Ostroff]

Here's another one:

[Heath McKnight]

Originally Posted by Marvin Emms

Thanks Boyd, you are a saving my sanity.

That is how I guessed they were counting them, and the problem is, its not self consistant. If you use the method in one direction, you get 1080x960, if you use it in the other direction, you get 540x1920 for the full sensor.

Either interpretation is true, and neither is indicative of the performance, its just a convenient method to make people think they can compare the performance directly with the FX1 and its siblings.

I think you mean 960x1080 and 1920x540; it's horizontal, then vertical.

heath

[Thomas Smet]

This kind of reminds me of a bayer pattern where every other green pixel is interpolated. In this case pixel 1 and 3 are real pixels while pixel 2 takes a chunk from 4 different pixels. So does this mean the luma resolution is kind of on par with what a single chip camera would do?

a single cmos camera may have a 1920x1080 chip but every other green pixel is interpolated from the bayer pattern. This leaves you with 960 true green pixels.

In the 3 cmos camera you start out with 960 but because of the way the pixels are lined up you get an interpolated result equal to 1920.

I would think the results would be somewhat close except for the chroma which would be better then the bayer pattern chroma.


If this new method works so well I wonder how it would have looked if they used a 960x540 chip instead of a 960x1080 chip? The vertical resolution could gain just as much as the horizontal. I'm sure it wouldn't have been as good and I am glad they went with a 960x1080 chip. If the vertical is doing the same clearvid interpolation isn't that like a 1920x2160 oversampled source?


One question. How are the in between pixels formed? Is it a (1+1+1+1)/4 blend or some other way the pixel shows up. If a single pixel is pulling in 4 different points of light which one would it use?

[Michael Struthers]

"IMHO there hasn't been a great film made in the last decade"

Wow, finally someone has reached the old geezer stage of "it was all better back in the day" phase...film opinions are like....uhh....noses, of course, but c'mon...

How about

"Amoros Perros"
"Requeim for a Dream"
"Traffic"
"Sixth Sense"
"American Beauty"
"Fight Club"
"Mullholland Drive"
"Eternal Sunshine of the Spotless Mind"

Hell even
"Y tu Mama Tambien"
"Elephant"
"Crouching Tiger Hidden Dragon"
"Lord of the Rings"
"GhostWorld"

Not too mention all the great docs....

[Simon Wyndham]

I've seen most every film that came into the USA from the late-50's

Well, mostly the overseas films that are brought into the States are ones that distributers think will appeal to US audiences. Hence why they look very similar to US output.

[Steve Mullen]

Originally Posted by Emmanuel Plakiotis

http://www.hdvinfo.net/articles/sony/firstlookv1u.php
My question, how are you certain that the pixels are square and not wide like the Z1?

I'm not certain at all.

The problem is we have drawings and a pitch specification. Plus, I've never seen a CMOS chip that did not hve square pixels. And, in many cases, the chips have equal X and Y resolutions.

They certainly could be diamond pixels laid into a 16:9 array.

[Steve Mullen]

Originally Posted by Marvin Emms

... Neither is indicative of the performance, its just a convenient method to make people think they can compare the performance directly with the FX1 and its siblings.

You were certainly correct about the HOW, but I think given that Sony is using ClearVid tek -- there's no way they can avoid not comparing the resulting resolution with that of previous products.

It seems you are saying ClearVid is a way of getting good test numbers and do not represent the real-world. This is the same argument I made about pixel-shift. And, I was proven correct as dynamic testing showed a significant drop from static tests.

The FX7/V1, unlike the FX1/Z1, will measure the same under static and dynamic tests. But, I agree with you, that as soon as Adam Wilt turns it to a diagonal, EITHER resolution will decrease or aliasing will increase, or BOTH.

I've said it before and I'll say it again: a camera must have sensors that have AT LEAST the number of pixels as the recording format. There is no way to get the missing pixels if they are not there.

However, given that Sony is not supplying camcorders with 1440x1080 pixels -- although I suspect they supply such CCDs to Canon -- the question is will testing reveal ClearVid CMOS to perform better than pixel-shift CCD technology. And, Sony clearly will have to compete with Canon and JVC.