"Club Penton" - Ask questions to Hollywood insider "Penton-Man"

[Doctorossi]

Originally Posted by Penton-Man

I think that’s because ‘the degree’ to which you will appreciate it (the final outcome), is dependent upon a variety of factors including how the scene is lit, the glass used, any camera-based filtering, how much action is involved, etc.

I don't follow you.

Originally Posted by Penton-Man

Doc, on your display at home, could you identify any difference in resolution between any or all of the IMAX scenes of The Dark Knight Blu-ray as compared to the 35mm sequences?

Grain-structure and processing artifacts, yes. Spatial resolution... hmm... I'd have to double-check, but from memory, no.

[Bobby Henderson]

Simply put, the Bayer pattern grid of square pixels is an inefficient way of storing image detail.

A digital video camera with a native resolution of 1080p in its sensor(s) really isn't going capture all the detail that can be represented in that number of pixels. If you record the imagery at a higher resolution with a 4K video camera and then down-sample it to the 1080p level the imagery will indeed have improved detail.

The same principal is used in professional publishing. Photographs are typically scanned at ppi levels of at least 1.5 times the halftone screen or line screen number. Such imagery will always have better spatial resolution than an image scanned merely at a 1:1 ratio of ppi to line screen.

I've posted links to this great publication from Arri and will do so again:
http://www.efilm.com/publish/2008/05/19/4K%20plus.pdf

Anyone who reads that article should come away with a clear understanding of the great benefits that result from scanning film imagery at higher resolutions and then later down-sampling rather than merely doing 2K crap in a 1:1 scan and reproduction ratio.

[syncguy]

Originally Posted by Bobby Henderson

Simply put, the Bayer pattern grid of square pixels is an inefficient way of storing image detail.

A digital video camera with a native resolution of 1080p in its sensor(s) really isn't going capture all the detail that can be represented in that number of pixels. If you record the imagery at a higher resolution with a 4K video camera and then down-sample it to the 1080p level the imagery will indeed have improved detail.

.......

This is absolutely right.

The Nyquist-Shannon theorem, in information theory, explains this phenomenon. According to this theory, the sampling rate should be twice the frequency of the original information signal.

In order to reproduce at 2K (about 1080p), the scanning rate should be 8K (since the information is on two orthogonal, independent axis).

This can be imagined easily, if we think of photographing a 2000x1000 chess board pattern with a 2000x1000 pixel camera. If the pixel grid of the camera is perfectly aligned with the chess board pattern, the camera can reproduce the pattern correctly. However, in the worst case, if the camera pixel grid is 1/2 pixel out on both axis, it cannot reproduce the pattern correctly. If the camera had 4000x2000 pixels, it could have reproduced the pattern more accurately.

So, this also explains that 4k or 8k scan is much better than a 2k scan.

[Doctorossi]

This is all great information, but I'm not talking about scanning and Nyquist frequency; I'm talking about origination. I'm also not talking about Bayer patterns; I'm talking about 1:1:1 color pixels.

[Doctorossi]

Originally Posted by syncguy

In order to reproduce at 2K (about 1080p), the scanning rate should be 8K (since the information is on two orthogonal, independent axis).

How do you get 8k? The last time I checked, 2 x 2 = 4.

[Bobby Henderson]

This is all great information, but I'm not talking about scanning and Nyquist frequency; I'm talking about origination.

Um, if you're only "originating" with a 1080p HD camera (such as a Panavision Genesis for instance) you're dealing with a device that can't really capture all the detail that's capable of being squeezed into a 1920 X 1080 resolution image (didn't I say this already?). "Originate" the same material in 4K "digital" (or 35mm and then scan the film in 4K) and then reduce it down to 1080p. That down-sampled imagery will contain more spatial detail than anything produced in a 1:1 ratio.

That Arri article spells out the issue very clearly.

[syncguy]

Originally Posted by Doctorossi

How do you get 8k? The last time I checked, 2 x 2 = 4.

A pixel grid has two dimensions. The information on one dimension is independent from the other (i.e. two orthogonal axis). Hence the number of pixels should be increased (or doubled) in both dimensions. Therefore, 2x1K need to be scanned at 4x2k.

Edit: 4x2k is called 4K (4096x2160)

This is one of the reasons that 4K scanned film-based movies look better than the movies shot on 1080p video cameras on blu-ray especially on larger screens.

[NutsAboutPS3]

Originally Posted by Bobby Henderson

I've posted links to this great publication from Arri and will do so again:
http://www.efilm.com/publish/2008/05/19/4K%20plus.pdf

Anyone who reads that article should come away with a clear understanding of the great benefits that result from scanning film imagery at higher resolutions and then later down-sampling rather than merely doing 2K crap in a 1:1 scan and reproduction ratio.

Thanks for the article, it is certainly a very good article and an interesting read. However, while it shows that a 4k capture is clearly superior to a 2k capture, it doesn't address the question of whether downsampling that 4k capture to 2k will provide a better result than a 2k capture.

Originally Posted by syncguy

This can be imagined easily, if we think of photographing a 2000x1000 chess board pattern with a 2000x1000 pixel camera. If the pixel grid of the camera is perfectly aligned with the chess board pattern, the camera can reproduce the pattern correctly. However, in the worst case, if the camera pixel grid is 1/2 pixel out on both axis, it cannot reproduce the pattern correctly. If the camera had 4000x2000 pixels, it could have reproduced the pattern more accurately.

However, when you downsample that 4k or 8k scan to 2k, you will have exactly the same aliasing problem as when you tried to capture the chess board pattern at 2k. Unless you are happy to shift the image by half a pixel, you will still not be able to make an accurate reproduction. And if your image contains 2 pictures of chessboards, one perfectly aligned with the 2k pixels, and one half a pixel misaligned, there would be no way to reproduce both of them correctly by shifting the image, no matter how high a resolution you capture at before downsampling.

I have a vague recollection of penton having addressed this issue in a post a while back, with mention of a test where people found they could not distinguish between a direct 2k master, and a 4k master downsampled to 2k. How you do the downsampling would be a factor in such a comparison, as the downsampling algorithm can effectively apply sharpening to the image, and if this were the case you would need to apply the same sharpening to the 2k master to have a valid comparison.

[wallendo]

Originally Posted by syncguy

A pixel grid has two dimensions. The information on one dimension is independent from the other (i.e. two orthogonal axis). Hence the number of pixels should be increased (or doubled) in both dimensions. Therefore, 2x1K need to be scanned at 4x2k.

This is one of the reasons that 4K scanned film-based movies look better than the movies shot on 1080p video cameras on blu-ray especially on larger screens.

I don't get this. 4K resolution is FOUR TIMES 2K resolution. As I understand it, when talking about film scanning, the "K" number refers to the number of horizontal pixels; i.e., 4K has 4000 horizontal pixels. However, unless you are using wide (non-square) pixels, you would also be doubling the number of vertical pixels. 2 X 2 = 4.

[Doctorossi]

Originally Posted by syncguy

A pixel grid has two dimensions. The information on one dimension is independent from the other (i.e. two orthogonal axis). Hence the number of pixels should be increased (or doubled) in both dimensions. Therefore, 2x1K need to be scanned at 4x2k.

Right. Which is called a 4k scan. Again, where do you get 8k?

[Penton-Man]

Originally Posted by Bobby Henderson

Simply put, the Bayer pattern grid of square pixels is an inefficient way of storing image detail.

A digital video camera with a native resolution of 1080p in its sensor(s) really isn't going capture all the detail that can be represented in that number of pixels. If you record the imagery at a higher resolution with a 4K video camera and then down-sample it to the 1080p level the imagery will indeed have improved detail.

The same principal is used in professional publishing. Photographs are typically scanned at ppi levels of at least 1.5 times the halftone screen or line screen number. Such imagery will always have better spatial resolution than an image scanned merely at a 1:1 ratio of ppi to line screen.

I've posted links to this great publication from Arri and will do so again:
http://www.efilm.com/publish/2008/05/19/4K%20plus.pdf

Anyone who reads that article should come away with a clear understanding of the great benefits that result from scanning film imagery at higher resolutions and then later down-sampling rather than merely doing 2K crap in a 1:1 scan and reproduction ratio.

(I bolded the pertinent phrase)

Yes, I agree with everything you’ve said but, just for accuracy purposes and so that people here don’t get the wrong idea about “a 4k video camera” compared to an HD video camera, I think just a little elaboration is needed. As it seems like this is yet another example of where some folks may have redefined terms due to lack of knowledge…..with perhaps? a little nudge from the respective camera companies involved.
And I must admit that I often fall into the same trap as you do in regards to using the terminology “4k video camera.”

It’s like this.
For years, the camera community has defined “HD” video cameras as those that have a pickup device that produces 1920 x1080 worth of information in a horizontal direction on all three primary color channels. Example being most 2/3 inch cameras (such as the Sony F23), which generally have 3 (three) 1920x1080 sensors or, for the Panavision Genesis camera, which contains one 5760 x 2160 sensor (which has the same number of pixels [1920] devoted to each of the primaries, i.e. true RGB)

So, by extension of the historical definition, a “4k video camera” should imply a camera that makes 4K worth of information in the horizontal direction on all three primary color channels. Well…..they don’t.
Stacey can correct me if I’m wrong but, his “4k video camera” has one 4k sensor with a Bayer pattern. A more correct terminology would be to call his acquisition device a “4k RAW” camera.

Still……..a fine camera, for an exceptionally fine price.

[Penton-Man]

Read………….
https://forum.blu-ray.com/showthread...ua#post1633904

Now…………
Joshua’s response regarding the viewing of his motion picture –

Well, to say the least, he wasn’t happy because he felt that the ‘colored’ imagery with the color filters turned out hardly as subtle as he had intended it to.

So, after he saw the public preview, he went to his producer and asked that he take back the film and get the overt color filtered sequences either removed or bleached out to make them look more subtle.
The producer informed him that it was too late to do that, unless other sold-out previews were cancelled and he (the producer, namely George Skouras) was unwilling to do that.
Allegedly, by Joshua’s own words, the producer told him….”It’s your fault. You made it cost too much money, so I’ve got to get my money back.”

Next installment ----How South Pacific manifested itself in terms of financial success and how it influenced Joshua Logan’s further Directorial career in the movies…….in particular with Jack Warner of Warner Bros. Studios.

[Penton-Man]

Originally Posted by Kris Deering

On another note, Stacey Spears and Don Munsil have just put together an OUTSTANDING Blu-ray calibration disc that will be available for purchase very soon. It is also being bundled with the upcoming Oppo BDP-83 Blu-ray player. If you want to get the most out of your HD display, this is an OUTSTANDING disc with clear explanations on how to use the test patterns to get the most of your display.....

Where’s Stacey or Kris?
I want my two dollars! (for my plug of the calibration disc ^)

http://www.youtube.com/watch?v=mXIBjo8gWEE

[Doctorossi]

Originally Posted by Penton-Man

A more correct terminology would be to call his acquisition device a “4k RAW” camera.

More correct, but still problematic, as Red's definition of "RAW" is not the industry standard format.

If you ask me, the Red One is a very good black & white camera.

[Penton-Man]

Originally Posted by David Forbes

Steelers for the win! (I'll send you a picture of my Terrible Towel once the Ravens have been roasted!)

You don’t have to, as it seems like it will be issued in high definition packaged home media (sans cable/satellite compression artifacts) for your pleasure…………….
https://www.blu-ray.com/news/?id=2461

[cwesley]

I have a question as to the flickering of the black bars. Is this normal. It will change shades of black within a scene. It can be very distracting.

Originally Posted by Kris Deering

I think Jeff may have nailed it with this one. Your display's contrast abilities change dramatically with the overall APL of what is on the screen. So movies that tend to be brighter or have a lot of bright imagery may appear to have darker bars. This could also be a trick of the eye as things appear darker when they have lighter objects next to them. TDK is a very dark movie so the bars don't have a lot of contrasting objects near them, but they are not elevated.



Below black material in film isn't even a consideration if you have a display properly calibrated so that black is at digital 16 as it should be, you would never see it anyways. Anything that is digital 16 or below would be the same value of black on your display.

On another note, Stacey Spears and Don Munsil have just put together an OUTSTANDING Blu-ray calibration disc that will be available for purchase very soon. It is also being bundled with the upcoming Oppo BDP-83 Blu-ray player. If you want to get the most out of your HD display, this is an OUTSTANDING disc with clear explanations on how to use the test patterns to get the most of your display. There is also a great montage shot by Stacey himself with a 4K Red camera!

[Doctorossi]

Originally Posted by cwesley

I have a question as to the flickering of the black bars. Is this normal. It will change shades of black within a scene.

Sounds like you may have some sort of 'dynamic contract' circuit engaged?

[cwesley]

What should I do about it? My tech knowledge is limited.

[Doctorossi]

Do you have a manual for your display device? Otherwise, you might begin by searching its menu system.

[cwesley]

I have 80g PS3 and a Sony Bravia 1080p display. Should I be looking for a dynamic contract circuit?