Jump to content

Mtf Of Blu-ray/hd-dvds


Recommended Posts

For the technically inclined, there's an interesting picture of aliasing created by a camera designed without an aliasing filter, in the section The Nyquist sampling theorem and aliasing near the top of this webpage. To the left of the picture is included this text:

The 14 megapixel Kodak DCS 14n, and its successors, the Pro N and Pro C, have no antialiasing filter. MTF response is outstanding, but remains high (0.31) at the Nyquist frequency.
Don't videos always have some degree of blurring? I mean if everything was resolved down to single pixels then it would look all aliased and pretty bad, like when you play video games without anti-aliasing - only much worse because the real world has more detail.

I agree.

mlx we are not talking tv cameras here.

Alebonau,

there are several reasons for mentioning tv cameras, including:

1. The 2nd para of the post you provided, attributed to Josh Zyber, starts with these words:

Video sourced material on a DVD is much sharper than film originated

material.

This is obviously a reference to video cameras.

2. It is only for tv cameras that we have access to graphs that show the MTF of the lens, and the MTF of the camera including its lens. With such graphs some sort of discussion and analysis is possible. The manufacturers of film cameras do not appear to publish such data - it is perhaps a less competitive industry in the MTF stakes. However it is important to bear in mind the words in bold at 1 above that a DVD is sharper made from a video camera than made from film. It can be said that studio TV cameras provide the pinnacle of 1920x1080 camera technology that is in everyday use. They are certainly not cheap. We expect our studio talkshow host and their guests, to appear with pristine sharpness.

I do not dispute that high-end professional cameras are capable of very good performance. There would therefore be no purpose served in my responding to various weblinks and quotes you have provided that provide evidence of good performance. I already agree!

But that is not really what this discussion is about, anyway, as I understand it. It is about measured Modulation Transfer Function for material converted into a digital representation, notably the 1920x1080 format, particularly for detail that lies in the region of performance from about 50% of the Nyquist frequency up to the Nyquist frequency. The MTF is far lower than most people assume it would be. I am not certain that that message has registered because I continue to see paragraphs and paragraphs devoted to how good professional equipment can be, and yet virtually no comment on Modulation Transfer Function at spatial frequencies near the Nyquist spatial frequency. Brisvegas1 seems to appreciate this. Most contributors seem not to.

Putting this into an audiophile context, audiophiles expect high frequency performance of a compact disk to extend up to 20KHz, which is fairly close to the Nyquist limit of 22.05KHz (the samping rate of compact disks is set by the red book standard at 44.1KHz). A variation at 20KHz of 0.1 dB would not be unusual for a CD player.

But in a video context, the response at 10/11ths of the Nyquist limit is more than 0.1 decibels up or down.

The horizontal Nyqusit limit of 1920x1080 video is half of 1920, or 960. 20/22.05 of 960 is 871. Looking at the HD studio camera performance graph (top of page 6 of the pdf ) we see that the MTF for 871 lines is about 40%. This is drop of 8 decibels. Even for a microphone that would be a poor result.

How about the performance of CDs at 15KHz? Audiophiles would expect extremely solid performance at such a frequency. Let's see. That's 68% of the Nyquist frequency of 22.05 KHz. In the context of a digital 1920x1080 camera that corresponds to its performance at 653 lines. Reading from the graph we see that the overall MTF of the hoizontal resolution of a professional studio grade camera at 653 lines is about 65%. Ouch. That is a drop of 3.7 dB. Hardly audiophile standard. But perhaps not all that bad for a set of loudspeakers at 15KHz.

We still need to keep all of this in context. Although the human eye can detect variations in intensity levels of about 2% it can also make unconscious adjustments. A 35% drop in intensity is very noticeable but if it is only for the more detailed parts of an image, the eye will simply seek out the details. And the image will not look as sharp. If the detail starts out as very high contrast the eye will still be able to see it at a reduced MTF. If it is very subtle to begin with, the poor MTF may be fatal. The detail will be lost in the greyness, and the video noise, or the introduced film grain effect.

A graph Optical modulation transfer function (MTF) of the human eye appears towards the bottom of this webpage. MTF is shown for pupil sizes from 2 mm (bright lighting; f/8), to 5.8 mm (dim lighting; f/2.8). The iris dilates if the eye is subjected to a period of dim lighting. MTF is less under these conditions. It is well known that people look different in the cold hard light of day than in dim lighting. Wrinkles or other skin blemishes are no longer hidden!

Thudd's question at the next post is quite a sensible question and it has an answer [based on whether the cgi can move and still look smooth]. Many of today's movies incorporate cgi.

Cheers,

MLXXX

Edited by MLXXX
Link to post
Share on other sites
  • Replies 426
  • Created
  • Last Reply

Top Posters In This Topic

Let me add an analogy into the discussion.

First let's get a 1920x1080 computer monitor and display on it pixels of alternating colour - black and white for arguments sake - so that the screen is displaying a large 1920x1080 B&W checkerboard.

Now let's take our XYZ film camera and point it at the screen so that it exactly fills the camera's frame and take some snaps.

Now let's process that in whatever ways are necessary to produce a HD/BR disc.

Stick the disc into your player and see what comes out on your HD screen. Do you see an perfect 1920x1080 checkerboard? Without doing the experiment I can't say for sure, but I'd be pretty confident in saying no. As cameras and sensors improve the final image will get closer and closer to the original computer monitor but as soon as we start sticking analogue devices - like lenses - into the image path then the information in the final product will not match the original source.

I'd be interested to see where CG movies come into it. I would guess that since they could be rendered at the exact resolution of the output device, and transferred digitally the whole way, that the final HD/BR disc would indeed have a fully resolved 1920x1080 image on it. Whether they're done that way or not, or if they're rendered at a higher/lower resolution and up/downscaled to the resolution of the intended format, I don't know.

Edited by Thudd
Link to post
Share on other sites
Let me add an analogy into the discussion.

First let's get a 1920x1080 computer monitor and display on it pixels of alternating colour - black and white for arguments sake - so that the screen is displaying a large 1920x1080 B&W checkerboard.

Now let's take our XYZ film camera and point it at the screen so that it exactly fills the camera's frame and take some snaps.

Now let's process that in whatever ways are necessary to produce a HD/BR disc.

Stick the disc into your player and see what comes out on your HD screen. Do you see an perfect 1920x1080 checkerboard? Without doing the experiment I can't say for sure, but I'd be pretty confident in saying no. As cameras and sensors improve the final image will get closer and closer to the original computer monitor but as soon as we start sticking analogue devices - like lenses - into the image path then the information in the final product will not match the original source.

I'd be interested to see where CG movies come into it. I would guess that since they could be rendered at the exact resolution of the output device, and transferred digitally the whole way, that the final HD/BR disc would indeed have a fully resolved 1920x1080 image on it. Whether they're done that way or not, or if they're rendered at a higher/lower resolution and up/downscaled to the resolution of the intended format, I don't know.

We aren't talking home movies here or "happy snaps", we are talking Major Motion Pictures which are filmed to be screened on screens as large as IMAX. The resolutions are way beyond 1920x1080 (imagine that put on imax :blink:). By the time the film is downscaled to that resolution any deficiencies have been long since filtered out and you are getting a pure 1920x1080 print.

PS you wouldn't have a black and white checkerboard either you would have some shade of grey.

Link to post
Share on other sites
  • Moderators
That's it Al. I'd really like to see some more from Hosko.

We aren't talking home movies here or "happy snaps", we are talking Major Motion Pictures which are filmed to be screened on screens as large as IMAX. The resolutions are way beyond 1920x1080 (imagine that put on imax :blink:). By the time the film is downscaled to that resolution any deficiencies have been long since filtered out and you are getting a pure 1920x1080 print.

unfortunately not, something people seem to be able to understand I don't think. sigh..hosko ? he probably got bored and went home !

mlx we aren't talking digital slrs here, but anyways I remember the kodak dcs 14n from 4-5 years ago, from around the time I bought my canon 10d. the 14n is of quite different construction to anything we're talking here and a very old design now, with digital slrs really having marched on since. if your keen on talking digital slrs theres a digital slr thread in the off topic sub forum plus I guess a multi tude of camera forums if something you like yakking about.

Link to post
Share on other sites


Guest JimboTHX1138
We aren't talking home movies here or "happy snaps", we are talking Major Motion Pictures which are filmed to be screened on screens as large as IMAX. The resolutions are way beyond 1920x1080 (imagine that put on imax :blink:). By the time the film is downscaled to that resolution any deficiencies have been long since filtered out and you are getting a pure 1920x1080 print.

This sounds accurate to me. A master film print has approx. 2-3 times the resolution of a BD/HD-DVD. The copies that get sent to cinemas have a little less. Anyway, even if half the resolution is lost due to MTF, and the DVD or BD/HD-DVD has been made with the master film print, which they are, it would easily leave 720x576 for DVD or 1920x1080 for BD/HD-DVD to be fully resolved.

Link to post
Share on other sites
... we are talking Major Motion Pictures which are filmed to be screened on screens as large as IMAX. ...

I believe movies to be screened on IMAX are specially produced. IMAX format uses a print that is more than 35mm.

Actually I'd remind everyone the topic is about the MTF of Blu-ray and HD-DVDs.

I think it has been established that professional cameras have good performance, though I think Owen would like to remind people that even professional lenses have tapering MTFs at high spatial frequencies.

The central issue to my mind is: how you funnel the information from an analogue source [or a high pixel architecture digital format] down to 1920x1080 for Blu-ray or HD-DVD and somehow get a high MTF.

I think the basic answer is that the transfer cannot be done with a high MTF. I think the primary reason for this is sampling theory, and the need to give a smooth result that looks as smooth as film, but is in fact digitally based.

This may seem an anti-climax but it is the basic argument that has been repeatedly put, raised by Owen in the progenitor thread: you cannot get anything like 100% MTF between adjacent pixels in the digital 1920x1080 format.

For some reason, most contributors seem to want to talk about aything but MTF.

Edited by MLXXX
Link to post
Share on other sites
The central issue to my mind is: how you funnel the information from an analogue source [or a high pixel architecture digital format] down to 1920x1080 for Blu-ray or HD-DVD and somehow get a high MTF.

This is no longer relevant once a movie gets to this format. The only way it is relevant is if you are filming at 1920x1080 in which case we are not talking about material filmed for blu-ray or HD DVD and is therefore also irrelevant.

PS Movies such as 300 where screened on IMAX without being specially reduced.

This may seem an anti-climax but it is the basic argument that has been repeatedly put, raised by Owen in the progenitor thread: you cannot get anything like 100% MTF between adjacent pixels in the digital 1920x1080 format.

and this is clearly wrong. Hell I could cheat and point out CGI films where MTF is not even an issue, but it is also not true for Movies filmed at high res and transfered to these formats.

Link to post
Share on other sites
Hell I could cheat and point out CGI films where MTF is not even an issue ...

This is the point Thudd queried this afternoon.

Momaw, you are free to cheat by referring to a CGI movie. Could you please enlighten us as to the MTF for a premium CGI movie in 1920x1080 format at 960 lines horizontal resolution (the Nyquist limit) on a Blu-ray disk?

Link to post
Share on other sites


This is the point Thudd queried this afternoon.

Momaw, you are free to cheat by referring to a CGI movie. Could you please enlighten us as to the MTF for a premium CGI movie in 1920x1080 format at 960 lines horizontal resolution (the Nyquist limit) on a Blu-ray disk?

since there is no camera or optics involved at any stage then there is no Modular Transfer Function issues whatsoever.

Link to post
Share on other sites
since there is no camera or optics involved at any stage then there is no Modular Transfer Function issues whatsoever.

In one sense this is correct. There is no 'issue' in that the source is sickly sharp. [forgive me this lapse into juvenile-speak]

However it is not quite as simple as that. The cgi generator needs to anti-alias.

The extent to which it anti-aliases is a question of taste, and/or the purpose of the cgi.

If the source is a training video of still images, that are text or simple graphics that can be aligned with the 1920x1080 grid, the Modulation Transfer Function could be set at 100%: a fully black pixel next to a fully white pixel.

Alternatively, if it is ok the cgi looks like a cartoon, but the cartoon has to move; in order for the movement to be smooth, we will need to lose about 50% of the apparent resolution. Without being all that precise, we could suggest that MTF could be set at about 50%. All edges would be softened, using software [no pun intended].

But if it is intended that the cgi pass as a real world image [e.g. a famous actress is involved in an accident just before filming is complete and an image of her has to be inserted into remaining scenes artifically by cgi] it is a different ballgame. The image will most likely need to be blurred further, further reducing the MTF; so as to blend in seamlessly with the material captured by a camera, and not to contain a hint it is based on a sample grid of pixels.

Edited by MLXXX
Link to post
Share on other sites
In one sense this is correct. There is no 'issue' in that the source is sickly sharp. [forgive me this lapse into juvenile-speak]

However it is not quite as simple as that. The cgi generator needs to anti-alias.

The extent to which it anti-aliases is a question of taste, and/or the purpose of the cgi.

If the source is a training video of still images, that are text or simple graphics that can be aligned with the 1920x1080 grid, the Modulation Transfer Function could be set at 100%: a fully black pixel next to a fully white pixel.

Alternatively, if it is ok the cgi looks like a cartoon, but the cartoon has to moves in order for the movement to be smooth, we will need to lose about 50% of the apparent resolution. Without being all that precise, we could suggest that MTF could be set at about 50%. All edges would be softened, using software [no pun intended].

But if it is intended that the cgi pass as a real world image [e.g. a famous actress is involved in an accident just before filming is complete and an image of her has to be inserted into remaining scenes artifically by cgi] it is a different ballgame. The image will most likely need to be blurred further, further reducing the MTF; so as to blend in seamlessly with the material captured by a camera.

Ah but the argument was that 1920x1080 could not be fully resolved in blu-ray or HD DVD period. This is clearly not the case.

Link to post
Share on other sites

Have you seen a training video of still text and graphics on Blu-ray?

That is the only type of context where you could get 100% MTF.

Link to post
Share on other sites


That's it Al. I'd really like to see some more from Hosko.

I'm not going to begin to say I understand the theories that some people are putting forward, all I'll say is that it probably holds true for digital cameras.

However the vast majority of Hollywood films made in the present day are shot on film. Film is made of emulsion which has silver halide crystals inside, the upside of it being organic and not digital is each crystal acts independently from each other. When light hits each crystal the silver ions build up a collection of uncharged atoms which forms the basis of the latent image. Again I'll say that each crystal acts independently from each other.

Could someone now explain how Nyquist relates to this organic phenomenon.

I studied film not science and this discussion has gone well beyond the realms of what people in the industry discuss, it reads more like a SMPTE white paper then a discussion on technique.

Link to post
Share on other sites
Have you seen a training video of still text and graphics on Blu-ray?

That is the only type of context where you could get 100% MTF.

That's nonsense and you know it.

I studied film not science and this discussion has gone well beyond the realms of what people in the industry discuss, it reads more like a SMPTE white paper then a discussion on technique.

Yup. And the fact they keep going back to 1920x1080 cameras as their proof when we know they have no relevance to Blu-ray or HD DVD.......

Link to post
Share on other sites
That's nonsense and you know it.

I don't think I can put what I've said any more clearly. Perhaps if you reread this thread in the future it will make more sense to you than it does now.

Film is made of emulsion which has silver halide crystals inside, the upside of it being organic and not digital is each crystal acts independently from each other. When light hits each crystal the silver ions build up a collection of uncharged atoms which forms the basis of the latent image. Again I'll say that each crystal acts independently from each other.

Could someone now explain how Nyquist relates to this organic phenomenon.

In the transferring of the film to a sample grid of 1920x1080 for use on an HD-DVD or Blu-ray disk.

Edited by MLXXX
Link to post
Share on other sites
I don't think I can put what I've said any more clearly. Perhaps if you reread this thread in the future it will make more sense to you than it does now.

In the transferring of the film to a sample grid of 1920x1080 for use on an HD-DVD or Blu-ray disk.

I think you may need to reread it as you are fixated on equipment that is not used for blu-ray or HD DVD - either in the original capture of the source, the source itself or the process in getting it to the formats.

Link to post
Share on other sites


I have been away and have not read the posts in this tread, so this is not in response to anyone.

....

Owen

Edited to correct inoperative links.

The gist I'm getting is that "full HD" doesnt have 1920x1080 resolvable resolution? How does this affect viewing distances (discounting PC's and CGI's I think :/ ) if there isnt really the resolution there to resolve?

Link to post
Share on other sites
  • Moderators
I believe movies to be screened on IMAX are specially produced. IMAX format uses a print that is more than 35mm.

Actually I'd remind everyone the topic is about the MTF of Blu-ray and HD-DVDs.

I think it has been established that professional cameras have good performance, though I think Owen would like to remind people that even professional lenses have tapering MTFs at high spatial frequencies.

The central issue to my mind is: how you funnel the information from an analogue source [or a high pixel architecture digital format] down to 1920x1080 for Blu-ray or HD-DVD and somehow get a high MTF.

I think the basic answer is that the transfer cannot be done with a high MTF. I think the primary reason for this is sampling theory, and the need to give a smooth result that looks as smooth as film, but is in fact digitally based.

This may seem an anti-climax but it is the basic argument that has been repeatedly put, raised by Owen in the progenitor thread: you cannot get anything like 100% MTF between adjacent pixels in the digital 1920x1080 format.

For some reason, most contributors seem to want to talk about aything but MTF.

yes this thread is about hd-dvd & blu-ray, not digital slrs, tv cameras or screenies and your display capabilties.

and sorry to burst your bubble but we are yet to see anything from you or owen on anything inregards the capabilties of move production that disputes abilities to get 1920x1080 on hddvd or blu-ray. all you guys have talked about are digital slrs !

on the contrary I have provided much info, none of it disputed with any rebutall or with any contrary evidence to say any different.

you guys have not provided any evidence to show mtf in filming movie blockbusters reduces res to below 1920x1080 for hi-def disc formats.

and inregards owen, he has infact provided jacksh!t to this disussion in regards movie making to dispute 1920x1080 res for movies for the hidef disc formats. he and you mlx have provided absolutely nothing, which is really not suprising and I totally understnd the reasons why.

hosko on the otherhand, who actually knows what he is talking about, from his knowledge & experiece has already made clear lenses used in movie can easily achieve 4k and I myself totally understnd why you or anyone has been able to provide any evidence to say any otherwise.

Link to post
Share on other sites
  • Moderators
The gist I'm getting is that "full HD" doesnt have 1920x1080 resolvable resolution? How does this affect viewing distances (discounting PC's and CGI's I think :/ ) if there isnt really the resolution there to resolve?

that is the very misguidded owen gist, baswd on digital slrs with nothing to do with movie camera. and is actually quite wrong if you read the info in the earlier posts of this thread you'll see the systems used to produce holywood movies are certainly capable. and owen knows it, which is why we should all watch really big screens with our noses glued to them to fully resolve the full 1920x1080 resolution .

Link to post
Share on other sites
  • Moderators
I'm not going to begin to say I understand the theories that some people are putting forward, all I'll say is that it probably holds true for digital cameras.

However the vast majority of Hollywood films made in the present day are shot on film. Film is made of emulsion which has silver halide crystals inside, the upside of it being organic and not digital is each crystal acts independently from each other. When light hits each crystal the silver ions build up a collection of uncharged atoms which forms the basis of the latent image. Again I'll say that each crystal acts independently from each other.

Could someone now explain how Nyquist relates to this organic phenomenon.

I studied film not science and this discussion has gone well beyond the realms of what people in the industry discuss, it reads more like a SMPTE white paper then a discussion on technique.

great to see another post of insight from you hosko, and i'd certainly encourage anyone else who has studied movie making with some knowledge and experiece on the matter to contrbute.

ha ha now perhaps no that would be fun to see owen & mlxs smpte paper on how mtf on some digital still slrs relate to movie camera capabilties !

Link to post
Share on other sites
www.digitalpraxis.net/zippdf/scene-to-screen.pdf

Thanks for that link, Owen. I found the discussion on low contrast mid-brightness level capture fascinating. Amateur video cameras seem to want to maximise contrast the whole time, which reminds me of aggressive compresssion of audio levels.

Actually I was wrong film does have spatial frequency response.

Yes. The difficulty is how to encode it as 1920x1080 sample points that are smooth and film-like without losing information. I think the only real answer is to use a denser sampling grid to bring all available infomation to the fore. That will happen in future years. For now we will have to be content with 1920x1080, which is a huge improvement over 720x576. :)

Having said this, a lot of older movies may not have that much extra infromation to deliver that cannot be fitted in a 1920x1080 'conduit' with the required levels of anti-aliasing filtering. And modern movies may cut corners during the editing or special effects phases, or try to capture a larger depth of view than a prime lens supports, or try to capture scenes with a very small or a very large aperture setting.

It may be complaints by videophiles that will result in film production being undertaken at high technical standards, partly with an eye to the future and revenues from licensing pristine versions for transfer to high pixel density consumer formats.

you guys have not provided any evidence to show mtf in filming movie blockbusters reduces res to below 1920x1080 for hi-def disc formats.

If the penny hasn't dropped for you by now Alebonau, I suspect it never will. Never mind. The point Owen was making was a technical point about MTF particularly at around the Nyquist sampling limitation frequency. He stated it is not 100%, and he is right. There may not be much point in repeating our arguments at this stage.

... ha ha now perhaps no that would be fun to see owen & mlxs smpte paper on how mtf on some digital still slrs relate to movie camera capabilties !

Mmmn, perhaps not. :blink:

I appreciate the efforts people have made in contributing to this thread, so far. We have been able to present our views in a calm environment. I have learnt from researching some of the issues. (There is some highly technical material on sampling theory and the human eye that I must admit I found quite fascinating, even if only being able to skim it at a superficial level.)

Link to post
Share on other sites

you do realize 1920x1080 actually refers to the number of lines of resolution and has nothing to do with the number of pixels - which your whole argument is based on (well that and technology that is completely irrelevant)?

Once again MTF is *only* relevant where optics are used to capture an image (and it is completely irrelevant when converting an image that has the resolution of a motion picture to the very small format of 1920x1080) and as such has ZERO bering on a CGI film. Case closed. Your premise that 1920x1080 cannot be resolved is wrong and plain BS.

Link to post
Share on other sites
I wondering what agenda you and owen actually have to be spreading such mis-information?

There is no agenda on my part, momaw. If you do not understand, or agree with the subject matter, so be it. Judging from your string of posts you are not [yet] on the wavelength. Others are (such as Brisvegas1).

It cannot be exepected that everyone will agree.

Edited by MLXXX
Link to post
Share on other sites
  • Moderators
you do realize 1920x1080 actually refers to the number of lines of resolution and has nothing to do with the number of pixels - which your whole argument is based on (well that and technology that is completely irrelevant)?

I wondering what agenda you and owen actually have to be spreading such mis-information?

yep I wonder too.. especially since it is based on no factual information, just their own opinions and theories, never mind atleast we know in this case.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    No registered users viewing this page.





×
×
  • Create New...