DVD optimal size

[dplaton]

The DVD size (let's say PAL) is 720x576
but the displayed TV size is around 650x530
so first you loose a lot of pixel,
and second you magnify an already poor analog capture.

I wonder what solution it would be better:
cap at 640x480 ?
cap at 768x576, resize and add borders?

[The Geek]

The DVD size (let's say PAL) is 720x576
but the displayed TV size is around 650x530

That's about right. The visible area is less than the full resolution.

so first you loose a lot of pixel

Well, they're just not displayed, if that is what you mean.

and second you magnify an already poor analog capture.

What ? Nope, that's wrong.

I wonder what solution it would be better:
cap at 640x480 ?
cap at 768x576, resize and add borders?

Neither. Capture full resolution, and encode it to 720x576.
Resizing and adding borders is btw the worst you can do. You lose alot of pixels by resizing (as you make it smaller than it really is), and therefore noticeably lower the quality.

The Geek

[dplaton]

Yes I make the picture smaller but I can see the whole picture,
and a smller picture has a better resolution.

Why make a larger picture with poor resolution and a significant
part of it is not showen on TV screen?

I want to see not to have hiden pixels.

[Video Dude]

I agree with The Geek.

When you watch cable or broadcast TV there are hidden edges that you don't see, they are broadcast that way. So if you capture full resolution and encode at 720x576 you will mantain the original format. It is not until people start capturing that they even realize that there are hidden edges.

Another thing is if you resize and add borders and then view it on a LCD monitor or monitor with no overscan, you will now be stuck with a black picture frame around your picture. Think about the future.

Also, if you are capturing interlace materal, then you should never really resize if you want to keep it interlaced. You would leave it interlaced for viewing on TV. Resizing interlaced material can mess up motion and produce artifacts.

[The Geek]

Yes I make the picture smaller but I can see the whole picture,
and a smller picture has a better resolution.

Wrong !

You can't see the whole picture if you make it smaller. Otherwise you couldn't make it smaller. The TV has 576 lines vertical, and when you resize them to 480 or whatever you wanted them to resize, you lose the difference. So, less quality.

So, a smaller picture does *not* have a better resolution. It has less resolution, and less does not mean good. A higher resolution is better, not a lower one.

Why make a larger picture with poor resolution and a significant
part of it is not showen on TV screen?

Because you're wrong. A larger picture has a better resolution. And the significant part of it *is* shown on the TV screen. Only a bit of the borders are invisible.

The Geek

[Boulder]

Originally posted by Video Dude
Also, if you are capturing interlace materal, then you should never really resize if you want to keep it interlaced. You would leave it interlaced for viewing on TV. Resizing interlaced material can mess up motion and produce artifacts.

Depends on how you do it.

Besides, using a 16-pixel overscan area increases compressibility big time, which is a very useful thing especially with interlaced encodes.

When you say that you'll suffer from having black borders around your image on displays that have no overscan, think about the fact that almost every DVD has the borders as well;)

[dplaton]

1. An interlaced material can be resized without damage if you
resize each field and put the fields back again.

2. The black border problem is not a problem even on a PC you can
feed the monitor via a frameserver.

3. Think about the future, in short time the DVD will be obselote
this is the realy big problem.

4. Yes the TV has 576 lines but the camera capture hasn't this is the problem.
VHS, S-VHS, miniDV, etc. emulate a TV picture
A cassette played on the native device looks better than a <digitised> one.
Why?
Of course, has a lot more information, but, I think, there is some trick in emulating the TV picture.
If this is true I want to find the trick.

If VHS ..... has 240 <lines> it will be a big difference in outputing
the 240 <lines> on 576 lines or on 500 lines (plus some 50 lines of 576 lines are <invisible>).

I will draw some sqares, I will record on camera, and I will out-put
on TV to see exactly what is hapening.

Thanks all

[Boulder]

Regardless of your source (VHS or antenna, cable, whatever), you should always capture at the vertical resolution 480(NTSC) or 576(PAL). With VHS, the true horizontal resolution is apparently somewhere between 352 and 480 pixels.

My capture card (PVR-250) captures at 720x576 and resizes internally to 352 or 480x576 if I choose those lower resolutions, so I prefer capping at 720x576, whatever the source is.

[Malcolm]

I agree with the Geek + Boulder.

Btw. 704x576 (480) is also a valid DVD resolution. You have to crop(!) your captured source from 720 to 704. this way you can 'delete' some of the unshown content on the right + left.

the following is from a post i've writte long time ago :)
encoded width automatic stretching-factor when displayed at 4:3 (~704)
720 -> x1 =720 -8 left, -8 right =704
704 -> x1 =704
544 -> x???
528 -> x1.333 =704
480 -> x1.5 =720 -8 left, -8 right =704
352 -> x2 =704

greetings,
Malcolm

[dplaton]

OK Capturing at the max resolution is best.
What I want to know if I resize down to the visible image
what is to gain and what is to loose?

[Wilbert]

4. Yes the TV has 576 lines but the camera capture hasn't this is the problem.
VHS, S-VHS, miniDV, etc. emulate a TV picture
A cassette played on the native device looks better than a <digitised> one.
Why?
Of course, has a lot more information, but, I think, there is some trick in emulating the TV picture.
If this is true I want to find the trick.

If VHS ..... has 240 <lines> it will be a big difference in outputing the 240 <lines> on 576 lines or on 500 lines (plus some 50 lines of 576 lines are <invisible> ).
VHS doesn't have 240 lines in the vertical direction (this 240 is the lines of horizontal resolution; see analog capture guide). If you record (pal) a test image with 576 lines in the vertical direction you will see that all of them are recorded and visible.

[dplaton]

I don't want to crop, because at the full size on TV is automaticaly
croped. This is what I want to avoid, cropping.
If I cap at full resolution and resize down I will improve the <vizual> rezolution?

[Boulder]

You could try capturing at full resolution and then resizing to 352x576 if you have a VHS source. You probably won't see any difference to a 720/704x576 encode, especially if you use LanczosResize to downsize.

With TV captures, I'd go for 704x576.

[Malcolm]

no

[The Geek]

If I cap at full resolution and resize down I will improve the <vizual> rezolution?

You don't get it, do you ? You can't improve resolution by downsizing it. How you do think that could be possible ? Downsizing means you reduce the resolution. That is what it means.
And the visual resolution is always the same, it doesn't change. All you seem to do by downsizing is to fit the video in the lower resolution. Of course that means quality loss, as the video originally was in higher resolution.

The Geek

[dplaton]

What is in my mind, and is somehow a missrepresentation:

a photographic image projected on a larger screen will look poorer than projected on a smaller screen.

The image has a constant resolution but when magnified the resolution gets low.

I wnat to project the image on a smaller screen in order to obtain a visual effect that the picture has a better resolution.

But I understand that resizing will not improve the vizual aspect.
It's that correct?

It is possible to achieve a better aspect (more information on surface unit) and also the whole picture?

[Boulder]

I'd say that the final quality depends on the resolution the source initially has. VHS doesn't contain information to represent a full PAL image, it contains much less than that. If you capture VHS at 720x576, you actually use the existing information to create new data to cover the capture area.

This is why I recommend trying capping at 720x576 and then going back to 352x576. It's very likely that if you capture at 352x576, your capture card will do the same my card does, and that's not nice. It's better to do the resizing manually, that way you can be sure which resizer is being used. I have a feeling that capture cards use bilinear which softens the image quite a bit.

[dplaton]

Thanks, I'm not concerned about 720, 704 size but 576 that seems to me too much for VHS (S-VHS, miniDV, etc).

I'm looking for a simili 4:3 to 16:9 transformation that don't distort the image and use the number of lines of the original source, in order
to obtain a better vizual result.

[The Geek]

a photographic image projected on a larger screen will look poorer than projected on a smaller screen.

A photographic image with a given resolution, yes. For example, a 352x288 picture on a 15" monitor in full screen looks better than the same 352x288 picture on a 20" monitor in full screen.

The image has a constant resolution but when magnified the resolution gets low.

Yes. That is coz the pixels get bigger when you magnify, and at some point you can see that.
But when magnifying, you keep the entire picture, you keep all the pixels.

I wnat to project the image on a smaller screen in order to obtain a visual effect that the picture has a better resolution.

That is the other way , but doesn't really work the same way. If you have a picture of 720x576 and a visible area of let's say 640x480, and want to resize the picture to fit the smaller area, then you lose 96 rows vertically and 80 columns horizontally of video information.
The picture can't look better as it's still 640x480, the pixels are still the same size, and the 640x480 don't hold the same amount of information as the 720x576. Some video information is lost. All you can achieve by resizing is to make the borders visible that normally are in the overscan area.

But I understand that resizing will not improve the vizual aspect.
It's that correct?

Correct. It will actually degrade it.

It is possible to achieve a better aspect (more information on surface unit) and also the whole picture?

No. You already have as much information on the surface as there can be. More than that is not possible. And by resizing the picture you still have the same amount of information on the surface, but not the whole picture anymore.

The way it is, is the way it is supposed to be. Always has been.

Thanks, I'm not concerned about 720, 704 size but 576 that seems to me too much for VHS (S-VHS, miniDV, etc).

Well, you're wrong. 576, that is exactly what VHS, S-VHS, miniDV etc. are. Any row less than that means quality loss.
Horizontally VHS has about 240 lines, so 720, that is way more than that. 352x576 would do it, too.

[dplaton]

Let's make a simple experiment:
open a given avi in a player not in full screen mode,
resize the player window, the avi resolution doesn't change,
but the <visual> resolution change, the smaller the picture the better image.

Let's use the zoom function that is available on some TV set
(only the vertical size is changing -> distortion)
and a 4:3 transmision is <resized> to 16:9
again the <visual> resolution change.

So what is doing the resize function in a video editor (AviSynth, VirtualDub)?
Why a smaller image full displayed on a TV set
is worst
than a bigger image partialy displayed on a TV set?

Again I'm speaking about visual not intrinsec resolution.

[Malcolm]

o.k. i'm unsubscribing from this thread now. ;)

greetings,
Malcolm

[The Geek]

Let's make a simple experiment:
open a given avi in a player not in full screen mode,
resize the player window, the avi resolution doesn't change,
but the <visual> resolution change, the smaller the picture the better image.

You are still wrong and don't get it, and your simple experiment failed anyway.

I have 1024x768 on my monitor, and it will display 1024x768 on the whole monitor at all times. So, it is technically not possible do downsize the player window while maintaining the same resolution as the avi. I just used a DVD for it, and it doesn't look better at lower resolutions. The more, the better, and at its native resolution it looks best.
Now you can make a simple experiment. Make a screenshot of the original sized AVI and make one of your downsized "according-to-you-with-the-same-resolution"-AVI and load them both in photoshop or paint or whatever and look for yourself. The resolution does change. And less resolution, less quality.

Let's use the zoom function that is available on some TV set
(only the vertical size is changing -> distortion)
and a 4:3 transmision is <resized> to 16:9
again the <visual> resolution change.

What do you mean ? Is this about how anamorphic 16:9 will be displayed on a 4:3 TV ?

So what is doing the resize function in a video editor (AviSynth, VirtualDub)?
Why a smaller image full displayed on a TV set
is worst
than a bigger image partialy displayed on a TV set?

Because the visible area is always the same. The resolution that you see is the same, no matter if you have something in the overscan area or not. And the smaller image, "fully displayed", doesn't fully display the original picture, as it got resized.

Again I'm speaking about visual not intrinsec resolution.

I think I don't get your visual resolution thing at all.

The Geek

[theReal]

I have another important reason why you don't want to see the invisible underscan area (at least for tv productions, not for movies intended for cinema):

Every cameraman knows that some part of the picture will not be shown on tv sets - that's why he will fit his shots into the tv safe area. By watching the material in underscan (=with the usually invisible area) you destroy the picture composition (too much headroom, maybe unwanted stuff visible right/left of the normally visible area.
Bad comparison:
It's like eating the disgusting crust of a cheese only because it's there --> cut it off and throw it away so you can enjoy the delicious cheese as it was intended ;) :D

[Sergei_Esenin]

dplaton, you have no idea what you're talking about. You think you're talking about video source resolution, but you aren't. You're really talking about display resolution and "display density"--the apparent resolution of display devices, which is in the same vein as the dpi resolution of a printer.

Higher source resolution is always better, period, in every circumstance. This is the fundamental truth which makes DVD more detailed than VHS, HDTV more detailed than DVD, 35mm film more detailed than HDTV, etc. If you continue to argue otherwise, even after the explanations given above and the example I'll give below, then you'll be displaying either an intractable stubbornness or a profound ignorance of some very basic concepts. If this is so then you just can't be taught or are unwilling to learn, so please try to understand this.

Here's my example. I view TV, DVDs, and my PC screen on an Infocus X1 projector. It works like a movie theater projector--it's placed at the back of the room towards the ceiling and projects an image onto a screen in front. It has a resolution of 800x600, so anything with a smaller or larger resolution is scaled to 800x600 and there are always that many pixels projected on the screen.

Now, here's where your mistake comes in. I view the 800x600 image on a 100" screen, but since this is a front projector I could easily focus it to 50" or 150". *It would still show 800x600 pixels resolution* whether displayed at 50" or 150". What would change is the size of each pixel and thus the apparent density of details--at 150" each pixel would be big and blocky and the picture wouldn't look very watchable; it would look like there's not much detail at all. That's why I watch at 100", because each pixel is much smaller and the apparent density of pixels and level of detail looks much better--and yet the resolution is still the same, 800x600. Likewise, if I weren't satisfied with the apparent detail I get when projecting at 100", I could refocus the projector to produce a smaller 50" image. Each pixel would be smaller, so apparent image density and apparent detail would increase--and yet the resolution would *still* be the same, 800x600.

Likewise, if I wanted to maintain a 100" screen but increase the level of detail, I could get a higher resolution projector like 1024x768 because a higher resolution at the same screen size means smaller pixels, higher pixel density, and better detail. By your logic however I'd want a lower resolution projector, like 320x240 (yuck)--which clearly isn't the case, because each pixel would be gigantic, and pixel density and the level of detail would be nonexistent.

Now, all of that assumes a fixed, identical viewing distance. The reason is because changing viewing distance also changes apparent pixel density and thus apparent level of detail. That's why a TV looks good and detailed from a few feet away, but looks awful and splotchy up close.

What you've been talking about has nothing at all to do with source resolution. Higher source resolution is always better and more detailed. Display size and pixel density are completely different issues.

[jggimi]

I'm going to jump in at this point and add a comment, which I think is necessary in order to help guide this discussion.This discussion has been skirting the edge of Rule 4 (http://forum.doom9.org/forum-rules.htm). Let's not call each other stupid or wrong, even if we don't "get it." OK?You've already pushed Malcom away. This does not foster the exchange of ideas.

I'm not picking on any one individual. Everyone please, just be nice, people.

Thank you.

[ronnylov]

Sometimes especially when encoding to lower bitrates it may be a good thing to remove the picture from the overscan area because you just can't see it anyway on your TV. But you don't must resize, you can crop and then replace the cropped are with new black borders. To do this effeciently you should crop multiples of 8 or 16 pixels, because mpeg encoding use 8x8 blocks and 16x16 macroblocks.

For example: Your source is 720x576 full screen. You can crop 16 pixels at top and bottom and 24 pixels to left and right and you get a 672x544 pixels picture, then add 16 pixels borders on all sides and you have a 704x576 picture with black borders hidden in the TV overscan are. 704x576 is a valid PAL DVD resolution and you get better picture quality at the same bitrate and you can't see the border when watching on a TV.

If you want to resize and add borders this is also possible but you may loose some picture details (depending on the source). Then the picture will look slightly outzoomed compared to what you originally would see on a TV. When I want to do this I use FitCD to do the calculations. FitCD homepage: http://shh.sysh.de

[dplaton]

Maybe I'm stupid in puting the problem, let's try again:

1.I have a source resolution let's say 640x480;
2.Capturing at 640x480 is considered improper;
3.Capturing at 768x576 it's OK for color reasons;
4.There is a degradation regarding clearness;
5.How to make the image clearer?
I consider 768(720)x576 a fake resolution and
I think that making the image smaller it will be clearer.
I treated 720x576 as a size not as a resolution,
(an 640x480 enlarged image)
so I considered that resizing will give me back a smaller picture
but a clearer one.

The simple question is about to capture and display proper a camcorder
analog capture.
I prefer a clearer image to one that fill the whole screen.

[Wilbert]

The simple question is about to capture and display proper a camcorder analog capture.
Explain why you think that the resolution of an analogue camcorder is 640x480. Which camcorder are you referring to (8mm, VHS-C, Hi-8, etc.)?

[trevlac]

@dplaton

Ya know I never liked pixel measurements (ie 640x480) as a statement of resolution.

Problem is that 640x480 is a frame size on a common display device, a PC monitor, but people like to use it to talk about resolution.

But resolution really depends upon how much space is in between each of those darn pixels when you 1st convert from analog . And! how much fine detail was in the analog. Hopefully each spec of detail is bigger than at least 2 pixels wide. If not, you lost it for sure.

Now if you downsize past each spec being bigger than 2 pixels wide, you are loosing more resolution. If you go lower but not too low, you (theoretically) loose nothing. Going hi (theoretically) you loose nothing. In practice, the resize (resample) methods do change the details and you may loose some by resizing (up or down).

The sizes you are talking about (for an analog camcorder) a probably not too low for the specs of detail.

So generally, a change in the frame size does not change the "true" resolution at all.

Preceived Resolution - A term I am making up. At a real big size and CLOSE UP a picture will no doubt look worse. Sit close to the TV to see. Your eyes can and try to pick up details that are not there. Sit farther back and you don't have a problem. Same goes for smaller images up close. :)

An interesting note about this and HDTV. I believe ... To really get the biggest bang, you need a higher resolution source/monitor AND it has to be bigger than your current set (or you need to sit closer) to get big results. Trouble is big high def == $$$$

PS: I don't have a big (or small) HDTV, so I may be talking out of my butt on this last point. :)

[dplaton]

Thank you all

[theReal]

I consider 768(720)x576 a fake resolution It's not fake at all, it's broadcast standard...
If you capture less lines then you'll lose even more of the picture information.

[dplaton]

First 720x576 is not a resolution by it self is a resolution for a given screen size. Display the same picture at the 640x480 "resolution" on a 13" screen and on a 21" screen and keep the watching distance the same.

Resolution is density like printer's dpi.

The source has a given resolution let's say 600 dpi (it is just an example).
If you set the capture resolution at 800 dpi dosen't meant that the
captured avi will have a 800 dpi - that is the fake resolution - no, you will get a 600 dpi put on 800 dpi => a poorer image. The density of information doesn't increase.

I wrongly suposed that resizing down will restore the original resolution (density) and the image will look better.

There are two different items:
1 is the photographic image quality and that was my target;
2 is the TV transmition requirements requirements.

Again my ideea was that resizing down I will restore the real/original
resolution. But resizing reduce the size and doesn't increase the resolution.
In the optical world resizing down can restore the resolution but in the digital world NO. This seems to be the conclusion.

This thead is originated on a Trevlac thread on size.

[peterhuh]

I think i can understand what you mean, dplaton.

You refer resolution as density. Then it would be true we will get higher quality, if we resize a bigger image into a smaller one, as this will increase the density while maintaining the same resolution. (eg. changing a jpeg image from 96 to 120 dpi)

But when we talk about video, what most of us mean when we use the word 'resize' is 'resample'. Now, its a fact that when we resample a larger image into a smaller one, we will result in some loss of quality. (eg. resample a 768x576 video into 320x240).

Also, your simple experiment to prove that resizing a large video into a smaller one will result in higher quality is seriously flawed.

Firstly, unlike a printer, neither our TV nor the computer screen will change it density even if we resize our media player. Thus, with a constant pixel density on your screen, any down sampling of an image will result in loss of quality.

Secondly, I agree with you, when you say 'clearer'. Most of the time, when i resize my PowerDVD player smaller, i can percive a better image quality. This is only true if we use a natural image. Because most DVD transfers and broadcast are done in analog. Even with the best analog eqipment, some loss is inevitable. Thus, if are able to get a 'perfect' image, any down sampling will result noticible degration. A 'perfect' image can be obtained if we create digitally.

A very good example, is to do a Print Screen of doom9's forum with all the text in it, and paste the image in a photo editor like photoshop. If we view it at 100%(Actual Pixel), we can see all the text crisp and sharp. But if we to down sample the image, the text will become blur or unreadable. Similar comparison can also be done if we do a cap on video games such as Half-Life 2.

You are also right, but poorly expressed, when you use the pharse 'fake resolution'. In my opinion, what you mean is an 'up-sampled resolution'. (eg. transcoding a VCD into a DVD) I don't see the reason of doing that. But I undestand that you are trying to get back the original resolution, in this case the VCD resolution.

But the problem is, we can't tell if our capturing card is able to capture video that conforms to CCIR 601 standerd, or some clever up-sampling done by the capturing card's driver, just by looking at the video. I also suspect that my Huappuage WinTV card captures at a lower resolution, then auto-magically resamples to a higher resolution that conforms to 601 standerd. So the only way to retain all detail is to capture everything, that is full resolution.

Finally, before I summaries and answer your first question, let's see what are the important points of the methods we have dicussed here.

1) Cap at 768(720)x576:-
-No futher loss of detail
-Conforms to CCIR 601
-Direct transfer to DVD w/o further resampling
-As director's intention for viewers
-No underscan for full resolution TVs such as Plasma

2) Cap at 768(720)x576 + Reseize + Add black borders:-
-See everything on most TVs
-Underscan for full resolution TVs

3) Crop only:-
-Uses less bit to compress

4) Cap at 640x480:-
-Uses less bit to compress
-Good to view on computers
-Up smampling required for tranfer to DVDs

As you can see, cap at 768x576, then resize and add black borders is a solution only for most TVs. There is a chinese saying "zhi biao bu zhi ben". Which simply means you are finding a solution that fixes a current problem, but not the root of the problem. The problem you are having now is, that you want to see everything on your TV.

So, in my very humble opinion, you should adjust your TV instead. I believe almost all CRT TVs has a service man mode where one can adjust the image to their liking. This solution will not only fixes any video you will be displaying on your CRT TV, the video resolution will also be keep within the CCIR 601 standerd.

I hope you will accept my opinion. :)

[theReal]

Firstly, unlike a printer, neither our TV nor the computer screen will change it density even if we resize our media player. Thus, with a constant pixel density on your screen, any down sampling of an image will result in loss of quality.

That's exactly the point!

Also if you think that i.e. VHS doesn't display the full information - the picture that comes out of the VHS machine is "blown up" to the full amount of lines/pixels . Unless you have a method to exactly reverse that blow-up you will lose some of the real information as well in a resize (and there is no way to exactly reverse an analog blow-up).

It's just like if you take a divx video, convert it to uncompressed video and re-encode it to divx again you'll get worse quality in the end!

[Wilbert]

I also suspect that my Huappuage WinTV card captures at a lower resolution, then auto-magically resamples to a higher resolution that conforms to 601 standerd.
It's the other way around. See analog capture guide (chapter 5).

I don't see the reason of doing that. But I undestand that you are trying to get back the original resolution, in this case the VCD resolution.

dplaton seems to think the resolution of his source is 640x480. But he never bothered to answer my previous post about that.

[dplaton]

Dear Wilbert,


Explain why you think that the resolution of an analogue camcorder is 640x480. Which camcorder are you referring to (8mm, VHS-C, Hi-8, etc.)?


I don't think that the camcorder (JVC GRX 25 - SVHS PAL with digital signal processing) has a 640x480 resolution. This resolution is from a Trevlac post and refers to NTSC.

The question was derived from my amateur photographic experience.
Opticaly when you resize you increase the density.

The confusion stands that 720x576 is not a resolution and resizing is not increasing the density but is a resample.

The words used like resolution and resizing were not what they are.
Resolution=density
Resizing=same amount of information put on a smaller size.

Thanks you for all the answers

[dplaton]

Once again.


Firstly, unlike a printer, neither our TV nor the computer screen will change it density even if we resize our media player. Thus, with a constant pixel density on your screen, any down sampling of an image will result in loss of quality.


It is not about the screen pixel density but about the clip density, the film resolution like ASA 100 ASA 400 or DIN 19 DIN 27.

Capturing at higher "resolution" is in part equivalent to digital zooming, and that is bad, the higher "resolution" the worst.

So IMO it is better to capture at lower "resolution" 704x576 for PAL and to crop in order to


get a 672x544 pixels picture, then add 16 pixels borders on all sides and you have a 704x576 picture with black borders hidden in the TV overscan are. 704x576 is a valid PAL DVD resolution and you get better picture quality at the same bitrate and you can't see the border when watching on a TV.



My strugle was for getting a better picture, a picture with the original resolution - real resolution (density) not a fake resolution (a digitaly zoomed one).

And as the next step to add borders in order to complain with the DVD/TV "resolution".

I wonder if capturing in RGB 32 at <<640x480>> could be a better solution.

[trevlac]

Originally posted by dplaton

Capturing at higher "resolution" is in part equivalent to digital zooming, and that is bad, the higher "resolution" the worst.

So IMO it is better to capture at lower "resolution" 704x576 for PAL and to crop in order to


Hi again ....

This is close, but not quite true ... so I thought I'd add some info :)

When you digitize an analog signal, capturing more pixels from a given signal length is better and is not like a digital zoom.

The secret is that even though capture devices seem to give you a choice as to the 'pixel density' you can use to capture a given width ... they always capture at a fixed rate (usually 2x higher than DVD spec) ... and then size DOWN. Always.

So it is not a zoom because they are not going up in size.

My opinion is that it is better to limit the number of resizes (although don't kill youself over it), because any resize mucks with the image. If you cap near your output size, and pad or crop, that would be best.

--------------

BTW: If you are capturing a PAL signal and going to watch it on a regular TV ... you should stick with 576.

Reason is that the odd lines are not from the same point in time as the even lines ... so going from 576 to 480 a resizer would need to adjust for distance and time ... hard to do. THEN when you watch it on a PAL TV some other device would have to do the reverse.

[dplaton]

Some considerations

We have 2 kind of resolution

1. Absolute resolution which is density, a phisical one like the dpi for printer (photographic paper, film granulation, screen dpi, etc.)

2. Relative resolution like 720x576 or 360x288, the phisical caracteristics and the size of the device aren't change but the image is displayed using 720 colums and 576 rows or 360 columns and 288 rows.

That's correct?

Capturing at higher relative resolution it's better for sample consideration.

What is hard to tell is what is the absolute resolution recorded by the camera.

Idealy someone must to capture the recorded image at the absolute resolution and to ensure the corect sample.

Capturing at a relative resolution higher than the absolute resolution will lead to a digitaly procesed image where some line and columns are created by interpolation.
Let's take an extreme case a recorded image that has for a 15" screen 360 columns and 288 rows (that means an absolute resolution dots per sqare inch).
Capturing this recorded image at a relative resolution 720x576 and displayed on a 15" screen that dosen't mean a better image because half of the vertical and half of the horizontal lines are created digitaly by computations, and it seems that there isn't an apropriate algoritm which can create a realistic image. Even the broadcasting companies present poor VHS digitalised transmitions.

Can be done a reconciliation between absolute resolution and sampling?

So far it seems that the answer is No.

[Wilbert]

What is hard to tell is what is the absolute resolution recorded by the camera.
It's SVHS which has a bandwidth of ~ 5 MHz, which "equates" to a minimum amount of 520 pixels (horizontally).

Idealy someone must to capture the recorded image at the absolute resolution and to ensure the corect sample.
In an ideal world one should cap minimal 520x576. In practice, it's not possible to ensure "the correct" sample, whatever cap size you use.

Let's take an extreme case a recorded image that has for a 15" screen 360 columns and 288 rows (that means an absolute resolution dots per sqare inch).
Your source is analog, so it's a bit more difficult than this.