If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
My capture card is an ATI All In Wonder 128, and the settings only allow me to capture at 704x576. This is no good to me because I want to compress my movies as mpeg2 and put them on DVD.
If, in fact, your authoring software won't accept (or convert) the resolution, you can load the file into TMPGEnc and change the resolution to 720x576...
I could just have thin black borders at either side, which won't show; but this causes a problem: the image will now be distorted, so people will look slightly thinner than they should be. It also means the image is slightly lower res. I don't feel I should have to do this, as I can capture mpeg2 at 720 pixels so why not avi?
You say that scaling up from 704 to 720 will not affect the quality much, but actually you're totally wrong. Scaling an image up or down slightly causes a significant drop in quality. I'll try and explain...
When you capture video, your capture card is seeing the original signal and converting that to a grid of dots. Now, assuming your original signal is a good quality one, nice and sharp, you will get a nice crips image with each pixel given a unique colour, so you will have sharp edges and a lot of detail. Now, if you scale this image up or down, even slightly, then every single pixel has to be re-calculated in order to achieve the appearance of the same image using a different number of pixels. Now each pixel is no longer its original pure colour, it is a mixture of two pixels' colours. Consequentially, edges are not as sharp and the whole image looks blurry. This can be "compensated" with sharpening filters, but basically you will never have the original "1 screen pixel" = "1 image pixel" ratio again.
In case you're interested, I'll illustrate how even a tiny scale loses a lot of quality... Consider an video image 704 pixels wide. Imagine it's one of those scenes with a split screen and you have two people talking on the telephone. One person is in a payphone outside and it's dark, the other is indoors and it's light. As there's an even number of lines in the image, there will be a very crisp line down the middle of the screen, because the pixel column to the left of the center is light, whereas the pixel column to the right of the center is dark. Now let's say we scale it up to an image 705 pixels wide. You might not think that would affect the image quality much as it is only a tiny amount larger, but consider this: one whole pixel column has to be added to the image. Imagine this extra column is added in the middle of the image, so now the two columns that used to be the middle two columns are each "half" in the new middle column and "half" in the ones next to it. So the new middle column will be a mixture of the light and dark columns, making it mid-grey. So the line won't be as crips now. It's not just the center of the screen where this "blurring" occurs, every pixel on the screen will be a mixture of two of the old screen pixels, it will be less noticeable near the edges, but more noticeable in the middle, but the whole screen will lose its crispness.
This effect is very clear to see if you use computer graphics. Say you take a screen grab of a website (lots of thin, crips lines) and scale it up (or down) by just a few pixels. You will notice that the whole thing looks a lot more blurred than the original, as there can no longer be crisp lines. The only exception to this blurring effect is when you either scale down a hell of a lot, so there is sufficient information to give each of the new pixels its own unique colour; or when you scale (up or down) by a whole factor, for example when you double or half the number of pixels.
Sorry to rant, but I've been working with computer graphics for a long time and I know what I'm talking about! Hopefully someone out there will find all this as fascinating as I do!
As impressive as your personal knowledge may be, I'd suggest that, just for the sake of experimentation, you try my original suggestion of allowing TMPGEnc to convert your video to 720x576. You just may be pleasantly surprised to discover that the increase of horizontal resolution by approximately 1/5 of 1% in real life, doesn't noticeably distort the image. After all, it's only a video - and you'll never REALLY know until you've tried ;>}
Setarip is right, you really should just give it a go and check it out.
There are two things in your book above I can't get my head around:
How will inserting black borders make the picture look quished??
When I encode a wide screen DivX to say Mpeg 1 VCD and want to keep the wide screen format, I let TMPGenc insert black borders top and bottom. Now my picture does not look squashed top to bottom, it just looks normal. This is because the black borders just fill up the space to keep the correct Aspect Ratio on a 4:3 Aspect Ratio Screen/TV etc, like a DVD does when letter box is selected on a std TV.
The other thing I don't get is if your so experienced at all this. Why are you still using an AIW128, don't get me wrong it's a great card and I had a few in my time. I just thought you would have used a better model by now.
Anyway here's a page that clearly shows 720x576 with the ATi AIW 128 http://www.cvs.dk/aiw.htm maybe this will help.
setarip>> The increase of resolution isn't 1/5th of 1%, I worked it out as just over 2%. Aslo, you say that scaling the resolution won't "distort" the image. By "distort" do you mean stretch or lose quality? I never said I thought the image would stretch. I COULD try it, but whether it looks okay or not, I am a perfectionist, and I know that scaling an image reduces its quality. I'd probably never be able to sleep at night knowing all my videos are losing quality! I'll give it a go though. (Frustrated that I can't just capture at 720!)
Ndb>> Love this smilie: !
You ask: "How will inserting black borders make the picture look quished?? "
Allow me to explain. You're confusing aspect ratio with resolution. But aspect ratio and resolution are two independent factors. In your example, you're taking a 16:9 video and putting borders above and below it to make it look the same on a 4:3 screen. This won't distort the image because the borders are there specifically to compensate for the difference in aspect ratio. But in my situation, I am not changing aspect ratio, (I will be capturing both 4:3 and 16:9 videos using my setup, so the aspect ratio has no bearing on what we're talking about). It's only the resolution that will be changed in the scaling process, and resolution has no bearing on the shape/size of the image. Putting black borders either side of the screen will distort the image. To clarify this, imagine I capture my video at 360 pixels across. I then add borders either side to get it up to 720. But when these borders take up half the screen, you can imagine that the image will be very clearly distorted.
If you still don't understand tell me and I'll try and explain it some other way.
Your other question is, if I'm so experienced at this, why am I still using an ATI AIW 128? Well, firstly I'm not in any way experienced at this. I've been working with computer graphics for a long time, but I've only just got my first capture card recently. I'm totally new to digital video. (I only just started earning money recently so at last I'm finally able to buy the stuff I want!). Secondly, the only thing I wanted in a graphics card is good video capture support. One that can capture well, and output well. That's why I chose ATI, I was told that ATI was the best there is. For years I planned to get a Matrox Marvel, which some people told me was the best, but when I researched it and asked people who had used both, I discovered that ATI was the best choice (or so I thought!). I have no idea if there's a better card out there, but I'm very pleased with my ATI card. Could you tell me what card(s) are better, and why? Thanks.
The link you gave made no mention of capturing avi at 720x576, only mpeg2. I don't want to capture mpeg2 for 3 reasons:
1) First and foremost, I only want to encode with the best mpeg2 encoder - TMPEGenc. My ATI card's mpeg2 compression is not that good quality.
2) I want to be able to edit my movies in Premiere (specifically to get the sound, for music performances etc). Premiere won't import mp2 files I capture with my card.
3) If I capture as mpeg2 files, I can't edit them at all, or I'll have to re-compress them as mpeg2s afterwards, re-compressing anything reduces the quality. Pretty much everything I record I want to "trim" the beginning and ends off, so the recording is neater. Especially in the case of timered recordings, which I start recording a few minutes before the programme starts and stop recording a few minutes after the programme ends.
I'm not inserting the black borders, TMPGenc does it when it encodes for Video CD (i.e. VCD or SVCD etc). You see when we encode a DVD rip etc we get a picture of 720x576, but if it's wide screen (i.e. 16:9) we get black borders top and bottom and on some left and right (depends on a-ratio). So we always crop the borders before compression. This way we only compress the actual picture and don't wastes space encoding the black borders.
Now when we play the movie full screen the black borders are automatically displayed to fill the screen. Well TMPGenc does exactly the same when you come to encode for VCD etc. The picture res stays exacly the same with no stretching or distortion in the middle of the black borders.
This is exactly the same for DVD if I encode my movie to MPEG2 and write it to DVD the aspect ratio stays perfect and black borders are added around the space that the picture does not cover.
As to the link, I did not realise you were capturing to avi, as being as you could only afford an ATi AIW 128 I did not think you could afford the multi-GigaBite HDD drives that would be needed to capture uncompressed video at this res.
We are goiunf to wear our keyboards out at this rate
You still haven't told me what capture card(s) you think are BETTER than my "cheap" ATI card. I'd love to know, as I want the best capture possible so I may think about changing my video card if you can suggest a better alternative (although as I say, I am very pleased with the results I get from my ATI card. The video looks perfect to me.)
I use a 40Gb hard drive for capturing. 40 Gb is not expensive and it gives me about 20 minutes, which is all I need. I only capture TV programmes and home movies and stuff, not full movies. (I don't actually understand why people capture full movies from tape when they could go out and buy a perfect digital copy on DVD, or just rent one from a library and copy it.)
As you know, I'm new to this. I've never actually authored a DVD yet. But as I understand it, the resolution of DVD video is 720x576. This resolution is used for all aspect ratios. The disc also contains information that tells the player what the aspect ratio is. Presumably when I author DVDs I will specify what aspect ratio I want each video to be.
I have to admit your last post totally confused me and I don't understand what you were actually getting at. You say "we always crop the borders before compression, this way we only compress the actual picture and don't waste space encoding the black borders". I'm totally confused about this. When you capture video, you don't capture any black borders. The borders are only put on by your TV set, they are not transmitted as part of the TV picture. For example, when I capture a widescreen TV programme, I do not get any black borders at the top and bottom. And when I capture a 4:3 programme, I do not get any borders at either side.
You still don't believe me that the image will be distorted if you add borders either side! Think of it this way: you watch a TV programme on your TV. It fills the width of the screen. You then add black borders to either side of the image and watch it back on the same TV screen. The image HAS to be distored in order to make room for the black borders.
Man, this is confusing.
Is there anyone out there who understands what both of us is talking about and can explain everything to us and who's right?
Here's where you are getting confused, see your quote below:
__________________________________________________ __
You still don't believe me that the image will be distorted if you add borders either side! Think of it this way: you watch a TV programme on your TV. It fills the width of the screen. You then add black borders to either side of the image and watch it back on the same TV screen. The image HAS to be distored in order to make room for the black borders.
__________________________________________________ __
You pretty much answered yourself there, but then slipped. Your 704 res won't fill the width of the screen, you will have a space either side, this is where the black borders would be added to make it 720x576. now if your picture was 720 wide and you added black borders then you would squish it.
A std TV is not 720x576 res anyway it's 640x256 or what ever and a wide screen is not 720x576 either it's more like 720x256. So the 576 is just picture and black borders.
No matter how you capture it, it's gonna be resized to fit the TV it's viewed on using Black Borders.
Just try a small clip and wack it through TMPGenc and see what happens. I've done loads of VCD/SVCD and DVD and it just adds borders to make the picture fit without distortion.
> Your 704 res won't fill the width of the screen, you will have a space either side, this is where the black borders would be added to make it 720x576.
For what you say to be correct, when I grab video at 704x576 my graphics card would have to crop the sides off the image. Why would it do this? I have captured at both 704 and 720, and both are the same. My graphics card doesn't crop part of the image. Why would it?
I think that this is where you've been going wrong. You've assumed that because the resolution is smaller than 720, that some of the picture imformation has been cropped off the sides. It hasn't. All the picture information is still there, it's just a lower resolution. So the 704 image still fills the screen, and putting black borders either side of it WILL squish the image.
>A std TV is not 720x576 res anyway it's 640x256 or what ever and a wide screen is not 720x576 either it's more like 720x256
Standard TVs don't have fixed resolutions. They play whatever resolution the incoming signal is. Standard terrestrial TV and satellite TV are both broadcast at different resolutions. Games consoles and computers that plug into the TV all have different resolutions. A TV can display lots of different resolutions. When you play a 720x576 DVD on it, the resolution changes to 720x576.
DVD resolution is 720x576 (for PAL), so to ceate a DVD compatable MPEG-2 video file your movie will be made to fit this size.
If it's too small in width or high (or both), you can either resize/stretch or resize/resample to get to this res and fill the screen (i.e. no borders).
Now if you encode a wide screen movie, you don't make it fill the screen or everyone would be tall and skinny. You let it insert black borders top and bottom.
So why is that any different to black borders left and right. If I stick a movie of 640x480 through TMPGenc and want it to stay 640x480 whilst keeping to 720x576 for DVD.
I end up with a 720x576 movie but with a black frame with a 640x480 movie in the middle. No distortion, reduction in quality, no streching or shrinking, just a 640x480 movie in a black frame.
So you would get a 704x576 movie with black borders of 8 left and right which is quite normal. Go checkout a DVD ripping guide and see how much we normally crop of the left and right side, surprise surprise it's 8.
If you had every ripped a DVD you would understand that they are hardly ever 720 in width without black borders.
>DVD resolution is 720x576 (for PAL), so to ceate a DVD compatable MPEG-2 video file your movie will be made to fit this size.
Right. I agree. Any PAL DVD movie has to be 720x576.
>If it's too small in width or high (or both), you can either resize/stretch or resize/resample to get to this res and fill the screen (i.e. no borders).
Right. I agree. If your image is a lower resolution than 720x576 then you need to scale it up to 720x576 to get it to fill the screen on a DVD.
Seems like we agree on everything so far.
>Now if you encode a wide screen movie, you don't make it fill the screen or everyone would be tall and skinny. You let it insert black borders top and bottom.
The people would only be tall and skinny if you took a widescreen movie and converted it so that it could be watched back on a TV that isn't widescreen. Doing that would not touch the horizontal resolution, but would reduce the vertical video resolution, meaning a loss of quality.
>So why is that any different to black borders left and right.
I can see what you're thinking, but it's completely different. You're confusing the shape of the video to the resolution. In the above example, you have a video which is one SHAPE (wide) and you have to put borders on it to get it to fit into another SHAPE (4:3). But in my situation, both the source video and destination video are the same shape, they're just different RESOLUTIONS.
I'm not taking a widescreen video and converting it to play on a 4:3 TV screen, nor am I taking a 4:3 video and converting it to play on a widescreen TV. I watch everything on a widescreen TV, so the videos I grab that are 4:3, like my old home videos will have black borders either side, but these black borders will be put there by my TV screen, not by the DVD video itself.
> If I stick a movie of 640x480 through TMPGenc and want it to stay 640x480 whilst keeping to 720x576 for DVD.
If you take a movie that is 640x480 and want to convert it to 720x576 for DVD without scaling it then you will have to leave a black border around the 640x480 video. You will not lose any quality of the original video, but you will have a border, and the resulting video will be slightly distorted, and everyone will look slightly fatter and shorter.
>I end up with a 720x576 movie but with a black frame with a 640x480 movie in the middle. No distortion, reduction in quality, no streching or shrinking, just a 640x480 movie in a black frame.
I agree, apart from the image will be slightly distorted, making people look slightly fatter. But we weren't talking about 640x480, we were talking about 704x576, which is different as it has the same vertical resolution as the target resolution.
>Go checkout a DVD ripping guide and see how much we normally crop of the left and right side, surprise surprise it's 8.
Where is this guide? What do you mean by "we normally crop 8 pixels off either side"? What do you crop it for?
-------------------------------------------------
I agree, apart from the image will be slightly distorted, making people look slightly fatter. But we weren't talking about 640x480, we were talking about 704x576, which is different as it has the same vertical resolution as the target resolution.
--------------------------------------------------
No it won't because it will have a black border of 40 either side (i.e. 40 left and 40 right) and a border of 48 top and 48 bottom giving you a total of 720x576.
Comment