AVI to DVD MPEG (CCE or TMPGenc encoders)

[Spiet]

ok thnx a lot , i think i finally made it work. and now i'm able to use all the formats and convert them all to dvd mpeg2 with appropriate sound and quality

your guide rocks and thanks again for the fast answers on my problems.
cheers

spiet

[DJRumpy]

Thanks, and your welcome

[phbar]

Hi.

many thanks for such a clear guide, I managed to encode and burn a concert dvd.

However i do have a couple of wuestions.

1. GSpot reports the aspect ratio as 1.36:1. What sort of film carries this aspect? Which aspect ratio do I need to select?

2. The dvd I've burned shows Video "noise" at the top and bottom of the screen. How do I get rid of this? Do I need to set multipass to 2 or 3 scans to increase the quality?


Thanks for your help. It's much appreciated as I'm completely new to the processes involved in DVD creation.

PAUL

[DJRumpy]

Many times, the end user will resize their project incorrectly, giving you odd aspects like that. In those cases, assuming it looks about 'right' as far as width and height, just pick the closest aspect (1.33 in your case).

2. The dvd I've burned shows Video "noise" at the top and bottom of the screen. How do I get rid of this? Do I need to set multipass to 2 or 3 scans to increase the quality?

You can trim the noise out by cropping off a bit of the top and bottom. You do this by changing the LanczosResize command. The syntax is as follows:

LanczosResize( targetwidth, targetheight, left, top, SourceWidth, SourceHeight)

The Target width and target height are the resolutions your resizing TO, so you would leave those as they are. The left is used to crop area off the left of the video, top crops from the top. The SourceWidth and SourceHeight values then make up the remainder of your video, minus anything you've subtracted using the LEFT and TOP values. For example, if you have a video that is exactly 100 pixels wide, and 100 pixels tall, and your LEFT value is 10, and your TOP value is 20, then your SourceWidth could be no larger than 90, and your SourceHeight could be no more than 80. This is because you've subtracted the 10 left most pixels from your 100 width video, and you've subtracted the 20 top most pixels from your 100 pixel height. This would leave you with the top and left of your video cropped off, and leave you with the bottom right corner of your video. Hopefully you can see how chaning these 4 values can crop esentially any part of the video, and then resize it to your TARGETWIDTH and TARGETHEIGHT values.

If our example 100x100 video had video noise that was approximately 5 pixels in height at the top and bottom (crop 5 pixels from the top, and 5 pixels from the bottom for a total of 10 pixels), and I wanted to resize it to 720x480, my command would look like this:

LanczosResize( 720,480,0,5,100,90 )

In the above example, we're resizing our CROPPED output to 720x480. We're cropping 0 pixels from the left. 5 pixels from the top, keeping all of our original video's 100 width, and then keeping 90 pixels in height AFTER the 5 pixels we already cropped. This in essense ignores that bottom 5 pixels, which in effect crops those off too ( 5 + 90 + 5 = 100 height...we're only keeping the 90 part).

The command can be difficult to understand since it crops, and resizes. If this is too difficult for you, you can break it up into two different parts. The CROP and the RESIZE. The crop command syntax is very simple:

CROP( left , top, right, bottom )

Left and Top are always 0 or higher. Right and Bottom are always 0 or negative values. To crop 10 pixels from the left and right sides, and 5 pixels from the top and bottom, it would look like this:

crop(10,5,-10,-5)

You would then combine this with the simple form of LanczosResize, which would only include the TargetWidth and TargetHeight like so:

LanczosResize(720,480)

Put all together in your script, it would look like this:

crop(10,5,-10,-5)
LanczosResize(720,480)

Once you get used to the long form of the resize command, you'll move away from this two-part method, only because the two-part method is a tiny bit slower.

Hopefully that explains these enough so you can grasp what needs to be done with yours.

[marrty]

Hi,
thanks for excelent guide. I followed it step by step, everything worked fine.
Only when I play a dvd at standalone dvd player(sony), the movie ussualy get frozen at the end. Source of my movies - avi,Pal,mp3 audio.
I use sony dru 700 dvd (latest firmware VY05) writer and these media- memorex, verbatim,ritek . I tried to reduce write speed, it didnot help. I also did a quality disk check by Nero for a memorex dvd - PI max 125, PO max. 3, but this dvd got frozen too.
Results are almost same for all of these media.
What am I doing wrong ?

[DJRumpy]

Try lowering the MAX bitrate to 9000 if your using TMPGenc. It's possible you have a bad bitrate spike. I've also run into a bad mpeg before that a player simply could not handle. Does the disc play fine on the PC?

[BSR]

check out this thread
Just a thought, something to try.
Good luck.

[marrty]

I am using CCE.
The discs donot play fine on PC too.

One more qeustion.
After I burn a dvd disc, can I just copy this disc to another one without loosing quality to play on dvd player?

[BSR]

After I burn a dvd disc, can I just copy this disc to another one without loosing quality to play on dvd player?

Yes, absolutly no video qulity loss.

[DJRumpy]

If it won't play on the PC either, then I'd guess it's a bad mpeg, or a bad disk. CCE typically doesn't cause bad bitrate spikes. It tends to obey your settings, so I would leave your MAX setting alone.

Marty, BSR is correct. If you created the DVD-R disk yourself, then you can simply copy the files off the disk with explorer if you were so inclined, or use any burner software to do a 1:1 copy. Typically, only store bought dvd's are encrypted, and dual layer (dual layer burners are out on the market, so check the size of yours). Assuming the dvd you want to copy is single layer 4.3 GB type, then yes, you can simply copy it directly to another disc. If you have a dual-layer burner, then you can copy any of them 1:1.

[phbar]

Thanks for your reply. I understand more or less what I've got to do (I think!).
2 more questions for you:

1. Where do I find the size of the original in pixels? Is it reported in GSpot? Approximately how many pixels are there to the inch/centimetre?

2. Does upping the number of passes drastically improve the quality of the output, or is it ni=ot really worth the three hour+ wait for the process to complete?

Thanks again for all your help.

PAUL

[DJRumpy]

gspot will only report the video's resolution. To crop the proper amount, you have to try different values. Start with 5, and go from there. Edge noise rarely takes up that many pixels.

For the number of passes, beauty is in the eye of the beholder. Only your eye can judge what is acceptable.

DVD video tends to look fine with 2 passes, even with AVG bitrates down in the 3500 range (using CCE. Ive never tried AVG settings this low with TMPGenc). If your video's AVG bitrate is >4000 then 2 passes is usually more than enough to look excellent. If the value is less than 4000, then you can consider using more than 2 passes. I would just try the 2-Pass first to see if it still looks good to you. CCE typiclly does an excellent job.

If your output is SVCD, or CVD, then you can consider more than two passes to squeeze more video onto the media. Since SVCD/CVD are much lower bitrate, and much more sensitive to macroblocking, those additional passes can really pay out in the end.

[phbar]

Thank you. Will try all this out tomorrow.

Cheers

PAUL

[marrty]

Thank you

[mlopes]

Can you tell me a simple(objectve) and fast way to now if my source is progressive or interlaced?Thanks for the help.

[DJRumpy]

Once again, in english?

"and fast way to now if my source" ???

[mlopes]

Do you speak portuguese?Well sorry if my english is bad.

[DJRumpy]

Sorry mlopes. A little fun at your expense

I'm guessing your wanting to know how to tell if your source is progressive, interlaced, or telecined? (hopefully you do, or I'm going to waste a lot of text here ). If not, just try again, or use a translater on the web to translate from Portuguese to English.

The most direct method, is to examine the frames in VirtualDub. Just drop your video into VirtualDubMPEG, and click through the video frame by frame (pick somewhere in the middle of the video. Try to find a high motion scene as this makes the interlacing much easier to spot). If the film is progressive, each frame will look 'normal'. If the video is interlaced, each frame will have a 'combed' appearance (the interlacing effect will look like horizontal lines running all through the frame). If it's telecined, you'll see a series of progressive frames, and interlaced frames mixed together, in a 3:2 pattern (three progressive, two interlaced, three progressive, two interlaced). The pattern can, and will change, so it may not always follow an exact 3:2 pattern, but it still indicates a telecined video. This type of video can typically be converted back to progressive without losing any image quality (you actually improve the quality by restoring it to it's original progressive output).

Interlaced video cannot be converted to progressive without losing image quality.

Hope that answers your question. If not, let me know!

[mlopes]

Many thanks for the help, i think that i understand your answear.
My english is a litle bad because i`m not training it.
May be you can help me for the last time, i have bought a program called fit2disc dont now if you now it?Well in the final of the program i can save a script to open in CCE to encode the movie, but when i see the movie the dark scenes seems to have clouds(sorry about the term used dont now any other),because that, i dont like very much this program but i have almost sure that the problem is of the script given by the program fit2disc.Can you help on this?Many thanks again.

[razoon]

Hello DJ,

I burned a DVD and I followed more or less ur steps.

I tried first with TMPGens Plus but it just gave me the black screen with the word 'analysing' in the middle
So I tried it with TMPGens Express 3.0. I pulled and dropped the .avs onto it and to my surprise it not gave me the black screen at the preview. So I let Express make the DVD files. And I burned them with Nero.

It looked good on the pc...however on the television it gives me some odd distortions. Like 6 horizontal 'lines'....but only in the parts where there is camera movement... abit fast from left to right or up and down.
The lines look like uhmm...u know the part at the bottom of the screen when u not crop...those 12-14 lines when u capture from vhs. Well those 6 'lines' look like that each. And they are divided horizontally over the screen. But they appear only at the scenes with camera movement.

Do you have any idea what they are and how to get rid of them?

thanks in advance

[DJRumpy]

mlopes, try changing the setting in CCE under the VIDEO button. Specifically, the "Luminance Level" setting. If it's on 16 to 235, change it to 0 to 255, or vice versa. It's possible the problem was already encoded into your video. If that's the case, you can try darkening your output, but I'm not familiar with the software your using. It may not have an option to darken the video. If it doesn't, let me know, and I'll show you how to do this via AVISynth. It will not fix your problem, but it will minimiz the appearance of it.

razoon, your field order is probably reversed, although the problem you describe is a bit odd, and doesn't fit exactly with that problem. Your sure their not in your source material? Regardless, try switching the field order. I don't have a copy of express with me, so if you can't find the setting, just post a thread in the advanced forum, asking how to change it. Then try encoding a small sample, preferably from the scene where you see this problem.

[mlopes]

Well i have tried with another movie and the final result was very good without making any changes so the problem was the source i think.About this software(fit2disc) you must try it because is very good.It makes a lot of work for you like calculating bitrate, aspect ratio, etc.But may you explaining me the best way to configure aviscript to obtain the best results?Thanks once more.

[DJRumpy]

There is no best way, as each movie's requirements varies. You'll either have to stick with the one click software, or learn the hard way. I prefer the hard way

[Sakuya]

I'm wondering, what's the point of using AVISynth to resize? Won't TMPGEnc resize anyway? I have an AVI that's 640x480. That's 1.33 which means it is 4:3 right? Must I need to select "4:3 525 line (NTSC)" in Source Aspect Ratio in the Advanced tab? I usually select it as 1:1 (VGA).

Also, what is the setting "Output bitstream for edit (Closed GOP)" in the GOP Structure tab? I'm pretty sure it was checked by default and I always left it alone. When I'm encoding anime, is it good to select "CG/Animation" for the Quantize Matrix?

[DJRumpy]

You can resize in either, although AVISynth tends to be faster. TMPGenc used to work only in the RGB colorspace, which was very slow in comparison. I have no idea if that is still the case with the Express versions.

The Closed Gop setting ensures maximum compatability with most authoring software. Although closing each GOP can result in a theoretical hit in quality, you'll never see the difference at the bitrates we're working with. As to your matrix question, yes, you should select the CG matrix when working with anime. That matrix is especially good at encoding cartoon type material.

[mlopes]

Can you show me how to work with avisynth?

[razoon]

razoon, your field order is probably reversed, although the problem you describe is a bit odd, and doesn't fit exactly with that problem. Your sure their not in your source material? Regardless, try switching the field order. I don't have a copy of express with me, so if you can't find the setting, just post a thread in the advanced forum, asking how to change it. Then try encoding a small sample, preferably from the scene where you see this problem.

Hi DJ,

Uhm I think its in the source yes. I tried bff now but same bad result.
When I play the source avi in Xpress I see alot of comb effect at the scenes with camera movement. The faster the movement the more you see the comb effect.
I captured at 720x576, 2-pass vbr, interlaced. TMPGenc settings more or less as u described at the start.
Any idea what causes this comb effect?
Cus probably its the cause?

[razoon]

Welll the source is a vcr tape..I mean the capture is where it went wrong I guess

[DJRumpy]

You might be able to reverse the field order to fix the problem. Give it a try on a small encoded sample and see if it improves.

[razoon]

hehe...yes its fixed now ...TFF it is
Thank you very much DJ