AVI to DVD MPEG (CCE or TMPGenc encoders)

[DJRumpy]

You would typically have to listen to both the 25 fps version and the 23.976 fps version to notice a difference. It's a barely perceptable change in pitch when switching between those two framerates, which is why it's so common to convert between the two.

To determine if your source video is telecined, you need to drop it into virtualdub. You then drag the slider at the bottom to somewhere in the middle of the movie, preferably a good action scene. Now your going to use the frame advance buttons. They should be the 5th and 6th buttons from the bottom left. They allow you to advance through your video frame by frame. You should scan 20-30 frames to be sure you properly identify the video. You can also scan it in different locations in the video. Now what your looking for, are three progressive frames, followed by two interlaced frames in a repeating pattern of 3 progressive, 2 interlaced, 3 progressive (the pattern can vary somewhat). A progressive frame will look just like a still screen capture of your movie. An interlaced frame looks like someone dragged a comb through the picture from left to right (horizontally). You'll only notice it if there is motion in the frame.

(by the way, 100fps.com is an excellent source of information on all of these topics)
The link above is a pretty drastic example. Your video may not have anything quite so obvious. Just make sure the frames your checking have a good amount of motion in the scene, otherwise the interlacing effects can't be seen easily,

If all of the frames appear to be 'combed', then your video is true or pure Interlaced, and NOT Telecined. You should leave an interlaced video interlaced when it will be played back on a television. Deinterlacing causes a loss in quality, and it is not necessary for playback on a display that supports interlaced material (like a TV).

If all of the frames appear to be progressive (no combing) then you cannot neatly convert your video back to 23.976 fps, or even 25 fps. Leave it AS IS in regards to frame rate (leave the AssumeFPS command out of your script), and encode it using it's current frame rate.

If the video does appear to have a mix of progressive and interlaced, then it is most likly Telecined. You can perform Inverse Telecine to remove the duplicated fields, and restore it to the original 23.976 frames per second. This is a lossless process (meaning no quality is lost during the process), because we're simply removing the duplicated fields (see below).

A typical 23.976 video clip is made up of Frames. Each frame of video is made up of two fields, Top, and Bottom (T & B). Each field is displayed on a telvision like this:

#1 #2 #3 #4
------------------
1T 2T 3T 4T
1B 2B 3B 4B

The numbers represent the Frame # of our example video. The T & B represent the Top and Bottom FIELD that make up that FRAME of video (frame one is made up of 1T and 1B, Frame #2 is made up of 2T and 2B, and so on). The above sequence is a Progressive video, because the top and bottom fields for each frame are played back in proper order, and in the proper Frame. The FIELD ORDER for a video is simply whether or not the top field is played on the first pass, or the second pass. A TV makes two passes to draw each frame on your TV. The first pass draws one field, and then the other. For every frame of video, your TV actually draws the screen twice.

To describe it another way, if you numbered every horizontal line of your TV's display, starting with 0, the first pass, it would draw all of the EVEN numbers, and then the second pass, it would draw all of the ODD lines. That's why you sometimes here video fields referred to as Even and Odd, or Top and Bottom. Ok, now that you hopefully have a grasp on that part

In order to convert a 23.976 FILM to 29.97, they duplicate certain FIELDS in the video to increase the number of total FRAMES per second. They do this in a manner that is unnoticable to the nakid eye. A telecined video FRAME sequence would look something like this:

#1 #2 #3 #4 #5
-------------------------
1T 2T 2T 3T 4T
1B 2B 3B 4B 4B

Note that we now have an extra FRAME ( #5 ) where we had only 4 frames before, because we blended two frames (#3 and #4) to create a 3rd new frame.
Notice that there are two (2T) fields and two (4B) fields (these are the duplicate fields I mentioned above), and fields from frames 2, 3, and 4 are blended together.

This is what gives you the 3:2 pattern we talked about for Telecined video. Frame 1 is progressive, Frame 2 is progressive, 3 and 4 are interlaced because they contain fields from two different frames, and frame 5 is progressive.

P=Progressive
I=Interlaced

(1-P) (2-P) (3-I) (4-I) (5-P) (1-P) (2-P) (3-I) (4-I) (5-P)

Or just showing the frame type: P P P I I P P P I I

or 3P : 2I : 3P : 2I : 3P : 2I

With me so far?

Now, to restore a 29.97 fps Telecined video back to FILM (23.976 fps), you simply discard the duplicated FIELDS in your video, and stack them back in proper order like they were in the first example. To do this with AVISynth, I'd suggest you download the Donald Graft filter Decomb.
http://neuron2.net/decomb/decomb521.zip

You would add the following to your script BEFORE cropping, or filtering for noise.

AVISource("film.avi")
Telecide(order=1)
Decimate(cycle=5)

You must put the decomb.dll file from the above ZIP link, in your PLUGINS directory (usually C:\Program Files\AVISynth 2.5\Plugins)

Do NOT use Decomb if your video is true Interlaced, or Progressive. Only use it if your video is Telecined. If you used it on Progressive or true Interlaced video, you would end up chopping valid fields out of your video in order to reduce the frame rate (this is called DECIMATING frames, and it's a bad thing).

Telecine/Inverse Telecine are fairly complex topics. I should probably turn this into a guide, and perhaps I will

Let me know if I've lost you anywhere, and I'll try to clarify.

[dbsnwo]

Thanks excellent Guide.

Using avisynth scripts does anybody know what the command lines should be if my original avi is 23.97 FPS 2:35:1 ratio when it is to be played on a 16:9 widescreen T.V. I would like my encode to look like 1:77:1 ratio so there are no top and bottom borders.

Also if i may ask, i once installed a codec pack that screwed up my playback and encoding what codecs does everybody have installed, i only seem to need AC3 filter AC3 Dec,Divx 5 and Xvid 1.03. for encoding.

V-dub
Avisynth
CCE 2.67
Pulldown

Thanks again superb thread...

[DJRumpy]

To convert a video from 2.35 to 1.77, you just figure out what the correct width would be for your AVI's current height.

If your avi is 640x272, which is a 2.35 aspect ratio (640 / 272 = 2.35), then you would simply multply the height by the desired aspect ratio, or divide the height by the desired aspect ratio, and round to the nearest 4. Since 640 / 1.77 = 360 (rounded to the nearest 4), and our avi is only 272 pixels heigh, this won't work (you can't crop what isn't there ). We'll multiply 272 * 1.77 instead. That gives us 481 (272*1.77=481), rounded to the nearest 4 is 480. You would crop your avi to 480x272, and then resize it to whatever output resolution you needed.

I'm surprised you would want to chop up a video this way though. If you eventually buy a widescreen tv, you will barely even see the lines on a 2.35 video, and 1.85 and 1.77 aren't even visible. If you must chop your video up, try using a 1.85 aspect, or even a 2.21 aspect.

272 * 2.21 = 600 (rounded to nearest 4)
272 * 1.85 = 504 (rounded to nearest 4)

You don't loose quite so much video with these two aspects.

Regarding your codec question, install only what you need. There is no standard set that everyone should have. If you insist on a basic set, then I would suggest XviD, DivX, AC3, DTS, Ogg, Huffman (for captures), and VOBSub for any subtitles.

Cheers

[dbsnwo]

Thanks DJRumpy i will use 1:85:1 aspect...

[Jonas_Vedder]

Hi

Thanks for the long explanation DJRumpy..I understood now...I looked carefully my movie in VirtualDub, and because of your explanation, I can securely say that it isn't telecined..it's progressive!!
So, I'll leave it as it is right?

Thanks again, and congratulations once more for the excelent guide.

PS: Shouldn't your guide be in the "HOW TO" section instead of being in a forum???Maybe more people would see it!

[DJRumpy]

It's actually in the Guides section, but the guide is actually created in the Forum.

Your correct regarding your video. If it's progressive, you should definatley leave it as is.

[Phoenix2443]

"Frame Rate: Set this to your .AVS input files' frame rate"

When you say this, do you mean the Input file of the AVS or the AVS file which is the input to TMPGEnc? My avi was 25fps but my AVS script is written to recode it as 23.97fps. This is sort of important. Its simply the way its worded.


I decided to use TMPGEnc because my audio and video files were constantly becoming out of sync with one another when I used CCE. I dont understand why.

[DJRumpy]

I'll see about rewording it.

It should be whatever frame rate your AVS file is putting OUT, or to put it another way, whatever framerate your final output is going to be.

[Jonas_Vedder]

DJRumpy..I have a problem in here...

Sometimes when encoding with CCE, I noticed that the audio sometimes wasn't in synchro with the video...
I use DVD-Lab to author my mpeg2 videos..there is an option there where I can put any delay to the audio, but is there a better way to do this instead of doing it with DVD-Lab? Why does CCE sometimes put a delay in the audio?My avi's don't have the audio and video unsynchronyzed so why does CCE sometimes do this?

[DJRumpy]

I can only gues that it's a problem with CCE. That assumes that the video itself plays in syc, and that the AVS script plays in sync.

The version I have (2.67.00.09) doesn't exhibit this problem, and I've had it for well over a year. Yours is the second complaint I've heard. I'm guessing either a bad "patched" version of CCE, or just a bad version. You might try posting in the main forum, or even the CCE support site to see if you can find anything. You could always try another version of CCE to see if it is indeed CCE that is causing the issue.

[rated_z]

First off just like to say that this is a great guide and well explained.
My problem is with AC3 file and importing it into DVD Maestro where I get a Media Detection Error. My source avi has ac3 audio. I used VirtualDubMod to Demux the ac3 file, checked with ac3fix.exe which turned out fine and then used Besweet to do the PAL to NTSC conversion. Since that ac3 file gave me the error in DVD Maestro I tried saving the audio as WAV and then did the convertion in BeSweet and still got the same error. Last thing I tried was extracting the ac3 with ffmpeggui and still got the same error. The only audio I can import is the WAV file. What am I doing wrong?

I'm posting in this forum instead of the authoring forum because I followed this guide.

[DJRumpy]

Your issue is with Maestro, which I don't personally use, and this guide doesn't cover, so no promises. You would probably get a better response in the authoring forums. That said, a few questions:

What is the specific detail/text of the error in Maestro? Surely it provides more information than "media detection error"

What frequency is your AC3? (it is 48Khz right?)

[rated_z]

The Maestro error reads "Media Detection Error (0xc10b0001)" I have no problem with importing the video files. The ac3 audio is 48 KHz. I'll try the authoring forum. Thanks DJRumpy for the quick reply.

[DJRumpy]

A quick search of google indicated that it might be some sort of dongle copy protection. Does your software have a dongle you have to insert?

[Scars Unseen]

# 2nd Choice:
clip=AVISource("Input.avi", False)
aud=WAVSource("audio.wav")
AudioDub(clip,aud)
ResampleAudio(48000)

What do I put in if I have two wav files to encode(dual language tracks)?

[DJRumpy]

As far as I know, AVISynth doesn't handle multiple audio tracks. To encode those, you should join the wave files using any of the multitude of tools out there. Just search google for "join wav files". After joining them, convert the WAV files to AC3 using AC3Machine, or whatever tool you prefer.

Encode your video as if it had AC3 audio using the above guide.

[Scars Unseen]

Don't think that would work very well, since they are two seperate language tracks, rather than 2 parts of one track.... splitting the AC3 file afterwards could be tricky...
I'll probably just create a seperate AVS for the second audio track. Thats for the advice though

[DJRumpy]

You mean your audio files are not split into multiple parts, but rather you have multiple tracks that are not split?

[Scars Unseen]

Yes. I have a Japanese and English .ogg file(encoded into 2 wav files, of course)

[DJRumpy]

Then I would just encode them to AC3, and use the AC3 instructions. You don't join them in any way since they aren't split. The only oddity is the fact that you have to audio tracks. Depending on your dvd authoring software, you can either multiplex them together before hand using any muxing tool, or if you dvd authoring software supports it, just add the ac3 tracks during the authoring step.

[TooLFooL]

hi everyone, what great info! after reading these 10 pages of posts, one question still lingers on my mind...

why exactly do you need avisynth and avisynth scripts?

why cant you directly view the avi or direcly encode it to mpeg? why must you create an avisynth script and view/encode that?

just wondering, that part was never really explained. thanks everybody!

[DJRumpy]

The AVI's can be viewed directly on pretty much any PC, and on a very few number of DivX compatible players. These players tend to be somewhat picky in the codecs and encoding/decoding features they support though, which makes this difficult because the video codec scene is constantly changing and advancing. Unless these players come with easy upgrade ability in regards to the codecs and encoding features that they support, they would quickly become outstreched by the media they are supposed to play. The alternative is to convert the AVI to a standard like DVD or SVCD/CVD.

Because the resolutions used in an avi do not match the resolutions used by DVD, SVCD, or CVD. You must resize them at a minimum, and often add letterboxing to ensure the aspect ratio is correct for the media you are burning them too. It's extremely rare to find an AVI that actually matches the resolution and formatting required of DVD or SVCD/CVD video.

In addition, the typical AVI does not support aspect ratio. They use a 1:1 pixel aspect ratio (what you see is what you get). MPEG however does support aspect ratios, meaning that they can be resized, stretched, sqaushed any which way, and when played back, the embedded aspect ratio tells the player exactly how wide and tall it should be.

For example, a typical AVI you find on the net would have a 640x272 resolution. In order to put that on DVD, it must be resized to 720x480, 704x480, or 352x480 (NTSC). The audio is also often recompressed in a more portable format like MP3 to further reduce the size. If the movie is lengthy, it's also often split into multiple parts so it can easily fit on a CDR. Any letterboxing is also removed during the process to reduce bitrate requirements. All of these things must be undone to convert it back to DVD format.

As media PC's, and media players for TV become more popular, I would expect that these conversions will take place less often. I for one have a media pc now, and I haven't converted anything to DVD in months. I can get full 6.1 surround via my tuner and fiber outs on my pc, and 480P output from my media pc to my HDTV that looks excellent. The only items I would consider converting would be avi's that I want to permanently archive. Those are few and far between, as I simply go out and buy what I want to keep.

[TooLFooL]

so does avisynth do things that can't be done with tmpgenc? i just didn't understand why you can't directly encode from avi to mpeg with tmegenc and without avisynth.

also i have i more question!...
from what i have read most avi's aren't letterboxed, but i do have an avi that IS encoded with letterbox. the aspect ratio is 640x272 (with letterbox). question... should i use virtualdub to crop and eliminate the letterbox before i reencode it, or is that something that can be done on the fly with the avisynth script?

i have a damn-near-dvd quality avi complete with ac3 sound, and i would like to make a dvd so i can watch it!
thanks for the help!

[DJRumpy]

You can encode the AVI directly with TMPGenc, but TMPGenc would also need to resize your AVI. TMPGenc is simply slower since it works in the RGB colorspace. AVISynth is more flexible because you can use pretty much use any filter written for it. Don't like the way a particular filter looks, or the final output? Just use another. There are hundreds of plugins available for it. With TMPGenc your stuck with whatever filters are written into the program.

I get the impression you don't like AVISynth. If you don't like it, then don't use it. It's just that simple. Use TMPGenc instead. It's slower, but it offers a simple GUI interface for the newbie.

Regarding your AVI, it's highly unlikely that you have an avi that is 640x272 and it still contains letterboxing. Your probably only seeing the black borders that are added automatically by your media player. Almost all of them will simply show black around the video when it is played fullscreen. If in doubt, just open it in VirutalDub and see if you can see letterboxing there (it won't show black borders). The aspect ratio for a 640x272 video is 2.35. They just don't get much "wider" then that. If you really had a video that had letterboxing at that resolution, it would have an even wider aspect than Widescreen Anamorphic 2.35 video. Very doubtful.

If for some bizarre reason, you did, you would use the Crop command in AVISynth to crop the letterboxing. Verify it in VirtualDub first.
[/quote]

[TooLFooL]

you're 100% correct! i'm stupid! i loaded it into virtualdub and the letterbox is gone! sorry!
and it's not that i don't like avisynth! this is my first venture into this kind of thing and i was just trying to understand exactly why you NEED it. but i guess i understand now.
anyway, after 15 hours, my avi is encoded to an mpg, but i have another question!... there are very slim yellow borders running up the left & right side of the video. are these for 'overscan'? will my tv take them off? i hope so! thanks again, so far, my mpg looks great, i'm almost there!

[DJRumpy]

15 hours? Even a slow PC (say P3 and higher) should enocode it in an hour or two, or are you talking total time you've spent working on it?

The yellow borders are unusual. They will probably NOT show up on a TV. If in doubt, burn your project to RW media if your dvd player will support it. If not, you can crop the top and the left side of your video using the resize or the crop command. Let me know how you want to proceed, and I'll show you how. Hopefully your player supports RW, and we won't need to crop anything.

[TooLFooL]

yep, 15 hours to encode mpeg from avi with tmpegenc following your settings exactly i have an athlon 1700.
i do have a dvd rw, but now the problem is the only 2 authoring softwares i have are myDVD and pinnacle 8. apparently myDVD cant handle ac3 @ 448 bitrate, and for some reason when i pasted ac3 into pinnacle it is showing the audio stream extending about 25 minutes past the movie.
this was a 2-part avi that i joined, and ac3fix said stream was fine needing no correction. my plan was to go ahead and try it anyway using pinnacle, but because the resulting file will be too large to fit on a dvd-r, it refuses to even make a disc image. so i guess now i need to seek other software? what do you suggest? thanks again!

[DJRumpy]

Your bitrate calculations must have been off. If your Average setting is correct, and you also included the audio at 448Kbps, your project should have fit exactly onto a disk. I'm thinking that any authoring software will probably report them as being too big, but I've never used pinnacle. Maybe it errs on the side of caution. How big are the files when you have both of them selected in Windows?

I would suggest SpruceUp, or DVD Lab if you want to try something else. Both will handle AC3 audio. SpruceUp is no longer produced, but you can find it on the net for download.

[TooLFooL]

when i select both files they show total size as 4.29 GB. i did calculate with 448 audio. bitrate calc shows average as 4859 and max as 9341, but i used 9500 as max as per your guide. pinnacle shows total file about 70 minutes too long to fit when set to 'best video quality', but if i choose automatic it will compress about 70%, which defeats my purpose!

any idea why mpeg compilation takes so long for me, or why the audio stream appeared 25 minutes longer than video?

ill try to find software you suggested. (if this avi wasn't such excellent quality, i surely wouldn't waste this much time!)

oh, also, i was playing with tmpegenc and noticed setting under advanced/video arrange method that says 'no margin (keep aspect ratio)' where i chose 'center (keep aspect ratio)' as per instructed. i think this will eliminate the yellow bars on the sides, but im not gonna spend another 15 hours finding out!... thanks again!

[Jonas_Vedder]

Hi DJRumpy

I have a question about CCE! Can't CCE convert to mpeg2, mpeg1 files?
I dragged to the CCE window a mpeg1 file and this msg appeared:

GdipLoadImageFromFile failed : 3

What does this mean??

Can we just convert with CCE only avi's files??
How can I convert mpeg1 files?
Thanx