Star Wars Conversions-Hit me with your best shot...

[borghe]

the only problem with what you are saying is that when the flm is telecined it is still completely done from film at 23.97fps. Video at that point is moot. All you are doing with the IVTC process is taking it back to the films original frame count and look.

The one, simple, and only thing that causes jerky video is a change in the 3:2 pattern. Now the 3:2 pattern may certainly change because of, say.. going from a pure film segment to going to a mixed film/video segment, but technically if you can sync the IVTC process at the point with the correct frames from the finished film, it will still look fine.

You may be on to something, but the only thing that causes jerky video is a 3:2 pattern shift that the IVTC algortihm doesn't pick up or adapt to in time.

[borghe]

back to the encoding, I have two qustions.

First, I was reading that some dvd players have problems if there are over 72 chapters. Star Wars in this collection has 81. Can anyone explain to me what problems some players might have with this? Should I just lose some chapter stops? I already have the first disc chaptered (35... whew...) but it is easy enough to drop some of those if need be.

Second, can anyone tell me how with either Scenarist, ReelDVD, or IFOEdit, I can force subtitles during Greedo? I have the subtitles created and properly timed, but unless someone has the subtitle track turned on they won't play. I know other DVDs have done this (ironically enough Ep1 and 2 DVDs do this), just wondering if anyone can tell me what command I use either in one of the apps or what command I add with IFOEdit to force English subtitles during that conversation. Thanks.

[vhelp]

@ borghe,

I think you need to re-think that 3,2 pattern issue again.. cause you're wrong!

Ok, say you have your normal video 24fps sequence, now you have to
throw in some animation/bluescreen/computerGenerated sequences. Well,
this is NOT 24fps film. Those sequences are hard framed.. that is, either
30 fps or they are Progressive. These sequences (will call them LM for
last minutes sequences or footage) are usually Progressive or Interlaced.
If Interlaced, they are not 24fps Film.

So, now you have 24fps normal video, and you want to mix into the Telecining
process, these LM sequences. When you the Telecine process does
the film spilting of those last 2 frames, those bluescreens don't usually get
converted because they would cause jerkes in moviements and woudn't
be smooth w/ the normal video. So, somewhere's, the process is either
stopped or is auto adjusted during those LM sequences, hence, making
the total Film smooth.

Now, when you take that same Film, and capture it at your 29.970 fps rate,
you are capturing the Film's essense. That is, it's smoothness. But, when
you go and apply an IVTC, if those LM sequences that were put through
the Telecining process were Interlace or Progressive (non-Film) you WILL
be cutting those sequence frames where there were NO Telecining done in
those sections of the foootage, hence you jerkyness (no smooth parts in
those blue screen sequences)

So, lets say in the LM frames, that they are 30 (29.970) frames and they
are Interlaced (usually are for smoothest play) the IVTC will be cutting into
one of those Interlaced frames, but rather incorrectly. But, lets say that
those LM were in fact IVTC along w/ the 24fps Film. Again, when you
cut (undo) w/ the IVTC on an Interlaced sequence, you'll throw out a frame.

The best way to understand this is to lay out some video frames and imagin
them in this situation, OR, use your captured source to illistrate this, by going
from one frame to the next, and when you get to a blue screen (usually a
computer generated seq. ) note if thse are in fact IVTC (3,2) pattern right
into the 24fps Film. If they are, and the 3,2 pattern rolls smoothely w/ the
source, then you have no problem w/ using an IVTC on the whole film. And,
you shouldn't then. So, if you are STILL seeing trouble w/ you IVTC'ed
results, then the only other cause of your jerkyness is Lost Frame during
capture!! Lost Frames will DEFFINATELY throw off your IVTC ta-puwey

Hope that better explain those IVTC blues.
-vhelp

[borghe]

As far as I understood it, telecine transfers a 24fps progressive film sequence to a 29.97fps interlaced video sequence using 3:2. IVTC should in all reality take you back to where you were at before the telecine since it does the exact opposite of the telecine. If I convert aabbaabbaabbaabb to aabbbaabbbaabbbaabbb, technically if I remove ever fifth frame I would be back at the exact same point.

Telecine: 24fps film -> 29.97fps video
Inverse Telecine: 29.97fps video -> 24fps film

If the production video used for an fx shot is never used in the telecine process, it transitively couldn't be affecting the IVTC. The only way what you are saying would make sense is if they used that production video in the telecine process instead of the film transferred version that is used for projection. If this is the case then what you are saying makes sense, but if the only use the film during the telecine process with no video then I don't see how what you are saying would have any effect.

I am not trying to look like an authority on this, I just don't understand how when you see the film in the theater it looks flawless, when it gets transferred to video it looks flawless, but when you try transferring it back to film it looks messed up?

What I do know for certain is that they (or actually the telecine machine) do change the 3:2 pattern during the telecine process in cases when the original 3:2 pattern would cause serious artifcating. They then change back when another anamoly would occur. This is what causes the change in 3:2 patterns. I also know that an IVTC process that doesn't adapt correctly, say removing every third frame from a sequence like aabbbaaabbaabbb, will cause jerky motion. It just seems like the IVTC process not adapting to a pattern change seems like a more likely cause. But again, I'm far from an authority.

BTW, does anyone have an answer to my forced subtitles question above?

[Mythos2002]

Borghe: Are you getting any jerkiness in your encodings? I'm going to take a close look at my captures and I'm going to try a fresh capture to make sure there are no lost frames and make sure there isn't any jerkiness in them or the original source itself. I don't know if it means anything, but I tried the check option under TMPGEnc's IVTC filter at 24fps and get the message, "Synchronism has broken after this frame.".

Is it better to capture the black bars in Virtual Dub or whichever capturing program you use and then clip them in your Encoding program? I've been doing it in Virtual Dub. I'm also going to read up on how to take the audio out so that I can encode it separately in BeSweet.

I'm just wondering if it is just better to go through the time consuming process of manually IVTCing the whole film.

I wish I had an answer for you on the forced subtitles. I think flaninacupboard would be one of the best to ask about that. They will probably tune in.

I can't wait until June 6th. That is when I'll start summer vacation and have a lot more time (hopefully) to dedicate to this.

Mythos

[vhelp]

@ Mythos,

If you are seeing any jerkiness in your IVTC (after encoding) then perhaps
you should look closly at your .AVI file to see if there are any dropped
frames. No need to re-capture it just yet. First see if their are any dropped
frames.

Usually, when the source is Telecined from start to finish, the encoding
would be an accurate 3,2 patter all the way through. But, on certain scenes,
there this may not be so ie, blue screen and computer generated stuff like
that and so forth. But, don't forget that if those other scenes are a mix of different FPS, then you'll probably get this jerkiness from time to time.
I'm basing the blue screen thing due to experience when I did several
captures from satellite of the old Star Wars (full screen versions) These
had lots of jerkiness ie, in the space craft and chase scenes. Then again,
they could have ben due to the re-edits for TV airing. They could have
falled up the Telecine process or purposely did this to screw us. But, in any
case.. Anyways..

I don't know of any apps that can detect dropped frames after you've already
captured your source. When you capture your source, that's when you
should be alert of any dropped frames ie, vdbu and avi_io show you dropped
frames (of course you know this already) but still.
Its probably a safe bet that this is the cause of your jerkiness. Other's here
may be using an ADVC-100 where you wouldn't get dropped frames. But,
capture cards are another story.

If you can't get the full 720x480 capture w/out dropped frames, then try
352x480 instead. Still good quality, and fewer (if any) dropped frames.
You don't want ANY in an IVTC process.

Good luck thus far.
-vhelp

[borghe]

mythos - I have no jerkiness.. I DID have jerkiness when I did the following:

capture from DV, frameserve from VirtualDub which does the deinterlacing and the frame conversion, then pulling that into TMPEGEnc which did IVTC. I wanted to IVTC (including deinterlace) before I did anything to the frame, but I couldn't find the appropriate plugins for VirtualDub.

The answer was to go to AVISynth, I use the decomb plugin. You can find my script on page 6 of this thread. I then just created scripts for all five sides, adjusted the trim parameters in each file, used virtualdub's crop preview to check my cropping (it is always 276 pixels high, but the top y offset would sometimes change by up to 3-4 pixels from side to side), then created a sixth avisynth script that merged the five avisynth scripts into one movie. I then feed this into whatever encoder I use. The results are no jerkiness at all.

As vhelp said, first make sure you aren't dropping any frames. If you are, there will be jerkiness no matter what as the IVTC will have no choice but to guess wrong. As long as you are not dropping frames, I would use AVISynth and decomb. I have watched my entire rough dvd and the picture is better than the special edition video tapes by far (even with an a/b comparison with my friends).

As for subtitles, I ran into work yesterday to play around with scenarist and found that within scenarist it is possible to force individual subtitles on. Very helpful. Although I am still wondering what the procedure is to do it outside of a $22,000 program. Any thoughts?

[Robster]

I'm currently working on burned-in permanent subs on my Star Wars captures using the Video Vegas LE package included with my Canopus card. They look great, much more part of the film than player-generated ones.

[Mythos2002]

Borghe,

Did you run your first AVS script through Virtual Dub or though TMPEGenc? I know you ran the one that combines them all in TMPEGenc, but I wasn't sure about the first five. I have to find the decomb filter. I guess I can check Shelob's for that.

I've heard a lot about audio. Are you guys capturing the audio together with the video and then separating them or are you capturing the video and going back to capture the audio? I've looked through Virtual Dub's docs and can't find much on that.

Thanks
Mythos

[borghe]

robster - you are right that burned in subs will always look better than subpictures.. I went with subpictures though because that will give me clean screen captures on scenes with subs.

mythos - I ran five separate clips through avisynth and then ran all five scripts through avisynth again. My film assembler script is one line:

AVISource("swotep4s1.avs","swotep4s2.avs","swotep4 s3.avs","swotep4s4.avs","swotep4s5.avs")

That's it. I didn't use virtualdub at all for frameserving, only avisynth at two points in the same chain.

I can honestly say I am done with the capturing and encoding. I have finished chaptering everything inside scenarist, finished the subtitles. All that is left are the menus. Here is a directory with the pics I was thinking about using for the chapters. Yes there are 81 pics in there, yes there are 81 chapters. I emulated the chapter layout of the laserdiscs perfectly. It was long.

For those who are curious, those are all grabs off of my disc. Those are almost all I frames. Most turned out really good. All I did to the pics was cut off the 120 pixels of black bars and shrink the pix vertically to account for the anamorphic stretching of the 720x480 frame. Comments?

[Mythos2002]

Borghe,

That's great that you are finished and happy with the results. I've been trying to use AVISynth but have never been able to no matter what guides I read. What version are you using and what plugins do I need? Also, how do I open AVISynth since it isn't a standalone program? Thanks.

Mythos

[borghe]

that's the same things that confused me at first also. don't feel bad.

the only plugin I used was shelob's Decomb. That contains both Telecide() and Decimate(). Doning DV transfer from the discs, I then created the scripts on the other page for each movie (laserdisc side in this case). You write the scripts exactly as I have outlined. You then create a sixth script that contains the loading line I mentioned above. You should now have six .avs files in a directory. All you do now is load that .avs file as an avi file in your encoder. A word of warning: If you are using MainConcept, make sure you move the DirectShow import filter down below the AVI filter for .avs scripts, otherwise it will crash.

It seems weird at first just loading the text files, but that's how you use avisynth. Trust in the force, it works. I use the five scripts to basically create perfect movies from each side of the disc, then use the sixth script to combine those together into one perfectly edited movie. The encoder uses that sixth script. After avisynth is done you are left with a movie running at 23.97fps. Make sure that you encode it that way and signal 3:2 pulldown on playback.

[Mythos2002]

Borghe

I'm still having trouble getting any script to work. I have the decomb plugin, but don't know where to put it. I put a copy in the VirtualDub plugins folder, but it doesn't work. I couldn't figure out if I need to put it in TMPENGenc or not. Even if I have a one line script like:
avisource ("f:\empire.avi")

It doesn't work. I get variable not found or some other error. I have AVISynth 2.07 installed. Where am I going wrong? Thanks

Mythos

[GreyDeath]

Borghe,

I looked at some of your screenshots which looked pretty good, but I'm wondering if you could capture a shot from your transfer where on the original LD you would have a moire effect, say, in the beginning where C-3PO is wondering where R2-D2 is, or maybe the Rebel War Room. Both scenes have near horizontal lines which mixed badly with the interlace scan.

I just want to see what the effect of the AVSynth filter had on those parts.

[Hotskillz]

What is an ideal Laserdisc player to have for this process? Figure $50~$100 on eBay.

[Mythos2002]

Borghe,

I'm still working on the AVS scripting. Maybe I just don't have AVISynth installed correctly. I unzipped it and ran the registry install. Do I need to install something on my Windows drive? I have the program on one of my other drives currently.

What is an ideal Laserdisc player to have for this process? Figure $50~$100 on eBay.

It depends on which version you are doing and if you only plan to use your LD Player for converting a few LD's to DVD. I would definitely recommend Pioneer or maybe one of its clones like Mitsubishi. If you are doing the Special Editions, I would go with the Pioneer CLD-D704 as it has AC-3 audio. I'm doing the Definitive Collection versions which are the original versions that do not have AC-3 audio. I have the CLD-D703 which can be modified to be a D704. I don't need AC-3 as these are the only LD's I own and they only have digital stereo. The D703 has both coaxial and optical digital audio outputs. My player was $240.00 with shipping and the player is in excellent shape with the original box, manual, and remote. I wouldn't go too cheap on an LD player as it could have negative effects on your capturing. There are better Pioneers' than the ones I mentioned, but they are going to cost more. I'm not an LD expert, but I put in a lot of research before I bought mine and found that Pioneer was the LD Player king. Here is a really good website that has a pretty detailed list of LD Players divided up by brand and model number. They also have a lot of pictures and have detailed lists of all of the features for most of the players on the site. The link is http://www.laserdiscarchive.co.uk/

I hope that helps.

Mythos

[DJRumpy]

This thread got busy over the long weekend...

Mythos, if you had to run a registry poke, then you probably have an old 1.X version of AVISynth. The 2.0 versions usually just have a setup.exe to run. Your plugin doesn't have to be in an particular location. When you reference the DLL file, you can specify the full path to it
(example: LoadPlugin("c:\folder\pluginname.dll")
Or you can simply put in somewhere in your PATH, like your Windows, or System folder. Then you can call it directly, with no path:
(example: LoadPlugin("pluginname.dll")

borghe, just to clarify, vhelp is correct regarding the LD rip. There are true interlaced scenes in the video, which will not IVTC properly. These special affects shots really mess with the IVTC filters. I haven't tried the TMPGenc filter, but it claims to handle mixed sources like these. Your mileage may vary, but it might be worth a shot. Unfortunately, I have no idea how they took an interlaced frameset, and telecined it to 29.97. I mean, I know technically how it might work, but you'd end up with an interlaced field set, generated from an interlaced field set. Rather messy in my opinion. I suppose you could use AVISynth to deinterlace these frames, while running IVTC on the rest. It would be a more manual process, but the output would probably look better than IVTC alone could produce. You would have to use the TRIM command to splice your movie into IVTC sections, and Interlaced sections, treating each seperately. This kind of task is where AVISynth really shines. Has anyone tried this?

[borghe]

I wasn't aware that they did this on these telecines.. I Stand corrected. That seems really inane to do when you can just telecine the filme, but whatever.

As for that level of work, hmm.. Actually if I knew WHERE the frames were, it wouldn't be that much weork. It would just turn into a few extra files before the big gathering file. The trick is to find those scenes. Unfortunately my time (and patience) doesn't really facilitate this kind of scanning.

Decomb does have a way to perform this mixed method:

Telecide(guide=1,gthresh=30)
Decimate(mode=3,threshold=1.0)

That should treat the video as mostly film with some video. Maybe I will try encoding (yet again :P ) with decomb setup this way.

[DJRumpy]

I Stand corrected. That seems really inane to do when you can just telecine the filme, but whatever.

8) You've got it exactly. It seems somewhat 'tacked' on. Lucas probably did it to irritate the video community. He's good at that.

[Mythos2002]

I finally got a script to work and was able to play it in Windows Media Player. I used borghe's script with the exception of changing the filenames and directories of the AVISource and Plugin. I also had to cut the Trim command out so it would play. The only problem is that there is a green rectangle at the bottom of the picture and the top is cropped off. I'm getting ready to try my test clip in TMPEGEnc to see what happens. The clip is about 25 seconds long. I'll also try to play with the scripts a little more.

mythos - I ran five separate clips through avisynth and then ran all five scripts through avisynth again. My film assembler script is one line:

I have one question about this. How do you do it? Do you just drag the avs file into TMPEGEnc or open it in TMPEGEnc somehow? Please forgive such a newbie question.

Also, when you capture, how do you handle the audio? Do you select no audio and capture only video and come back and capture the audio later? Thanks.

Mythos

[DJRumpy]

LoadPlugin("Decomb.dll")
AVISource("swotep4s1.avi")
Trim(898,50171) # changes for each capture file
Telecide(guide=0)
Decimate(cycle=5)
LanczosResize(720,360,0,101,720,276) # y offset for cropping may change by a pixel or two from side to side
AddBorders(0,60,0,60)

Removing the TRIM command won't hurt your encode Mythos, as it applies to a specific file for borgh's encode, and not yours. I assume your capture is all in one file anway, so leave it off. The problem your seeing with the image being offset is probably due to a difference in the capture size you used, and the one borghe used. You should modify the LanczosResize params to fix it. You may also need tweak the AddBorders command, depending on your vertical settings. I'm not familiar with LanczosResize, so I don't know the params. If I had to guess, the first is the starting width (720). The second is the starting pixel TOP value (360). The 3rd value is the amount to crop from the left (0). The next value is the amount to crop from the top (101). No idea what the 276 implies, as it's not a direct subtract from 360 (ie. 360-101). Maybe Bourgh can shed some light. You can also try the built in CROP command. It would look something like this: Crop(0,104,720,360)
1st value (0) = amount to discard from left
2nd value (104) = amount to discard from top
3rd value (720) = remaining width (after subtracting 1st value)
4th value(360) = remaining height (after subtracting 2nd value from top)

No idea as to what is causing your green block. How big is it?

I have one question about this. How do you do it? Do you just drag the avs file into TMPEGEnc or open it in TMPEGEnc somehow? Please forgive such a newbie question.

Yes. You can drag/drop your .AVS script into TMPGenc, or open it manaully.

Also, when you capture, how do you handle the audio? Do you select no audio and capture only video and come back and capture the audio later? Thanks.

You normally capture the audio with the video, to avoid sync issues later on. They are easier to handle when captured and converted together.

[borghe]

all of your problems are probably from the LancoszResize. Here is the command:

LancoszReize(width,height,x offset,y offset,width,height)

The first width and height are the new size you want. The x and y offsets along with the second width and height are for cropping the image first.

LancoszResize(720,360,0,104,720,276)

is the same as

Crop(0,104,720,276)
LancoszResize(720,360)

According to the avisynth docs for 2.5x (not sure about earlier), cropping inside of the resize function can offer up to a 25% speed increase versus doing the two functions separately. As for LancoszResize, to my eye it seems to offer the best enlargement for an anamoprhic picture out of the available methods.

The first thing you will need to do is adjust the cropping. The resize will actually stay the same, but your cropped area will be different. What I did was actually loaded each original capture in VirtualDub and used Null filter to preview. I went to the cropping window and checked exactly where the black lines ended. This is where I came up with my numbers. Everything was perfect and from side/capture to the next things deviated 1 pixel at most usually. After I got those numbers from VirtualDub, I jut plugged them into the LancoszResize() function.

As for audio, I captured using my Digital8 camera as an analog pass-thru so my video and audio were captured at the same time into the same file. Consequently, after I finished all of my encoding in TMPEG, I could open the exact same avisynth file I used to encode and just load it into VirtualDub and save the WAV.. Everything matched frame for frame.

The only audio I am capturing separately is the scene specific audio commentary. However that is a little easier. Because my scenes line up virtually frame for frame with the laserdisc, I should be able to trim my audio capture well enough (it doesn't have to be perfect as it isn't "in sync" anyways) to place at the appropriate chapter stops. I was planning on encoding only each individual audio clip at AC3 2.0 128Kbps. this will result in around 10 or so ac3 audio files that I can then drag into another language track. The advantage to it not being a running audio commentray is that I will only end up with 15 minutes or so of voice only encoded at a much lower bitrate. It should only add less than 15MB to my already pressed for space DVD. I only have about 60MB left after the movie, and that's without the much wished for THX intro or the Fox Home Video intro but including all of the credits. So at least I get the commentary from the set in there.

[Mythos2002]

Borghe,
Thanks for the help. I have to take the LancoszResize line out for now because I keep getting an "Invalid video source" message when I try to open my script in TMPENGEnc with the wizard. If I try to open it without the wizard, I get the message, "Floating decimal point is divided by 0". If I leave that line out, it will open in TMPENGEnc ok.

I'm also having another wierd problem. When I go to the IVTC filter, it says that it is at 19.2 fps. I don't know how that happened, because the clip was captured at 29.97 fps and not altered in Virtual Dub. The AVI file even says 29.97 fps if you right click on it and look at its properties.

Any ideas? Thanks.

Well, I just figured something out. If I put AssumeFps (29.97), it stops the 19.2fps from appearing in the IVTC filter. I wonder if I should say AssumeFPS (23.976) or (24).

I have a very important question and I feel pretty dumb about it. I just tried the script again in TMPENGEnc 2 times. I used the lines:
Telecide(guide=1,gthresh=30)
Decimate(mode=2,threshold=1.0) Mode 3 gives a out of range error.
The first time I combined this with TMPENGEnc's IVTC filter and it was a disaster. The second time I omitted the IVTC filter of TMPENGEnc and it was nearly perfect. I think the reason there are a few stutters is because I had the black bars trimmed off in my capture as I was capturing. So, I shouldn't crop the bars off with Virtual Dub just use it to see what numbers to put into the resize command?

My current capture is 720x296 (since I cropped some off of the top and bottom bars) at 29.97fps with Huffyuv 2.1.1. The scene is the first 30 seconds of side 2 of Empire.

Here is my sample script.

LoadPlugin("g:\AVSFilters\Decomb.dll")
avisource("G:\Test\empire2.avi")
Telecide(guide=1,gthresh=30)
Decimate(mode=2,threshold=1.0)
AssumeFPS(23.976)
LanczosResize(720,480)

I'm puzzled by a couple of things. Do I need the AssumeFPS command? Since I have Telecide and Decimate, do I still need TMPENGEnc's IVTC filter? Do I need to set 23.976 fps (29.97 fps internal) in TMPENGEnc as well as 3:2 pulldown during playback or can I skip those two parts since I have the AVISynth script?

Do I need to use DVD2AVI for anything in this project?

Please forgive all of the questions. I've been trying since December to work on this but haven't had the time to read up on it or work on it. Thanks.

Mythos

[DJRumpy]

The divide by 0 error is due to the resize command. Your trying to crop more video than is actually there. You'll need to input the proper values to crop the remaining letterboxing. Given the info Borghe has suppiled:
LancoszReize(width,height,x offset,y offset,width,height)
Also knowing your video is: 720x296
You need to determine how much of the remaining height is letterboxing. If you have only 20 pixels of total letterboxing left (10 pixel top, 10 pixel bottom), and I have the params right for the LanczosResize, then your command would look like this:
LanczosResize(720,480,0,10,720,276)
If that doesn't work, just use the Crop and AddBorders command, which would look like this:
Crop(0,10,720,276)
AddBorders(0,102,0,102)
The difference is speed is minimal. You do know this is a 16:9 movie, but your letterboxing for 4:3? Anyway....

The AssumeFPS command should most certainly NOT be used. Your telecide and decimate commands are at fault. They are removing too many frames:
Telecide(guide=1,gthresh=30)
Decimate(mode=2,threshold=1.0)

The standard IVTC filter config should resolve it:

Telecide(guide=1)
Decimate(cycle=5)

This should work pretty well. It will scan for one duplicate frame (or two fields) in every five (cycle=5). You can change the GUIDE setting (to Guide=0 ) to let the filter ignore the 3:2 pattern, and cause it to simply scan for matches. This is desirable if the pattern changes often. Of course you would probably want to tweak it to get good results on this capture.

For Mixed Mode video (like Star Wars), I might suggest you do not use the guide=1, as it tries to identify the existing 3:2 pulldown pattern, and it can override a matched field if the pattern looks wrong (the pulldown pattern is very much 'all over the place' in this movie). For mixed IVTC/Interlaced clips, try the following:

Telecide(guide=0)
Decimate(mode=1,threshold=1.0)

Note, that gthresh=30 isn't required. It's the default. The threshold param may need to be changed, depending on your clip. Use show=true in the Decimate command to view the returned threshold values so you can set your threshold specifically for your clip. Note, that these options will not convert the framerate of your clip (it will remain 29.97). Per DGraft, this is the suggested configuration for mixed mode clips. I've never tried it, but it may be worth a look.

[borghe]

you have almost everything right djrumpy, although in regards to the decimate command, if he captured at 29.97fps, mode=3 on decimate should work.

If mode=3, instead of discarding the most similar frame, Decimate() will apply different decimation strategies for 3:2 pulldown material (film) and for pure video (nonfilm) cycles. Film cycles are decimated in the normal way. Nonfilm cycles are decimated by applying a frame blending decimation of the cycle from 5 frames to 4 frames. The resulting frame rate is the film rate (decimated by 1/5).

doing it the default way it won't correctly decimate video material as far as I can tell.

[DJRumpy]

Correct. It will leave the output at 29.97 fps. I'm not sure how it treats the progressive frames, but dgraft suggests this one for mixed mode input. Hopefully someone who's used this mode can shed some light.

[Mythos2002]

Thanks for the help.

The difference is speed is minimal. You do know this is a 16:9 movie, but your letterboxing for 4:3? Anyway....

That's not what I was looking for. I'm trying for 16:9.

I found out part of my problem with mode=3. It was AVISynth 1.0 even though I installed Version 2.0.8. I read how to remove the old version and went and installed 2.5.1. I hope it won't cause any problems since it is a beta. I'm going to try the mode=3 variable and see what happens.

Well, it was choppy with:
Telecide(guide=0)
Decimate(mode=3, cycle=5, threshold=1.5)

I'm going to try this later tonight and see what happens:
Telecide(guide=1,gthresh=30)
Decimate(mode=3,threshold=1.0)

I forgot to ask: Since I have this script, do I still need to change the framerate in TMPENGEnc to 23.976fps (29.97fps internal) and use the 3:2 pulldown during playback?

Thanks
Mythos

[vhelp]

Hi everyone.

@ Mythos2002..

Any time you utilize any IVTC in your video processing, in TMPG, you
should alway configure your settings w/ 23.976 and 3:2 pulldown.

I'm comming back w/ some final words on IVTC and Interlaced.. otherwise
known as "mixed" source content.

-vhelp

[vhelp]

@ everyone.

Ok..

I want to be a bit more anal w/ respect to IVTC and mixed frames.. if I may..

When you source is Film (and you know for sure it's Film throughout)
* you should verify that any computer animation; generated; etc.,
(first, that they exist in your source) ..as I was saying..
verifty that THOSE scenes are examined for Interlace !! THis is a must, at
least after you've encoded your source, and then fianlly, realized that you have
issues w/ the playing ie, studders, or jitters etc.

When you have discovered that your final encoded MPEG-2 play w/ these jitters
etc., you should examine your source in THOSE areas that produce these issues
for Interlace. Go back into the .AVI file (not the MPEG-2) and single-step
through the frames for Interlace frames. They can be like this:
* 1i, 1i, 1i ....(1 Interace, 1 Interlace.. ..)
* or other variation ie: 2p, 1i, or: 3p, 2i etc, etc.

The important point, is to find that they are Interlace.
If the source is 24 fps Film, but has mixed in Interlace frames within the
24 fps Film, than you are in for trouble.
When they perform the Telecine process on this 24 fps Film, those Interlaced
frames will be a part of it. That's the tricky part. This is where the IVTC
process fails mostly. Because, some fields will be destroyed or missfired.
Don't forget, that the Telecining process is Interlacing 2 frames. And, guess
what ?? when it comes to those frames that were interlaced, gues what's happening ??
Well, you figure it out. Then, apply that as a factor in your IVTC process.

Ok, but what is your source is Progressive in those scenes. This performs another
problem. Cause, these fields are usually 29.970 for smoothest play. But, they
can also be optimised to be 24 fps. never the less, those scenes that are
progressive will present problems as well, because they throw off the IVTC process.
You see, sometimes, they do the Telecine process on the Movie, but then do the
Computer Animation/generated cuts/pastes into the Telecined Movie AFTER'wards.

So, when you run your source (assuming now, that you source is a DVD) inside
DVD2AVI, and you press the F5 key to scan your source to see what is the nature
of the source ie, Film/Intelace/Progressive.. you may see some switching back
and forth between Interlace Progressive or even Film, depending on your DVD2AVI
settings.

When you encounter scenes w/ Progressive mixed in w/ Interlace (a.k.a., Telecine)
source, these were the last minute edits. And, that's why you see strangeness
in your IVTC pattern. Star Trek TNG (those that are on DVD) does this form of
practice. And, as a result, you can't perform a perfect IVTC (a.k.a smooth play)
Unless you do it for EVERY one of those scenes that were computer generated or
whatever, but you may EVEN run into trouble if they didn't cut/paste exactly at
the IVTC point. ie, if they inserted those pieces at Interlace point from the
Telecined Movie.

It's a comprimise. You have to decide what outweighs what. And, make the
decision to either encode w/ 29.970 fps and higher bitrate, OR, encode w/ an
IVTC and lower bitrate. It's your choice.

The above was some more descission on the issues w/ IVTC and mixed Interlace
frames and processing these as such for smoothest play etc. rada, rada, rada.

Well, that was my two extra cents worth, thus far.
-vhelp

[Mythos2002]

Thanks for all of the help. Is there anyway when you use those show=true variable in AVS scripts to print the results or do you have to sit there and watch each frame and write down the results?

This was the script I last used:
LoadPlugin("g:\decombfilter\Decomb.dll")
avisource("g:\Test\empire52903.avi")
Telecide(guide=1,gthresh=30)
Decimate(mode=3,cycle=5, threshold=1.0)
LanczosResize(720,360,0,104,720,276)

I didn't use the show variables on that one. The bad thing is that the debugging isn't always displayed due to the resize. Some of the information is cut off of the top, so you can't read all of the debugging info.

Anyway, the results seemed nearly perfect. There was a little stuttering, but it wasn't in a special effects shot this time. I'll keep at it though.

Mythos