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

[DJRumpy]

I don't know of a way to output these to file. You're just trying to get an average anyway. No need to be that exact. Also, as far as I can see, there is not Mode=3 setting. It may be defaulting to Mode=0, since you gave it an invalid mode.

You should try some of the other Guide, and Modes, to see if they work for you. Here's a quick rundown from the docs:

guide (0-3, default 0) can be used to greatly improve field matching when the source clip is known to be PAL or NTSC telecined material. To disable this option (blind field matching), set guide=0. For NTSC 24fps->30fps telecine guidance, set guide=1. For simple PAL guidance (tries to maintain lock to the field phase), set guide=2. For NTSC 25fps->30fps telecine guidance, set guide=3.

When this option is enabled, Telecide() can overrule a field match decision and use a predicted match based on the recent clip history. The gthresh parameter (below) is used to define how large a discrepancy between the predicted and calculated field matches is required to reset the pattern. Do not enable this option unless you know that the source clip corresponds to the selected guidance mode. If in doubt, leave guide=0.

mode (0-2, default 0) determines how Decimate() deals with the extra frame in the cycle.

If mode=0, Decimate discards the frame in the cycle determined to be most similar to its predecessor.

If mode=1, instead of discarding the most similar frame, Decimate() will either replace it with a frame interpolated between the current frame (usually a duplicate of the preceding frame) and the following frame, or it will pass the frame through as is. The choice between these two depends on the threshold parameter setting and on how different the frame is from its preceding frame (see below). Decimate(mode=1) is useful for hybrid clips where you do not want to reduce the frame rate but want to ameliorate the effect of duplicate frames that are emitted by Telecide() (frames that are normally removed with mode=0).

If mode=2, Decimate() deletes a frame from the longest run of duplicates. This mode is the most reliable with anime and other material where the motion may occur only in every second, third, or fourth (etc.) frames. If you use mode=0 on such clips, there is a danger that incorrect decimation may occur, causing jerkiness. Clips such as those described usually derive from 8fps or 12fps animation, as well as normal 24fps animation where slow motion results in repeated duplicates. Mode=2 is able to delete the correct duplicates in all these cases.

[DJRumpy]

Just as a follow up. I ran a few tests. Using the mode 1 option just does a pass through of the existing content (framerate will not change). It will IVTC telecined material, and deinterlace interlaced frames. I was suprised at the quality. This option would seem to work well for Star Wars. It will deinterlace the interlace sections, and I assume, IVTC the telecined sections.

Note: My input was 23.976 (already had IVTC run against it), with remaining interlaced content.

LoadPlugin("decomb.dll")

Telecide()
Decimate(mode=1)

I haven't had a chance to run this against a 29.97 non-IVTC source yet. If anyone gets a chance, let me know how it turns out.

[Mythos2002]

What did you use to IVTC it? So far, I haven't been completely successful, but its getting closer. I just have to work on it some more and figure out the best way to reduce noise.

Mythos

[Robster]

I thought I'd post some frame grabs to show the work I've done so far.

The grabs have been taken using PowerDVD from the actual DVD files following TMPGEnc VBR encoding. I still have more I want to try but I'm quite impressed with the results so far. This is from the 4:3 letterbox version. The final one I do will be 16:9.

Pictures 8 and 9 show the permanent subtitles I've added using Vegas Video 3 LE. They're a substantial improvement over the ones on the original LDs which were displayed in the lower letterboxing border.

http://users.surfanytime.co.uk/robgillespie/sw1.jpg
http://users.surfanytime.co.uk/robgillespie/sw2.jpg
http://users.surfanytime.co.uk/robgillespie/sw3.jpg
http://users.surfanytime.co.uk/robgillespie/sw4.jpg
http://users.surfanytime.co.uk/robgillespie/sw5.jpg
http://users.surfanytime.co.uk/robgillespie/sw6.jpg
http://users.surfanytime.co.uk/robgillespie/sw7.jpg
http://users.surfanytime.co.uk/robgillespie/sw8.jpg
http://users.surfanytime.co.uk/robgillespie/sw9.jpg
http://users.surfanytime.co.uk/robgillespie/sw10.jpg
http://users.surfanytime.co.uk/robgillespie/sw11.jpg

[miked00]

You all could save yourselves a lot of trouble if you looked online because there are sites that sell the Star Wars VCD's. These are professional quality discs too. Not sure if they are widescreen though.

[Robster]

VCDs have lower resolution than DVDs and have to split the film over two or three discs.

If you've ever compared a VCD to a well-encoded 'home made' DVD then you'd understand.

[geek rock]

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.

sorry if someone else has replied to this..

the easiest method, surely, is to have a subtitle stream that's blank except for the greedo subtitles, and have it set to "on" from the start of the movie, by default.

-Mark

[vhelp]

@ robster..

D/L'ed all your pics..
First, you pics look great I love seeing peoples results. IMO, it's very important
to see one's hard work demonstrated. Case you don't know, I'm going through a
Monitor phase issue. My NEC died out on me, and I'm using an older one and it's all
messed up (colors/brightness) Am working on getting an Flat-screen this comming week
yeahe !! You know.. at first, I thought.. man, his VHS pics look great, then I forgot,
oh, yeah.., LD's hehe.. hehe.. I had forgotten for a moment. I have some WS movies on
VHS, but not the Star Wars. I'm still waiting for some stores to hopefully carry
them - - k-mart, walmart, bj's etc. They all seem to only carry the FS versions.
Anyways..

burned-in permanent subs on my Star Wars captures using the Video Vegas LE package included with my Canopus card.

So, I gather that your SUBs are permanent then ??

Also, I noticed those pics didn't seem to have any noise in them. The ADVC was THE
RIGHT CHOICE hehe.. aren't you glad..hehe..

Question on IVTC..
Did you get ALL your IVTC issues resolved ??

Samples..
you know.. the next request will be, samples. They don't have to be any thing huge.
Just 30 seconds for a few clips - you know Especially those troublesome areas that
most were talking about, that is, IF you were able to IVTC those w/ no issue, else,
whatever you have is alright w/ us (me) hehe.. Yes, show us some samples of your
accomplishements hehe..

Later on guys.
-vhelp

[Robster]

The subs are indeed permanent. I overlayed them onto the raw avi using Vegas Video 3 LE (which came with the Canopus card). The subtitles in the Star Wars films are as much a part of the film as backgrounds or the effects. I've just put them back where they should be. I really dislike this trend of making all subs generated by the DVD player, though I do understand why ($$$). It's something that has got worse as DVD has become more a mass-market consumer product rather than a film buff's format. Also, player-generated subs look like total crap. Mine don't.

For the test I applied noise reduction with TMPGEnc, but I'm getting slight 'blurr' effects during some motion so I'm trying it again without. The original LDs always had a slight tendency to 'blurr' but I want to try and minimise it as much as possible. The Canopus captures are very nice indeed and I think I made the right choice with that card. It must be said though, that the player makes a huge difference with LaserDisc. My Japanese CLD-HF9G was pretty much at the top of the tree.

Those examples are 4:3 letterbox, but my final versions will be 16:9 enhanced. This doesn't result in any more detail or sharpness, but it does mean that on a widescreen TV you don't get the 'gridline' effect caused by the NTSC scan lines. The end result just looks better.

As for IVTC - well, to be honest most of the discussion here on that subject is going right over my head. I just tick the box for it and it works.

Once I'm happy with the results I will post a couple of short clips.

[steve996]

I thought I'd contribute my experience in converting the Definitive Collection LDs to SVCDs. I'd like to do the DVD route as some of you guys are, but thats another project. My goal has been to convert each movie onto two 80 min CDRs in a XSVCD format 702x480 (w/a 702x272 window) and an anamorphic format 702x480 with a 702x360 window. So far, the 702x272 format looks very good, clean and almost dvd perfect. Going the anamorphic route though, picture sharpness was really no sharper, if not actually a bit softer, with no additional detail, and I saw compression artifacts due to the increased screen resolution at the same bitrate. I used a Lanczos resize to 360 after IVTC in VD to do the conversion. In the end I decided it was not worth it and went back to the 272 format for now. If I had the storage capacity of a DVD, then perhaps it would be worth it, but clarity is more important to me than scan line elimination. Perhaps in a few years when Lucas does decide to release his trilogy on DVD, it will be anamorphic and I'll just buy it then.

After capturing the disks through VirtualDub at 702x272 with the huffy codec, I tried various conversion techniques to maximize the encoding quality. I took the captured avi, used VirtualDub to crop the blanks at the beginning and end of the capture, and then had VD do a IVTC in full uncompressed avi back to the hard drive. Making sure I had no dropped frames in capture, I used a manual offset in the IVTC menu to make sure the process was done correctly. Each LD disk side which is about 30 minutes takes about 22 GB of drive space. I then used Kwag's KDVD template in TMPGEnc, changed the max bitrate of the template to 2520, motion search precision at high quality(slow) mode, and encoded at CQ 77 to fit essentially 61 minutes/disk. Audio is encoded at 224 kbps/44.1/stereo with TooLame. I also applied a noise reduction filter in TMPGEnc set at 15:1:20 to reduce video noise, improving compression and a sharpening filter set at 32:52 to improve sharpness. So far the results are very good, still frames from a Panasonic RP82 in progressive mode on my Sony 61HS10 hdtv shows little to no compression artifacts, the result is almost indiscernable from the original.

When using the CQ mode, and trying to get the max bitrate possible while still fitting half of the movie on one disk, here what I found after many days of experimentation for Episode IV:

CQ 70 for a 704x272 format, audio @ 224, = 10.633 MB/min
CQ 74 for a 704x272 format, audio @ 224, = 11.696 MB/min
CQ 78 for a 704x272 format, audio @ 224, = 13.415 MB/min
CQ 68 for a 704x360 anamorphic, audio @224, = 12.037 MB/min

One of course will have to adjust the CQ value based on the total length of each disk, however a CQ of 74 is actually pretty good already, higher than that is just all the better.

[Mythos2002]

I hope I'm not going to have to spend more money and get a Canopus ADVC card. I can't seem to get the noise out of my encodes and they are a little too dark. I thought it was my monitor settings, but the brightness and contrast are high enough. The picture was also dark when I made a test DVD-RW and played it on my TV.

I've tried several things like brightening the picture, but that seems to show more noise. I've also tried with the Output to Yuv Basic setting under the Quantsize Matrix turned on and off.

As far as the noise, I've been using TMPENGenc's Noise Reduction set on 8 1 10. I've also tried other settings, but they didn't make a lot of difference. Are there better noise reduction programs or filters? My captures look fine except for the combing lines. Its just when I encode that I see noise and the picture being too dark. The only good thing about the darker picture is that you don't see the matte lines around the ships.

So far, my best setting for IVTC have been with using either Virtual Dub or AVISynth. TMPENGenc hasn't done that great of a job for me.

I fixed the noise. Soften block noise works wonders. The picture is not as dark and the noise is almost gone. I used 55 for both settings. The settings were 0 for all the rest of my encodes.

Mythos

This is a little off topic, but I have to ask. Does anyone have the LD's for both the Original and SE versions? You could make some nice hybrid versions of the films by taking out some of the dumb SE stuff and reinserting the original version material. The only problem would be a lot of extra editing and making the picture quality match. It would be nice if there was an encoder that let you change colors within individual frames. I know that would take an extremely long time, but you could remove the matte lines of any of the original material you reinsert.

[DJRumpy]

Try a temporalSmoother in VirtualDub, or AVISynth. They work wonders with analog noise, and they won't soften your output like the 'soften block noise' filter in TMPGenc.

[borghe]

Just an update to where I'm at. I have the film and avisynth script finalized. Now it is just a matter of finding the right bitrate vs. everything that will be on the disc. I was originally just going to have the movie at 4.8Mbps and soundtrack at 192Kbps, but then I felt it was kind of important to have the commentary from the discs. It isn't that great, but if I want to be proud of this, I should really do it right. So I was then going to have the commentary at DD1.0 64Kbps which would have sounded great. Unfortunately I am not a huge fan of DD1.0. Anyways, as I was editing together the soundtrack (and completely reworking their horrendous levels through it) I noticed there was a good 25 minute gap between two commentaries (chapters 38 and 56 I believe). Well, if 25 minutes of silence cost the same as 25 minutes of sound (bitrate-wise that is), I might as well doing something with it. I hunted down and found the 15 minute extended version of the Meco Star Wars disco track from the LP. The problem is if I encode that in 1.0 you'll lose some of the impact. So I instead went with DD2.0 at 128Kbps for the commentary track. The speech sounds fine and the song sounds as good as the MP3 I got it from. I am now trying to encode my movie again at 4.75Mbps to make room for the 128Kbps soundtrack. I figure giving up a miniscule fraction of video quality for the commentary and "such a great tune" is a small sacrifice.

I actually ecoded at this rate once, but in doing so it seems TMPEGEnc stuck the I frames at every 1/2 second, instead of before where it also stuck them at scene changes. I didn't tell TMPEG to close all GOPs and I am wondering if that's why. My first encode was awesome as I was able to get chapter stops inserted very accurately. With the 1/2 second GOPs it is very difficult for me to place the chapter stops to a satisfying degree. arrggh...

On that note, can anyone tell me what is the easiest way to have two computers encoding the same video at the same time and merge them back together? I have two decent PCs (1.4Ghz/768MB and 2.08Ghz/1GB) that I could have working on this at the same time if there was a flawless way to merge the videos afterwards.

[borghe]

oh, for the record, my IVTC section of avisynth looks like:

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

It seems to give me the best quality out of all of the IVTC methods. I do need to file down to 24fps as I am running extremely tight on space with the aforementioned commentary track. The way it is the second movies will probably have commentary in the much despised DD1.0 just so I can cram it down to 64Kbps being that both of them are longer than Star Wars by quite a few minutes.

Anyway, if you are capturing the disc through a DV camcorder, give the above IVTC method a shot. Not flawless, but hardly noticeable and only in a few places, and at the same time giving you more bang for your limited disc space (unless of course you are using two discs for the movie).

[DJRumpy]

I double checked the docs. There is no Mode=3 flag. I'm guessing it defaults to Mode=0, but I'm just guessing.

[borghe]

djrumpy - check my quote on page 7. There is definitely a mode=3 in decimate(). That quote is literally cut&pasted from the html docs in the decomb410b2.zip file. Also, there was a slight improvement on some sequences, most noticeably (just because of how bad they were) the dogfights at the end.

[DJRumpy]

borghe, this is from the 4.05 (legacy) docs:

mode (0-2, default 0) determines how Decimate() deals with the extra frame in the cycle.

Is it possible that the mode 3 is supported in the 2.5x version only?

[borghe]

yeah, I'm using 2.5x.. that is probably the difference.

[DJRumpy]

I resist version 2.5x only because my favorite plugin (mpeg2dec) doesn't include the same TemporalSmoother in the 2.5 plugin versions of mpeg2dec. Do you know of any versions of mpeg2dec that still include the 2.0 temporalsmoother, or equivelent? I don't like the temporal/spatial filters that come with AVISynth 2.0. Much too slow.

[Robster]

Guys, can you give me a quick rundown on how to use this Temporalsmooth thing?

I've got a 3min clip from my SW capture that I'm using to test on before committing a final encode in TMPGEnc Plus. I've got the AVI open in VirtualDub, I've set the temporalsmooth on with a setting of 5 (just to try it), save the file as AVI and instead of the 600mb file I started with I now have one almost 5gb in size!

[Robster]

Uhh - think I figured it out. You have to use it via frameserving - correct?

Anyway, tried it and about half way through TMPGEnc reported a read error. Arse!

[Mythos2002]

Anyway, tried it and about half way through TMPGEnc reported a read error. Arse!

That was a problem with TMPENGenc Plus version 2.511. Here is what it says on the Pegasys website:

"In an environment where multi-thread was enabled, choosing "Highest Quality" as Motion Search Precision would cause the encoding to stop halfway with an error saying Memory access violation error."

I ran into that problem several times when I upgraded to that version. Thankfully, that problem has been corrected with 2.512.

Mythos

[Robster]

I'm on 2.512!

This only happens when I tried frameserving. I've let the encode run overnight using TMPGEnc alone and it's completed fine.

[DJRumpy]

Try the temporalsmoother included with MPEG2DEC. It only has two arguments (this requires an AVISynth script).

LoadPlugin("mpeg2dec.dll")
AVISource("c:\blah\hablahblah.avi")
temporalsmoother(2,3)

The first param is radius, and the second is strength. Or it could be the other way around. Just read the docs. I never take either above 3, as you could start to see banding in the color ranges. This is part of mpeg2dec.dll (AVISynth 2.0)

[DJRumpy]

A follow up on the Mode 1 option. It does nothing with the progressive frames. It only deinterlaces the telecined frames using a Blend mode. The fps will not change. This would give you decent looking output for Star Wars, on the special effects shots that are true interlaced. You will lose the advantage of 23.976 fps, as compared to 29fps though.

Telecide()
Decimate(mode=1)

[borghe]

here's a new question. So I'm going through and inserting as many of my own I frames as possible. For the most part TMPEG has done a good job at detecting scene changes. However the one place it will always get hung up is a scene cut (not a wipe) that Decomb doesn't IVTC properly. It looks to be one field from the past scene and one field from the future scene. Consequently Decomb treats this transition frame as a new scene. What I am worried about is a) that my scene change will be off of this goofy frame, and b) that it will mess with my compression (even if only to a small degree). If I put the goofy frame at the end of a GOP it will probably get murdered (if only for a split second), and if I put it at the beginning of a GOP it will taint the B pictures directly after it.

So how do I correct that. I see two ways. One is if I can tell TMPEG to make it a P frame copy based on a previous I,P frame and then make the frame immediately before it a P frame. This seems like maybe the easier way and should get me good results on average motion scenes. The other way is to figure out a way to tell Decomb to treat these frames differently than it currently is. I have no idea how I would do this.

As it is, I am putting the I frame on the full frame and not the goofy split field frame. I want my chapter stops perfect and my GOPs to be as clean as possible. Give the bad frame to the end of a GOP I say.. Any thoughts?

[borghe]

screw it... I went back to mode=0. All of my bad mixed field frames magically disappeared going with mode=0. Unfortunately I am doing this right now through terminal services so I can't actually check the video on some scenes, like my supposedly noticing the IVTC of video at the end during the space battle, but I will check it when I get home. In the meantime I can say that all of my problems as mentioned above have gone away with mode=0. Also in looking through quite a bit of the raw capture (hundreds of frames), I couldn't find anything that fell out of 3:2. Will do some more work when I get home on the PC.... For the meantime though, back to mode=0.

[borghe]

Ok, one last question for a while, this one for DJRumpy and vhelp. this is in regards to the telecine process on The Definitive Collection in general. Was the Definitive Collection even telecined? The CAV discs store individual frames, and the frame rate is at 24fps. This can be seen by the counter moving and considerably different speeds on the laserdisc and Premiere when it captures (premiere running at 29.97fps). So you have progressive frames on the laserdisc running at 24fps. Where did the telecine happen? This has been bothering me for a bit now. A CAV laserdisc will either run at 29.97fps or 24fps. We can tell that Star Wars is running at 24fps. So the film isn't telecined on the disc. We also know that as part of the laserdisc spec that the players effectively "telecine" 24fps CAV discs (CLV discs are always telecined and at 29.97fps). So my question is, how does this video material you are talking about make it onto the laserdisc? They would have to effectively do an inverse 3:2 pulldown of the video material to put it as full frames on the laserdisc. Either way, an IVTC would technically restore the video back to the actual 24fps progressive video of the laserdisc.

If any of this is wrong, please let me know. The only things I am certain on are how laserdiscs work. This transfer is still confusing to me, but as far as I can figure, the transfer on the laserdisc is still effectively 24p, so it would seem to me that any video material would be moot (and thus not cause jerkiness) as it would have had to have been interpolated to 24fps anyway.

Ok, back to the conversion.

[DJRumpy]

So my question is, how does this video material you are talking about make it onto the laserdisc? They would have to effectively do an inverse 3:2 pulldown of the video material to put it as full frames on the laserdisc. Either way, an IVTC would technically restore the video back to the actual 24fps progressive video of the laserdisc.

I'm not quite sure what your asking here. Are you wondering how they got these interlaced segments into a 24fps disk?

I can only assume that the original theatrical release is all 24fps progressive. I've hit a simular problem on the Oz - Season One DVD. It has an irregular pulldown pattern (it's telecined from 23.976). Strange thing is, that the pulldown markers, when ignored by DVD2AVI, still gave me interlaced frames. Of course non of this is visible on a television. I've never had this happen with any rip. When you use the Force Film option, it should ignore the telecine flags altogether, giving me a 23.976 progressive feed. I finally had to leave it at 29.97 for the rip, and then run an IVTC filter after the fact (this too left jerks in the video). I'm assuming Star Wars would be a similar situation. I'm still puzzled about this, even after 4 days of examing the video feed.

[borghe]

but is the OZ video encoded at 23.97 w/pulldown flags or 29.97? The Star Wars laserdisc is encoded at 24p, the laserdisc player does the telecine.

I am not asking how they got the video material on to film. I am asking how if I "put" the laserdisc video back to the 24fps that it originated at (the LD, not the film), how it will appear jerky on the computer even though it doesn't on the laserdisc. For all intents and purpose the LD was never telecined, so there was never any time to put the 29.97 video material in there, at least without taking it down to 24p first.

If OZ is at 24fps and you have those kind of problems, that's just messed up.. sorry. but if it's truly at 29.97 then they did just what has been said here. they telecined 24fps and stuck some video sequences in. given that there target display was 29.97fps, I would imagine this was the case.