Sign In

Original Trilogy "Faces" LD Preservation & Upscale project

Author
Time
 (Edited)

I’m not the first person to attempt this, but I wanted to try my own hand at these while I’ve also been working on some other LaserDiscs.

I will be completing the entire original trilogy. I will also be including the exclusive bonus feature interview with George Lucas that was available on these discs, also in 1080p.

The films are presented in the “correct” aspect ratio @ 23.976fps with lossless PCM audio (44.1hz 16bit) in MKV format.

I am also releasing a BD-25 of each film (that has the audio re-encoded @ 48Hz for compatibility).

Version 2.0 of Star Wars (Episode 4) is now available.
Version 2.0 of Episode V is now available. (FINAL RELEASE)
Version 1.0 of Episode VI is now available.

Author
Time

Check out JawsTDS’ excellent Return of the GOUT upscale. If nothing else, use it as a reference to make sure your results are meeting or exceeding the quality there.

Preferred Saga:
PT: Hal9000
4/5: Adywan
6: NSP
7/8: Hal9000

Author
Time
 (Edited)

nightstalkerpoet said:

Check out JawsTDS’ excellent Return of the GOUT upscale. If nothing else, use it as a reference to make sure your results are meeting or exceeding the quality there.

Author
Time

I realized I made a goof and encoded this without dropping the tail off from disc two and encoding with just the LD audio track. I am re-encoding it now for a 1.1 version that is actually a seamless movie. My mistake, stay tuned for updates.

Author
Time

V1.1 of ANH is available

ESB has been captured and is in post-processing currently

Author
Time

If you’re going to upscale to 1080p, I’d use 23.97 as your frame rate, not 60p (59.94). You can use AVISynth to IVTC the 3:2 pulldown 29.97fps back to the original frame rate, or at least the original frame rate when telecined. The true original is 24fps.

Author
Time

clashradio said:

If you’re going to upscale to 1080p, I’d use 23.97 as your frame rate, not 60p (59.94). You can use AVISynth to IVTC the 3:2 pulldown 29.97fps back to the original frame rate, or at least the original frame rate when telecined. The true original is 24fps.

Currently, I’m using 60 on purpose because of the interlaced signal from the LD. This allows for a substantially cleaner image to be captured.

I have considered re-encoding it back down to 23.97 after post-processing the image

Author
Time

clashradio said:

You can use AVISynth to IVTC the 3:2 pulldown 29.97fps back to the original frame rate, or at least the original frame rate when telecined. The true original is 24fps.

Author
Time

ZigZig said:

clashradio said:

You can use AVISynth to IVTC the 3:2 pulldown 29.97fps back to the original frame rate, or at least the original frame rate when telecined. The true original is 24fps.

Ok… there is a misunderstanding here of some form. I’m confused as to what you both mean. I certainly understand that I can use AVISynth to IVTC a 3:2 pulldown 29.97fps back to 24fps. But I am not working with 29.97 content per say.

I guess where I’m at is - I’m capturing at 60fps because the content is natively 480i/60 (60 being the field rate, not the frame rate). I understand the frame rate of the movie was originally 24fps on film and then 29.97fps on the disc, but the field rate of the image is what I am concerned with, the LD content is displayed at 60Hz.

The disc is playing at 29.97fps, but that image is flickering at 60hz. I’m capturing the entire flicker by capturing at 60fps, then post-processing to remove flicker/interlace. I understand that functionally what I am left with is a frankenstined 30fps content with the frames more or less doubled and being played back at 60fps. However, I have found that because of the 60Hz rate that this was originally outputted from the LD player at, that this leaves me with a much better image. It seems to me that this has prevented a great deal of the “smear” effect people describe in LD captures.

I understand there a numerous different ways to work with analog interlaced signal, I am not understanding what benefit would be derived from taking this content back down to 24fps - 60 is to ensure the image remains as smooth as possible.

If I’m misunderstanding what you’re suggesting please explain why - I do want to offer the highest quality image possible.

Author
Time
 (Edited)

Interlaced means that every 60th of a second, half the image is shown. The first 60th shows the odd lines of the image, the second 60th shows all the even lines. So basically, you have 30 full images a second. But you know that already.

As the movie was shot at 24 fps (not 30 fps), the “trick” to transform 24 fps into 30 fps (so that it can be carried on NTSC) is by doubling (pulling up) some frames, which you can easily remove (“pull down”) with IVTC (in AVISynth or even VirtualDub).

You won’t get additional information, image or quality by keeping those pulled up frames.

Author
Time
 (Edited)

Since the 3:2 pulldown is completely reversible (unlike true interlaced video) I think you’ll find converting your 60fps capture back into frame accurate (capable of being GOUT Sync’d) full frames at 23.976 fps with no interlacing and no duplicates is much harder than if you just capture it at 29.97 and then IVTC it with a single line in an AVISynth script:

AVISource(“c:\path\to\my_29_97_LD_Capture.avi”).Tfm().tdecimate()

Or simply open it in Virtualdub, go to Video -> Filters -> Add -> IVTC and choose Reduce Frame Rate, Auto, Adaptive, Click OK and then save your file.

http://www.thestarwarstrilogy.com

http://www.the007dossier.com

Author
Time
 (Edited)

ZigZig said:

Interlaced means that every 60th of a second, half the image is shown. The first 60th shows the odd lines of the image, the second 60th shows all the even lines. So basically, you have 30 full images a second. But you know that already.

As the movie was shot at 24 fps (not 30 fps), the “trick” to transform 24 fps into 30 fps (so that it can be carried on NTSC) is by doubling (pulling up) some frames, which you can easily remove (“pull down”) with IVTC (in AVISynth or even VirtualDub).

You won’t get additional information, image or quality by keeping those pulled up frames.

I understand that and I don’t disagree with you - I’m suggesting only that by capturing both halves and generating product at 60fps, the end result appears with less blur, and little to no combing.

I know that the pulled up frames from 24 - 30 doesn’t give us any additional quality, but I believe there is a significant increase in combing if this is captured at 30fps.

Author
Time

I think Williarob was able to explain it better than I could. It is best to capture at the original source of the media: 480i @ 29.97fps. I don’t think real-time capture software can IVTC as good as AVISynth in post.

For video-sourced material maybe capturing at 60fps is good, but if I had to guess I’d say for film-sourced material capturing at 480i 29.97 is best. The 24fps movie is slowed down to 23.97 during the telecine process. Thankfully there are programs out there that can detect 3:2 pull-down pattern and get rid of the extra fields.

Author
Time

Williarob said:

Since the 3:2 pulldown is completely reversible (unlike true interlaced video) I think you’ll find converting your 60fps capture back into frame accurate (capable of being GOUT Sync’d) full frames at 23.976 fps with no interlacing and no duplicates is much harder than if you just capture it at 29.97 and then IVTC it with a single line in an AVISynth script:

AVISource(“c:\path\to\my_29_97_LD_Capture.avi”).Tfm().tdecimate()

Or simply open it in Virtualdub, go to Video -> Filters -> Add -> IVTC and choose Reduce Frame Rate, Auto, Adaptive, Click OK and then save your file.

I’ll give this a try and see what I can come up with

Author
Time
 (Edited)

This site was pretty much built on capturing these Star Wars Laserdiscs, so most of us old timers here have some experience. I captured my first mismatched OT laserdisc set in 2002. The Empire Disc was from the faces set, while the other two were from the Widescreen Collection. At the time, my captures were no where near as good as the ones being done here by Dr Gonzo, Cowclops, Citizen, etc, but nearly 20 years (and 5 LD Players) later I’m still preserving laserdisc titles to this day.

Imgur

Imgur

http://www.thestarwarstrilogy.com

http://www.the007dossier.com

Author
Time

Love it! I’ve got a whole lot of LDs that I’d like to preserve and I appreciate your help.

This is what I’m working with so far, and I’m continuing to improve

Imgur

Imgur

Author
Time

@ riftamos: it looks like your aspect ratio is too wide on these two screen caps.

@ willirob: what do you use for your comb filter? I’ve been using the EVAL adv 7842 board along with a HLD-X0. But have been reading about the Domesday Duplicator. Since it taps into the RF signal, it can make any low or average player give great results. I thought I had the best set-up by using the best player and best comb filter. I think it’s more geared toward PAL stuff at the moment.

Author
Time
 (Edited)

clashradio said:

@ riftamos: it looks like your aspect ratio is too wide on these two screen caps.

@ willirob: what do you use for your comb filter? I’ve been using the EVAL adv 7842 board along with a HLD-X0. But have been reading about the Domesday Duplicator. Since it taps into the RF signal, it can make any low or average player give great results. I thought I had the best set-up by using the best player and best comb filter. I think it’s more geared toward PAL stuff at the moment.

Yes, I agree - those are just a couple frames from the raw capture file. I’ve got to squeeze it back to the right size.

The Domesday Duplicator looks awesome, over on LDDB there was someone trying to get a bunch of them produced but I don’t think they ever got the project off the ground. Have to build one of these from scratch?

Author
Time
 (Edited)

I have a CLD-99 Which has a built in comb filter which can be adjusted or disabled via the remote. My capture card (an Osprey 100e) also has a configurable comb filter, I have an Extron YCS 100, an NEC Scan Converter IDC-3000, a DVDO iScan VP-50, a Faroudia digital video processor NR series, an Extron DVS 304 and finally I have a Toshiba DVD Recorder with an excellent comb filter that I can use to pass through the signal to the capture card. I’ve tried them all and they all have their own pros and cons. For example, with the comb filter on, the osprey card can completely remove the rainbowing, at the cost of introducing little checkerboard patterns of noise:

Imgur

So what I should probably do is capture each disc a few times via different routes and use TOOT (Two out of three) to average them… But what I usually do is just capture them once and then deal with the issues using software, effects and filters.

My reason for capturing the discs is that I am usually more interested in the PCM audio tracks, or using the video track as a color reference.

Usually there is a DVD or Bluray with much better picture quality available, so maximizing laserdisc video quality is rarely a priority for me. I just try to make it look presentable.

For rare titles, only available on laserdisc, like the US theatrical cut of Highlander, or the Star Wars trilogy, I seek out 35mm or 16mm film versions for the picture, and use the laserdiscs just for audio and color reference 😃

http://www.thestarwarstrilogy.com

http://www.the007dossier.com

Author
Time

I have the Osprey 827e cap card which allows for simultaneously video and [digital] audio captures. Ya I only cap things that aren’t on dvd/BD like Star Wars, The Keep, THX-1138. Or like you I’ll cap for color reference and use the BD as my edit video source.

My next project will be cleaning up the Grindhouse BD of ESB (color correction, noise removal, stabilization). I’ll use the digital audio track from the 1987 Japanese Special Collection LD.

For the TOOT method, do you just overlay the video tracks and adjust the opacity? What is the advantage?

Author
Time

I believe TOOT analyzes every pixel and whenever Two Out Of Three versions of the same pixel are the same and one is different, the one that is different will be ignored. That way, any random drop outs or anaomlies unique to each capture (and not on the original source) are removed.

So capturing the same disc 3 times, or capturing multiple copies of the same title can yield cleaner results. I haven’t used it in years, but I do have multiple copies of the same 16mm Moonraker TV Spot, which I plan to scan and run through TOOT to see how well it cleans them up. Obviously, unlike with laserdisc capture, the scans will need to be stabilized, color matched and aligned with each other before trying this, so that’s an extra step and may mean it can never find 2/3 matches… Worth a try though!

http://www.thestarwarstrilogy.com

http://www.the007dossier.com

Author
Time

@ Williarob: Not sure what NLE you use but for Vegas Pro, the top video track is what will be the main image. I’m guessing you have to adjust the opacity of each track? or is there some type of averaging of the three files?