Dragon’s Lair Limited Edition from 2002 - Preservation and Restoration Project

[The Thinker]

 

Dragon's Lair Limited Edition: https://mega.nz/folder/HoZSCL6S#fgXN5-FMnWTSd0vCUd8ETw

 

 

 

 

Dragon's Lair Limited Edition - Restored: https://mega.nz/folder/OlARAY4Q#cOzmwdxRxVe0SCRfKnWjtQ

 

 

---

🎞️ Frame and Duration Verification

File	Duration (s)	Frame Count	Frame Rate
DLE-NTSC-Side1.mkv	1853.318	55544	29.97 fps
DLE-NTSC-Restored.mpg	1853.318	55544	29.97 fps

 

 

File	Display Aspect Ratio	Sample Aspect Ratio	Active Pixels	Visible Image
DLE-NTSC-Side1.mkv	4:3 (≈1.33)	352:413	760	Full analog scan captured
DLE-NTSC-Restored.mpg	4:3 (≈1.33)	8:9	720	Same image, mathematically resized

 

 

 

I captured the Dragon’s Lair Limited Edition (2002) LaserDisc directly to an uncompressed YUV4MPEG2 (.y4m) master using a high-quality capture chain.
The raw file header reads:

YUV4MPEG2 W760 H488 F30000:1001 It A352:413 C444p16 XCOLORRANGE=LIMITED

This means the footage was recorded at 760×488 resolution — the full NTSC active image, preserving the entire analog scan area beyond the standard 720×480.
The frame rate is 29.97 fps (30000/1001), interlaced top-field-first, exactly matching the LaserDisc’s NTSC cadence.

The most important detail is the C444p16 format, which indicates 4:4:4 chroma sampling at 16-bit precision per channel.
This means every pixel contains full color information with no chroma subsampling (unlike 4:2:2 or 4:2:0 formats).
While the original LaserDisc signal itself is effectively limited to about 4:2:2 color bandwidth, capturing in 4:4:4 ensures that no additional loss or rounding occurred during digitization.

I also used broadcast-safe limited range (16–235), consistent with NTSC standards, and retained the interlaced field structure, which keeps motion and transitions authentic to the original arcade timing.

This approach provides an archival-grade digital master, ideal for future restoration or color correction.
From this .y4m file, I later created lossless FFV1 and visually lossless MPEG-2 versions for compatibility and distribution.

 

Modifié lundi à 11:41 par The Thinker

[ducon2016]

Can you share some of the captures, I can try few upscaling and cleaning algorithms to see which ones works best. The signal is pretty bad, wondering if something what would work best.

 

CUGAN

 

ESRGAN

 

RealSR

 

SRMD

 

Waifu2x

 

Modifié le 9 octobre par ducon2016

[ultimatepump]

Is this from the 20th Anniversary DVD box set?

[CharlieChan]

2 hours ago, ultimatepump said:

Is this from the 20th Anniversary DVD box set?

 

It's available from the laserdisc DomesDay rip here: https://archive.org/details/DomesDay_DragonsLairLE

 

Played here:

 

[ultimatepump]

1 minute ago, CharlieChan said:

 

It's available from the laserdisc DomesDay rip here: https://archive.org/details/DomesDay_DragonsLairLE

 

Played here:

 

Oh ok, just wondering because I have the physical anniversary discs and blu rays.

[ducon2016]

But is has been encoded with mpeg2 already. Is there a "lossless" laserdisc rip anywhere?

[CharlieChan]

17 minutes ago, ducon2016 said:

But is has been encoded with mpeg2 already. Is there a "lossless" laserdisc rip anywhere?

 

Where there is a DomeDay RIP there is a DomesDay LDF source.

 

30s of searching and, as if by magic:

https://archive.org/details/dragons-lair-enhanced-limited-edition-laserdisc-ntsc

 

The source is noted on the MPEG2 page.

[TheThinker]

4 hours ago, CharlieChan said:

 

It's available from the laserdisc DomesDay rip here: https://archive.org/details/DomesDay_DragonsLairLE

 

Played here:

 

The Domesday one was slightly cropped.  
I generated mine from the YUV4MPEG (.y4m) capture of the LaserDisc, and it was converted losslessly and without cropping.

I’ll upload some the files on Archive.org and update my post. 😎

[ducon2016]

6 hours ago, CharlieChan said:

Where there is a DomeDay RIP there is a DomesDay LDF source.

 

30s of searching and, as if by magic:

https://archive.org/details/dragons-lair-enhanced-limited-edition-laserdisc-ntsc

 

The source is noted on the MPEG2 page.

 

Thanks, I actually searched for hours the other day when I saw the 30+ GB discs for the NEC PAC systems. Except I did the wrong search and looked for .mmi discs. I guess I am really getting old. lol.

[TheThinker]

I am uploading the MKV and WAV files of the original prototype to Archive.org, so there won’t be any need to convert the LaserDisc to a lossless digital format again. It’s taking several hours, but once it’s done, we’ll have a solid base for quality enhancement, and I’ll share my improved version afterward.

[CharlieChan]

 

You mean you ran it through "ld-decode --NTSC --V4300D_notch_filter" to produce a .mkv as described on the IA LDF source page.

 

Well I guess that'll save a few decode hours for anyone with a potato PC, although the time to download it will probably make up for that - 🤣.

 

Look forward to the end "enhanced" results, which should ultimately be "playable".

[ducon2016]

2 hours ago, TheThinker said:

I am uploading the MKV and WAV files of the original prototype to Archive.org, so there won’t be any need to convert the LaserDisc to a lossless digital format again. It’s taking several hours, but once it’s done, we’ll have a solid base for quality enhancement, and I’ll share my improved version afterward.

 

The decoding has "skips", so we might need to overlap a few laserdisc captures before there are no skip issues and the capture is real perfect and "lossless". There are videos on youtube showing how some people did that. Right now this capture is the best available, but until we check degradation and others we don't know how close to "lossless" it is.

[The Thinker]

2 hours ago, CharlieChan said:

 

You mean you ran it through "ld-decode --NTSC --V4300D_notch_filter" to produce a .mkv as described on the IA LDF source page.

 

Well I guess that'll save a few decode hours for anyone with a potato PC, although the time to download it will probably make up for that - 🤣.

 

Look forward to the end "enhanced" results, which should ultimately be "playable".

That's correct. I chose MKV to best preserve everything and then we can go from there.

 

---

 

40 minutes ago, ducon2016 said:

 

The decoding has "skips", so we might need to overlap a few laserdisc captures before there are no skip issues and the capture is real perfect and "lossless". There are videos on youtube showing how some people did that. Right now this capture is the best available, but until we check degradation and others we don't know how close to "lossless" it is.

Property	Result	Verified
Frame rate	29.97 (30000/1001)	✅
Resolution	760×488 full raster	✅
Aspect ratio	4:3	✅
Field order	Progressive	✅
Colorimetry	SMPTE-170M	✅
Codec	FFV1 (lossless)	✅
Audio	PCM 44.1 kHz 16-bit stereo	✅
Frame count	55,544 (matches original)	✅
Duration	1853.3 s (≈ 30:53)	✅
Crop/scale	None	✅
Sync	Perfect	✅

✅ MKV and WAV files are fully verified, authentic, lossless transfers from the original Y4M and audio captures.
 

Modifié le 10 octobre par The Thinker

[CharlieChan]

20 minutes ago, ducon2016 said:

 

The decoding has "skips", so we might need to overlap a few laserdisc captures before there are no skip issues and the capture is real perfect and "lossless". There are videos on youtube showing how some people did that. Right now this capture is the best available, but until we check degradation and others we don't know how close to "lossless" it is.

 

I think you are talking about "drop-outs" rather than "skips". Unless your ld-decode methodology is flawed or underpowered. And there are certainly "drop-outs" in this capture, you can clearly see them in the video I posted above.

 

In order to attempt to remedy this, the (DomesDay) process is called "stacking" where you basically "compare" multiple disk captures (you really need a minimum of 4 captures, from 4 different disks) and if a defect is detected, the other members of the "stack" are checked. If the defect is on all of them, it's a mastering error, if it's not then it's a capture defect and is discarded from that source and the other source members are used for that frame.

 

Ultimately it's all about maintaining the best "perfect" frames and framerate to ensure your end result is a playable media on real hardware, or in true emulation. The current single capture seems to play fine and is frame accurate in the MPEG as far as I am able to tell on all the scene skips (i.e. no "skips" as frames are accurate).

 

It may be a little premature to start working from a single exported LDF sources, with lossless conversions (it's actually lossy) from radio frequency analog sources to digital MKV. For stacking to occur, you will always have to refer back to the LDF/TBC RF captures. As only 400 copies of this LE disk were produced and who knows how may no-rotted copies remain, it might be a long or even fruitless wait.

 

If you just wanna work from "the best we currently have" then sure, use the current single source, ultimately you are altering the source material with your cleanup in any case, so no purist is ever going to want to use it, lossless or not, and it certainly won't be accepted in MAME. But in lesser emulators it should be a great addition.

 

DomesDay documentation is very well written.

[The Thinker]

I have uploaded the files  😎

 

I’m not an expert with Hypseus, so please test to make sure everything works properly. If the frame file needs any adjustments, kindly fix it and upload the updated version.

Modifié le 11 octobre par The Thinker

[Max0011]

A friend send me a link to this post, good to know there still work being done to preserver this files, I have been working on restoring the prototype files

myself... is a work in progress - so far manage to recreate 2 scenes.

 

 

 

 

 

 

This year I also started to fix some of the mistakes ... enjoy:

 

 

 

 

 

 

 

 

 

[TheThinker]

4 hours ago, Max0011 said:

A friend send me a link to this post, good to know there still work being done to preserver this files, I have been working on restoring the prototype files

myself... is a work in progress - so far manage to recreate 2 scenes.

 

 

 

 

 

 

This year I also started to fix some of the mistakes ... enjoy:

 

 

 

 

 

 

 

 

 

 

Well done, Max0011 — it’s a great start!

I’ve opened this thread to everyone interested in contributing to the enhancement and preservation of the prototype.

 

I highly recommend that any future restoration, enhancement, or re-encoding — such as the DLE-NTSC-Restored.mpg version — should originate from DLE-NTSC-Side1.mkv (and its corresponding WAV).

 

This ensures consistency and preserves the maximum visual and temporal fidelity of the 2002 Dragon’s Lair Limited Edition LaserDisc, including its original analog artifacts, frame cadence, and full picture area.

 

Modifié le 11 octobre par TheThinker

[Max0011]

Currently working on a HD Hypseus version of DL, restored rip from the Blu-Ray, enhanced video, edited video and sound, thinking to do a prototype version next year for Hypseus.

 

 

 

 

 

[ducon2016]

15 hours ago, TheThinker said:

 

Well done, Max0011 — it’s a great start!

I’ve opened this thread to everyone interested in contributing to the enhancement and preservation of the prototype.

 

I highly recommend that any future restoration, enhancement, or re-encoding — such as the DLE-NTSC-Restored.mpg version — should originate from DLE-NTSC-Side1.mkv (and its corresponding WAV).

 

This ensures consistency and preserves the maximum visual and temporal fidelity of the 2002 Dragon’s Lair Limited Edition LaserDisc, including its original analog artifacts, frame cadence, and full picture area.

 

 

Ouch YUV4MPEG2 was used. Was C444 used or not? If not it needs to be redone.

 

OK will check it out. I want to run state of the art super resolution and frame interpolation on it. Some people tried but there were issues with the other versions that were posted here with some weird cases breaking. I want a 100% reliable version.

Modifié le 12 octobre par ducon2016

[CharlieChan]

Not only YUV4MPEG2, but the decode also contains the pseudo lead-in from the presumed spin-up on the LDF capture, which means the starting frames are way off and need to be edited out.

 

There are 55544 frames in your decode, whereas the disk only should have 55387, so you've picked up 157 invalid frames at the start of the mkv that completely mess up scene alignments.

[ducon2016]

4 hours ago, CharlieChan said:

Not only YUV4MPEG2, but the decode also contains the pseudo lead-in from the presumed spin-up on the LDF capture, which means the starting frames are way off and need to be edited out.

 

There are 55544 frames in your decode, whereas the disk only should have 55387, so you've picked up 157 invalid frames at the start of the mkv that completely mess up scene alignments.

 

Can we just get rid of the 157 first frames?

 

Also it is hard to believe this is the best capture of the original. The fringing and ghosting seems abnormal. And the bottom is messed up too. Is it a limit of the current LDF capture process or the conversion afterwards?

 

 

[CharlieChan]

3 hours ago, ducon2016 said:

 

Can we just get rid of the 157 first frames?

 

Also it is hard to believe this is the best capture of the original. The fringing and ghosting seems abnormal. And the bottom is messed up too. Is it a limit of the current LDF capture process or the conversion afterwards?

 

You can do want you want in the transfer, include disregarding the lead-in frames.

 

The 8 minutes of "prototype" footage was discarded decades ago and only (re)discovered on VHS for the Limited Edition, that VHS footage was tagged onto the end of the original laserdisc. Hence it looks like, well a VHS capture. Because it is. No other known capture exists as yet, I believe, it was used to reconstruct the early game as detailed on the DLP:

 

https://www.dragons-lair-project.com/tech/enhancements/dle2x_prototype_mode.asp

 

 

Modifié dimanche à 08:49 par CharlieChan

[CharlieChan]

Here's another interesting source I just found. https://archive.org/details/youtube-HMtN7cIlMYk   But it's already MP4 from 2 known existing prototype laserdiscs (alleged).

 

Now this gets interesting, as the frames/scenes may not necessarily be in the same order as the Limited Edition footage. But parts of this could be used to reconstruct the prototype footage.

 

But it's gonna take lots of time and dedication to match it to the LE frames, which will be required in order to play on the dle21 ROM from the DLP site. Note this ROM plays in Daphne and Hypseus, not just Dexter as the page states.

 

You wanted a restoration task, well here it is... Let's see who actually picks it up.

[ducon2016]

10 hours ago, CharlieChan said:

You can do want you want in the transfer, include disregarding the lead-in frames.

 

The 8 minutes of "prototype" footage was discarded decades ago and only (re)discovered on VHS for the Limited Edition, that VHS footage was tagged onto the end of the original laserdisc. Hence it looks like, well a VHS capture. Because it is. No other known capture exists as yet, I believe, it was used to reconstruct the early game as detailed on the DLP:

 

https://www.dragons-lair-project.com/tech/enhancements/dle2x_prototype_mode.asp

 

Not asking what I want, but asking for a solution since you seem knowledgeable. Should we cut the first 157 for the video to be usable as is with hypseus for example?

 

8 hours ago, CharlieChan said:

Here's another interesting source I just found. https://archive.org/details/youtube-HMtN7cIlMYk   But it's already MP4 from 2 known existing prototype laserdiscs (alleged).

 

Now this gets interesting, as the frames/scenes may not necessarily be in the same order as the Limited Edition footage. But parts of this could be used to reconstruct the prototype footage.

 

But it's gonna take lots of time and dedication to match it to the LE frames, which will be required in order to play on the dle21 ROM from the DLP site. Note this ROM plays in Daphne and Hypseus, not just Dexter as the page states.

 

You wanted a restoration task, well here it is... Let's see who actually picks it up.

 

The compression artefacts are too heavy and there is luma clamping, so will not touch those sources. I think what @TheThinker posted is a good start if the last 8 minutes of the videos are what is expected.

 

Haha, I wanted a restoration task for AI. Not for me. Let's see which is stonger of the VHS crappy capture or the overhyped AI 😄

[CharlieChan]

24 minutes ago, ducon2016 said:

Not asking what I want, but asking for a solution since you seem knowledgeable. Should we cut the first 157 for the video to be usable as is with hypseus for example?

 

Having experimented with creating video for Daphne/Hypseus in the past, it's all about frame accuracy and maintaining FPS. So cropping the (it appears) first 157 frames of video, and the appropriate edit to the audio, (you can perhaps use the existing .ogg file as it should match), then it should (in theory) work in Hypseus. Basically look to match the existing MPEG2 video frames, as that IA MPEG2 seems to have been done by someone who figured it out.

 

There are ffmpeg commands on how to correctly encode the .m2v and .ogg for Daphne and Hypseus on the Hypseus game LUA GitHub pages. So that would be my starting point. But ultimately testing the video files in the game and carefully watching scene changes are the way to be certain.

 

 

[xxOToTOxx]

Hi guys, great project. Charlie Chan is pretty much spot on, I looked at the MKV and there are frames that need to be dropped from the beginning from the LDF capture, by my math 157 seems correct.

 

Then you just need to encode the m2v to MPEG2 as described in the GitHub pages here: https://github.com/DirtBagXon/hypseus_singe_data?tab=readme-ov-file#conversion

 

The existing .ogg is probably fine as you are just changing the video. Then just drop the m2v in place of the existing m2v file and start up Hypseus let it re-index.

 

I'll be interested in how this all looks 👍

 

 

[ducon2016]

56 minutes ago, xxOToTOxx said:

Hi guys, great project. Charlie Chan is pretty much spot on, I looked at the MKV and there are frames that need to be dropped from the beginning from the LDF capture, by my math 157 seems correct.

 

Then you just need to encode the m2v to MPEG2 as described in the GitHub pages here: https://github.com/DirtBagXon/hypseus_singe_data?tab=readme-ov-file#conversion

 

The existing .ogg is probably fine as you are just changing the video. Then just drop the m2v in place of the existing m2v file and start up Hypseus let it re-index.

 

I'll be interested in how this all looks 👍

Does anyone know what hypseus is getting from mp2 that it cannot get from mp4?

[xxOToTOxx]

1 hour ago, ducon2016 said:

Does anyone know what hypseus is getting from mp2 that it cannot get from mp4?

 

Yes 😆

 

I've written a detailed explanation of this in several places, including Discord. So I not gonna repeat it here (again).

[ducon2016]

8 hours ago, xxOToTOxx said:

 

Yes 😆

 

I've written a detailed explanation of this in several places, including Discord. So I not gonna repeat it here (again).

 

Lol link please. I searched and could not find a real technical explanation. Every frame could be indexed precisely in mp4. So if we can read the mp2 indexing it can be converted to a frame number that can be then used as an index in the mp4. There is pre-processing work of course, but I would love to read what is the technical reason to see if I can be part of solving it 😄

 

[The Thinker]

On 10/12/2025 at 1:48 AM, ducon2016 said:

 

Ouch YUV4MPEG2 was used. Was C444 used or not? If not it needs to be redone.

 

OK will check it out. I want to run state of the art super resolution and frame interpolation on it. Some people tried but there were issues with the other versions that were posted here with some weird cases breaking. I want a 100% reliable version.

 

I captured the Dragon’s Lair Limited Edition (2002) LaserDisc directly to an uncompressed YUV4MPEG2 (.y4m) master using a high-quality capture chain.
The raw file header reads:

YUV4MPEG2 W760 H488 F30000:1001 It A352:413 C444p16 XCOLORRANGE=LIMITED

This means the footage was recorded at 760×488 resolution — the full NTSC active image, preserving the entire analog scan area beyond the standard 720×480.
The frame rate is 29.97 fps (30000/1001), interlaced top-field-first, exactly matching the LaserDisc’s NTSC cadence.

The most important detail is the C444p16 format, which indicates 4:4:4 chroma sampling at 16-bit precision per channel.
This means every pixel contains full color information with no chroma subsampling (unlike 4:2:2 or 4:2:0 formats).
While the original LaserDisc signal itself is effectively limited to about 4:2:2 color bandwidth, capturing in 4:4:4 ensures that no additional loss or rounding occurred during digitization.

I also used broadcast-safe limited range (16–235), consistent with NTSC standards, and retained the interlaced field structure, which keeps motion and transitions authentic to the original arcade timing.

This approach provides an archival-grade digital master, ideal for future restoration or color correction.
From this .y4m file, I later created lossless FFV1 and visually lossless MPEG-2 versions for compatibility and distribution.

 

Modifié lundi à 11:42 par The Thinker

[ducon2016]

8 hours ago, The Thinker said:

I captured the Dragon’s Lair Limited Edition (2002) LaserDisc directly to an uncompressed YUV4MPEG2 (.y4m) master using a high-quality capture chain.
The raw file header reads:

YUV4MPEG2 W760 H488 F30000:1001 It A352:413 C444p16 XCOLORRANGE=LIMITED

This means the footage was recorded at 760×488 resolution — the full NTSC active image, preserving the entire analog scan area beyond the standard 720×480.
The frame rate is 29.97 fps (30000/1001), interlaced top-field-first, exactly matching the LaserDisc’s NTSC cadence.

The most important detail is the C444p16 format, which indicates 4:4:4 chroma sampling at 16-bit precision per channel.
This means every pixel contains full color information with no chroma subsampling (unlike 4:2:2 or 4:2:0 formats).
While the original LaserDisc signal itself is effectively limited to about 4:2:2 color bandwidth, capturing in 4:4:4 ensures that no additional loss or rounding occurred during digitization.

I also used broadcast-safe limited range (16–235), consistent with NTSC standards, and retained the interlaced field structure, which keeps motion and transitions authentic to the original arcade timing.

This approach provides an archival-grade digital master, ideal for future restoration or color correction.
From this .y4m file, I later created lossless FFV1 and visually lossless MPEG-2 versions for compatibility and distribution.

 

Beautiful. Thanks for the confirmation, and great work. It will be a great starting source for all work until we can stack the LDs. One thing I noticed is some frames are doubled or trippled which can help reduce some of the noise and hopefully artefacts. I am still scouting the web for a good VHS deghosting or deringing first.

[TheThinker]

Can someone kindly cut the first 157 frames? Thank you. 

[ducon2016]

23 hours ago, TheThinker said:

Can someone kindly cut the first 157 frames? Thank you. 

 

Don't worry about it this is not a big problem. The VHS part needs a lot of work though, still experimenting. Did not find a satisfying solution yet that does not destroy the small details.