Sign In

Williarob

User Group
Trusted Members
Join date
9-Apr-2007
Last activity
16-Jun-2019
Posts
772
Web Site
http://www.thestarwarstrilogy.com

Post History

Post
#1285363
Topic
RAW DV VHS transfers of ISD Executor and maybe more.
Time

SF01 said:

Williarob said:

The biggest problem I found with DV capture is that is uses 4:1:1 chroma subsampling which really reduces the picture quality.

In 4:1:1 chroma subsampling, the horizontal color resolution is quartered to reduce the bandwidth by half. In 4:2:2 chroma subsampling the two chroma components are sampled at half the sample rate of luma: the horizontal chroma resolution is halved. This reduces the bandwidth of an uncompressed video signal by one-third with little to no visual difference.

https://en.wikipedia.org/wiki/Chroma_subsampling

You are completely right about 4:1:1, but you might have missed one important detail. The ISD is PAL and I’m capturing in PAL and in PAL the chroma subsampling is 4:2:0, not 4:1:1, which looks light years better.

Oh, I didn’t know that it was 4:2:0 in PAL. That should look much better 😃 Carry on!

Post
#1285347
Topic
RAW DV VHS transfers of ISD Executor and maybe more.
Time

The biggest problem I found with DV capture is that is uses 4:1:1 chroma subsampling which really reduces the picture quality.

I tested it for laserdisc capture, and found it wanting:

Airplane LD captured as DV (4:1:1):

DV

Same disc recaptured as Lagarith 4:2:2:

AVI

In 4:1:1 chroma subsampling, the horizontal color resolution is quartered to reduce the bandwidth by half. In 4:2:2 chroma subsampling the two chroma components are sampled at half the sample rate of luma: the horizontal chroma resolution is halved. This reduces the bandwidth of an uncompressed video signal by one-third with little to no visual difference.

https://en.wikipedia.org/wiki/Chroma_subsampling

Post
#1284941
Topic
Dr. No Rare Uncut Version 1st Beta/VHS release 1982
Time

JayArgonaut said:

captainsolo said:

Like all the non-scope Bonds for the SE DVD, GF was cropped to 1.78:1. While this is still around the safe limits for framing flat matted widescreen the SE DVDs did crop a bit too much on all five done this way and the 1.66 is the best ratio for the initial three in terms of composition.

Thankfully the 1.66:1 OAR was restored for the GF UE.

Here’s a fun fact: 1.66:1 is not the original theatrical aspect ratio for the first three Bond films, the correct AR is 1.85:1. Since the mid-fifties, the most common AR in the UK was 1.75, but Kinematograph Weekly specifically lists 1.85 as the intended ratio for the first three Bond films. This documentation is from a British trade journal at the time of principal photography so it’s as good as gospel:

Imgur

A big thank you to Bob Furmanek of the 3D film Archive (http://www.3dfilmarchive.com/) for pointing this out to me this weekend.

Now, like most of you, I like minimal cropping, so I’m glad there are 1.66:1 versions available, but it’s also nice to know how people saw it in theaters in 1962. Looking at an original IB Tech film frame from a 1962 print, we can see that there is slightly more picture information available, my math puts it at around 1.58:1 if you include all of it. I believe it was hard matted in camera to 1.66:1 so there is probably a little more picture information on the sides on the original negatives than made it onto this print.

Imgur

Post
#1282696
Topic
Project 4K80
Time

Servanov said:

NeverarGreat said:

I feel like there has to be an easier way to clean a static starfield.

Since the automated dust cleanup presumably erases stars, have you considered stabilizing the shot, running the cleanup, and layering it over the original on ‘darken’? Then if you had just a single frame manually cleaned, it could be overlaid on top of that with the blending on ‘lighten’, which would return the accidentally erased stars.

True. This is actually a trick we’ve been taught at my film restoration business. We make two copies of the frame, and on one we apply an automatic dust and scratch removal, which removes some detail, but then put the dirty layer on darken and the quality is kept while only removing the dust.

So, limited success on this one… The main problem is that both the dust and the stars are “light”, so using the automatic cleanup and blending it with the original won’t work here. Most of the dark dirt is lost in the blackness of space.

To illustrate the problem, here is a raw frame (this is from the pandown, to make it more obvious which stars are missing, but the same is true during the crawl where perhaps only a single star is accidentally erased):

Imgur

And here is the same frame after automated cleanup:

Imgur

Now If I put the original frame on top of the cleaned frame and set the blend mode to Lighten, all of the stars return, but so does the red line - in fact it looks identical to the original uncleaned frame. If I set it to Darken, it looks identical to the cleaned frame - the line is gone but so are all those stars. (This is a technique I’ve used before, and it works very well on a lot of shots, but unfortunately not this one.)

However, Using a static stars frame (one taken from right before the pan down), and placing it above the auto cleaned layer with the blend mode set to “Lighten” does work pretty well for most of the shot - though of course it can’t be used until the Star Wars logo has gone, or it adds stars to the black area inside the words:

Imgur

And it’s only good until the pan down, but that’s still a lot of frames.

Every frame will still need to be checked, but this should save me some time painting out the dirt and dust, so thank you for your suggestions.

Post
#1282627
Topic
Project 4K80
Time

The $325 has been raised, the 16mm print has been scanned, and is ready for use as a color reference and for repairing/replacing damaged frames. The Fuji print has also been scanned, so now it is just a matter of time, rather than money - unless somebody has 6 figures they can throw our way, so that we can quit our day jobs and work on this full time.

Cleanup is very time consuming, as this real time video demonstrates:

https://vimeo.com/338040925/ca9015d4f8

In 13 minutes, about 30 frames (or just over 1 second of film) is cleaned. The frame is advanced, forward and back repeatedly, because it’s very difficult to see which white specks are dust and which are stars without checking to see which ones persist across multiple frames and which are only there for a single frame.

It’s quite boring to watch and it’s quite boring to do, but it’s also why there are so few updates about the project. This opening crawl sequence is 2000 frames and will likely take at least two weeks to clean. We all have full time jobs and family responsibilities, so an hour or two per day is the most we can usually devote to projects like this one. Two films in, this is all routine to TN1 at this point, so unless a new technique or new software is being tried, or something weird and unusual is spotted in the footage, time isn’t usually wasted creating videos or posts to update people about what’s happening. If this post takes 5 minutes to compose, that’s 10-15 frames that didn’t get cleaned.

Progress is slow, painfully slow, but it is happening. Every day more frames are being cleaned, color corrected, stabilized, deflickered, and repaired or replaced. (Though not necessarily in that order!)

Different reels are being worked on simultaneously by different people around the world, which makes tracking progress difficult. But don’t worry, when we have some news, we’ll let you know.

Thank you to everyone who has donated, and to those who are still willing to donate. At this point, if you’d like to make a donation to help projects like this one, I would recommend you donate to one of Poita’s projects - last I heard he needed the money for Hard Drives, without which work cannot continue, so donating to him will actually help get some of his projects get off the ground, e.g.:

https://originaltrilogy.com/topic/The-Original-Trilogy-restored-from-35mm-prints/id/49756

paypal.me/poit

Though of course donations to us are also always welcome 😃 We always need more software, Hard Drives, and other computer hardware.

Post
#1281544
Topic
Star Wars Tosche station Restored
Time

Sky_ said:

Omni said:

What other version would they have? I’m pretty sure the deleted scenes as found on the Bluray are the best quality we’re going to get for a while.

Williarob said:

TN1 does not have a better version of the scenes. Surprisingly, the deleted scenes were not on any of the theatrical prints that were scanned.

I saw The Team Negative 1 of the deleted scenes, even though they was damaged like the ones on Bluray, they were much more Sharper a lot more detail, it was somewhere on these forums, cant remember the link, so I wrote to them and not had a Response, most likely they never saw my messages, that why i need help in getting contact with them… 😦

These ones?

https://originaltrilogy.com/topic/Star-Wars-Deleted-Scenes-Reincorporated/id/13512/page/9

That’s from the bluray. Unless you have access to the Lucasfilm archives, there’s currently no better source for these deleted scenes than the official bluray. If they look cleaner and sharper, it’s because we cleaned and sharpened them, but as far as I know, we never completed the project - as you know it’s a ton of work.

Post
#1275128
Topic
Star Wars 4K77 - Regraded - No DNR
Time

TK-423 said:

Total noob question here. Is a regrade essentially just a frame by frame recoloring? So this is essentially a version of 4k77 that an editor has gone through and recolored frame by frame for consistency or whatever?

So the original 4k77 release was “as is” whereas you could view this a version of 4k77 that has had specifically the color “remastered”?

Thanks!

Yes, I think that sums it up quite nicely. This has indeed been remastered, because Sanjuro was kind enough to send me his project files, and I could therefore apply his new grade to the lossless master and generate a new version - which is better than what usually happens, where somebody takes the h.264/5 version and regrades that, resulting in a slight generational loss.

Post
#1273261
Topic
Upcoming Skyfall airing on Syfy, possibly open matte?
Time

Skyfall does appear to have been broadcast open matte:

Bluray:
Imgur

SyFy:
Imgur

Certainly a little more image top and bottom.

Not sure how useful it is though, obviously it runs 180 minutes with commercials, it has a logo in the corner, and may have been edited for television.

Would it be better to try and get it from Epix or Amazon Prime?

Post
#1271340
Topic
Goldfinger Theatrical UK version
Time

kchrules said:

… this 16mm clip on eBay which I can’t afford: https://www.ebay.com/itm/16mm-IB-TECH-Film-CLIP-GOLDFINGER-1964-James-Bond-003-Now-with-DIAMONDS/113659583367

Yeah that’s been on the bay for a while now - $200+ for 33 seconds of footage! I keep waiting for the price to drop, but he seems to think that’s what it’s worth, and maybe it is, but I can’t afford it either.

Post
#1269027
Topic
Star Wars 4K77 - Regraded - No DNR
Time

DrDre said:

This is some really nice work, and in record time! Congrats! I think your shot by shot can also be used as a great basis for further tuning to cater to each and everyone’s personal preferences. Here’s my little effort, starting from the last shot of the film. I personally feel the color grading could use a little warmth to bring it more in line with the technicolor references.

sanjuro_61:

DrDre:

I applied this adjustment to the two sets of sample frames I found:

It might need a little tuning here and there, to get to a final version (I see a few frames with some excess magenta), if there’s an interest, but you get the general idea.

Here’s the LUT:

https://drive.google.com/file/d/1hDo_XIiozzDanx5B1WAn-GuG31UwHlYW/view?usp=sharing

Cool - thanks Dre!