Here is a side by side of the before/after on one scene:
After finishing up, I wrote up my procedure. Skip down about 75% through the article to the section "Restoration Details" to read what the steps are. It is amazing that by correlating the information across multiple frames, the resulting image is significantly sharper than any one frame. I mean, it makes sense in the abstract, but it is neat to see it actually work.
My avisynth script was 95% based on the work of "Video Fred":
From there you can find his youtube channel and see more restoration efforts:
Given a budget and a team of talented and motivated people, I'm sure Jackson was able to wring a lot more juice out of his old films.
> Here is a side by side of the before/after on one scene:
That is remarkable! Now I wish I had some old footage so that I could restore it :P
Your brain does this all the time - the resolution of your eyes is actually quite poor compared to what you perceive (as much as a 25x difference). It's quite easy to demonstrate:
> For the most impressive results -- I guarantee you will be amazed -- use a dim room with no light apart from the computer screen; a pretty strong effect will still be seen even if the room has daylight coming into it, as long as it is not bright sunshine. Cut a slit about 1.5mm wide in the card. On the screen, display http://www.inference.org.uk/mackay/itila/Files.html
> Stand or sit sufficiently far away that you can only just read the text -- perhaps a distance of four metres or so, if you have normal vision. Now, hold the slit vertically in front of one of your eyes, and close the other eye. Hold the slit near to your eye -- brushing your eyelashes -- and look through it. Waggle the slit slowly to the left and to the right, so that the slit is alternately in front of the left and right sides of your pupil. What do you see? I see the red objects waggling to and fro, and the blue objects waggling to and fro, through huge distances and in opposite directions, while white objects appear to stay still and are negligibly distorted. Thin magenta objects can be seen splitting into their constituent red and blue parts. Measure how large the motion of the red and blue objects is -- it's more than 5 minutes of arc for me, in a dim room. Then check how sharply you can see under these conditions -- look at the text on the screen, for example: is it not the case that you can see (through your whole pupil) features far smaller than the distance through which the red and blue components were waggling? Yet when you are using the whole pupil, what is falling on your retina must be an image blurred with a blurring diameter equal to the waggling amplitude.
-- p553 of http://www.inference.org.uk/mackay/itila/book.html
Despite the absolutely spectacular footage. I thought the documentary itself was OK. It uses audio from people in British WWI regiments to tell the story—all from BBC and Imperial War Museum recordings. It was a good execution given the constrains, but there's no real angle or conclusions to draw from the documentary.
The real gem was the interview with Peter Jackson after the movie. It covered the restoring process—much of which is covered in the article. They basically used computer software to move the frame-rates from 12 to 16 fps up to 24 fps. Then tweaked the highlights and shadows. There's a ton of work that went into color and sound as well.
I hope these techniques can be used to restore other old footage.
Paul McCartney was so impressed by Peter Jackson's work that he approached him to restore 55 hours of old Beatles footage. Rumor has it that the film will come out in 2020.
I liked how it wasn’t pushing a narrative, because you obviously come to the conclusion that war is hell.
Interesting because the thing that kills most documentaries for me is when they have a clear angle or are driving for a conclusion.
I find it odd that this article/interview is about something very visual and has no images in it so I wanted to see what it looks like.
Not exactly. They filmed using the 3 strip Technicolor process, meaning 3 strips of b+w film, each for a different color. This was also used in the 1925 Ben Hur, though they used only 2 strips.
Annoyingly, when that was restored, it was on line as a free film. Now it's on Amazon, etc. as a product.
An obvious difference is intent. The intent of They Shall Not Grow Old is (I assume) to memorialize historic events and enable modern audiences to understand them more viscerally. The intent of (most) deepfakes is deception.
Still, what should we do when the same technology that enables deepfakes also enables better They Shall Not Grow Old's, perhaps concerning subjects more poorly-documented than World War I? It's acceptable to make assumptions about the colors and sounds of a moment from 100 years ago and represent them indistinguishably from truths, but how about assumptions as to who was present and what actions they took and words they spoke?
All of which reminds me of this: https://twitter.com/rivatez/status/1077035912494759936
It's like, say, a movie or novel based on a historical event we don't know much about - it may aid understanding of the event, but the details are made up. It's fictionalized, "based on a true story", and to present it as fact or as history would be misleading.
Synthesizing texture, adding faces, meh. It's all an array of pixels to aid hallucination. The sooner we break out of video = impartiality, video = truth, the better off society will be.
Choosing what to record, how to crop it, what to show destroys videos supposed fundamental impartiality.
But seriously, would you be the first to stand up, when the last guy that did it was mowed down?
> was a Japanese holdout who did not surrender in August 1945. After the war ended Onoda spent 29 years holding out in the Philippines
(Also: It's amazing to me that they could convert hundred-year-old film to 3D.)
Edit: To be clear, I'm saying that since most 3D films aren't shot in 3D in the first place, this film probably won't be substantially different from other films, in 3D. It seems some people thought I was confused about what the movie was.
I "liked" this movie, as much as anyone can like seeing dead bodies. It really makes you realize how much most war movies, fiction or not, leave out. I wonder to what extent the lack of the true horrors results in a pro-war propaganda film, regardless of the intent of the director. Kubrick's Full Metal Jacket tries to be realistic in a sense, but still glamorizes war. It takes a documentary like "They Shall Not Grow Old" to give the full picture.
"The 3D in this WWI footage was faked in post."
Which is obviously true. I instead read it as:
"The reason the 3D looks about as good as you're used to from other films, is that most of those modern films _also_ faked their 3D in post."
When the camera pans, it gives the effect of looking through a window.
I look back at the experience now feeling like I time travelled.
I understand that Jackson isn't much of a techie but it would have been nice if they brought someone that worked on the restoration from the tech side along to explain their process.
I'll have to check that out.