Star Wars (1977) original Blu ray. Crappier than ever.

No, we're talking 1976-1977 Kodak 5247 film grain, which was just ****. And don't overlook the fact that the first four films (Episodes 4-5-6 and 1) were all shot using anamorphic lenses, which further degrades the MTF and apparent focus. And don't assume that grain reduction is as simple as you imagine: the guys handling the grain at Lowry applied different levels of grain reduction to whites, mids, and blacks differently, it was a non-realtime process, and they also added back an even amount of "reasonable" grain to the entire show. They also had the ability to apply NR to one part of a scene and not the other, or make it intense at the top and nearly off at the bottom (among many other combinations). Do a Google search on the Lowry Process® and see for yourself. It was very, very complicated and required good judgement, lots of time, and a ton of computing power. It wasn't just one "crank it up to 11" switch. Even now, with software like Resolve and Baselight, there's about 9 different ways to achieve grain reduction, even applying it only to a single color channel or to only the darkest or lightest parts of the shot. And it can be changed scene-by-scene, shot-by-shot, even frame-by-frame if necessary.

I wouldn't say that a film scanner "amplifies" grain per se. If anything, contrast amplifies grain, and that's just the nature of film imaging. I can refer you to many technical papers that reinforce what I'm telling you, but I got work to do. Go through the appropriate papers published by SMPTE and AMIA, and they'll verify all of this. What I will say is: scanning motion picture film negative and doing it well is very hard and not for the faint-of-heart.

Many, many film scanners work at a high resolution and then downsample on the fly, which has been a standard thing for 20 years that I know of.

 

On amazon Uk people have been leaving reviews of the new 4k blu ray versions before they have even been released. Its the usual paranoid drivel about 4/5/6 being the wrong versions etc.
This bizarre group of troublemakers will find fault with any version that is released.
I blame amazon for allowing reviews of products before they are released.
BTW the best video versions of 4/5/6 were on betamax. The vhs versions were shocking by comparison. thats only to be expected. It was VHS.
I had a Sony Beta hifi and the star wars films came out with superb digital audio. For the time PQ was stunning. But that was beta. a great format and way ahead of vhs. If only sony hadnt have been so arrogant about badging etc.
The grundig/phillips v2000 system was even better than beta with its double sided tapes. I wonder if Star wars was out on that format?

 

I always thought 5247 was a fine-grain stock... watching movies it was shot on, it never looked particularly grainy to me. How do you think it compared to, say, 5254 which I know was a very popular and well-regarded stock among DP's at the time?

What is MTF...?

Spherical lenses for the win!!!

 

It’s “modulator transfer function”, a supposed way of objectively measuring sharpness.

It was. But when you know everything and everyone, it’s ____ when you need it to be for the purposes of your argument.

 

These grain particles are still smaller than what a scanner can resolve.

I'm not saying it's simple, but it seems it can be done very well. I was impressed how good the DNR on 4K77 was, especially considering it was not done professionally. It's much better than the auto grain reduction from 10 years ago I am familiar with. Of course, that reduction always had only 1 frame to work with and could not average out over several frames.

Yes, sharpening (which is enhancement of microcontrast) does this, but grain aliasing introduced by scanners has been a known problem for decades. It was discussed a lot in the photographer world in the huge "film vs. digital" debate 10-15 years ago.
The conclusion then was basically that comparing digital imaging to scanned film is easily tipping the scale in favor of the digital sensor because it's quite hard to properly scan film, and scanning at the same resolution as the source's is not good enough. The grainy structure does not fit well into a fixed pixel matrix, not mentioning all the problems with color.

This is from 2000, but still valid and with good examples:
Grain aliasing

The aliased grain looks just like what I see in those 4Kxx versions without DNR.

The best option for stills is to use a drum scanner like the Heidelberg Tango, with a resolution about double of what we have discussed here so far. Results look very similar to optical enlargement. Requires wet mounting of single frames or short strips of film, making it impractical for movies. Many scanning services charge more than $200 per frame.
Even such a scanner misses faint details hinted at by a few specks of grain here and there. These are only visible with optical enlargement.

This might not matter with all the movement going on inside a cine camera when it comes to detail, but grain aliasing might still be an issue. However, using NR to bring it down again is much more practical than the huge amount of work required for a microscopic film scan. And that was all I'm saying, "no DNR" as demanded by some is not purist in this case, because it would not be there on this level when using a photochemical workflow.

Even scanning stills is not easy if you want to pixel peep.

@The Hermit
MTF is similar to the frequency response graphs for audio equipment. Finer and finer level of detail is recorded with less and less contrast, like audio equipment reproducing frequencies at the end of the spectrum at lower volume than the ones in the center of the range.
This is a nice (again older) page about it, and it has a schematic example of what is going on in the imaging chain (lens - film - scanner) right at the top: Understanding resolution and MTF

 

Both very grainy in modern terms, and in terms for digital mastering (which I do every day of the week). The grain went down massively in the early 1990s with the introduction of Kodak's T-grain Vision stock. My favorite film stock today is 5219, which is so clean, it looks like digital with a tiny bit of grain in it... but with all the color rendition of film.

MTF = Modulation Transfer Function.
What is a lens MTF chart and how do I read it | NIKKOR lens technology from Nikon

Engineers generally talk about resolution in terms of how many line pairs per millimeter it can resolve. My point is that the film stock, the camera lenses, how the film was exposed, and even the lenses in the optical path for the film scan will affect MTF and limit just how far it can go. Contrast has a lot to do with it as well: a scene with lots of fog & smoke in it won't look as sharp as a completely clean, bright shot with great black detail that's well exposed. Apparent resolution is also affected by how steady the film is, so if the film is bouncing around microscopically, that will reduce how sharp it looks in motion.

 

Everything you say has no bearing on how motion picture scanning is done every day in the real world. Read the materials available at the Arri and Filmlight websites and see for yourself. Again, SMPTE and AMIA have lots of technical papers on this stuff, and much of it has been known for 20 years or more.

Arri Archive Solutions

FilmLight | Products | Northlight

Society of Motion Picture & Television Engineers

The Association of Moving Image Archivists

Here's three good books on the subject:

Color & Mastering for Digital Cinema by Glenn Kennel
https://www.amazon.com/Mastering-Digital-Cinema-Industry-Handbook/dp/0240808746

Digital Cinematography: Fundamentals, Techniques, & Workflows by David Stump
https://www.amazon.com/Digital-Cine...855497&sr=1-1&keywords=digital+cinematography

Color Reproduction in Electronic Imaging Systems by Michael Tooms
https://www.amazon.com/Colour-Repro...=sr_1_1?s=books&ie=UTF8&qid=1541130852&sr=1-1

Do a few weeks of reading and get up to speed, and you'll begin to understand the enormity of this subject. I've been doing this every day for 20 years (actually 40 if you include standard-def), and I learn new things every day.

 

Do not speak until you get to his level! Of course, since he learns new things every day, you'll never be at his level. So just don't speak.

 

There was a Betamax Hi Fi format with digital audio? As far as I know Betamax Hi Fi was FM audio just like VHS Hi Fi.

 

So what was the need for grain management by Lowry? And why is their still grain on this film and other films on this stock?

They should be grainless because "These grain particles are still smaller than what a scanner can resolve."

 

In around two weeks, April 17, Rollerball is being released on UHD BD with HDR. Rollerball is one of more grainy movies I've ever seen, I wonder if there will be a noticiable improvement in sharpness compared to the two BD releases I own (Arrow Video and Twilight Time) and if HDR won't make grain ever more visible ala Close Encounters Of The Third Kind.

 

God, there's bucket-loads of grain on all the Star Wars films shot on film. One issue with higher-res is, the more sharpness, the more you can see the grain. At lower res, the grain tends to disappear.

In fairness to Kodak, an issue is that if we make an adjustment to contrast (stretching out the signal) in final color, this also tends to bring out the grain even more. So a small degree of grain reduction is almost always necessary. There's a way to do it carefully so that there's no artifacts and you're not softening the picture.

 

There was a Betamax format that recorded only digital audio, the PCM-F1 system, but with no picture. You could theoretically do Beta Hi-Fi at the same time, but I don't know of anybody who tried it.

Yeah, the grain in the 4K HDR Close Encounters was terrible. I haven't yet seen the 4K Star Wars that Disney+ has, but I'm guessing they fixed this kind of thing. A lot of 1970s and 1980s films are grainy as hell, some horribly so. I seem to recall the first Predator was really bad in terms of grain.

 

Rollerball on the two BD releases I have is even grainier than Close Encounters Of The Third Kind, so I don't know what HDR pass can do to Rollerball without turning it into a Paella. I hope they used it very conservatively and the main benefit of using it is Wide Color Gamut.

 

Amazon has a special "Buy 2 get 3rd" promotion that works on the 4K discs, but Empire 4K is OOS.

It means any of the three trilogies would cost $60 total - or get them all for $180 plus tax.

Get 3 for the price of 2.

 

Maybe that's what this reviewer is describing when he says there are many moments of "frozen grain." Yikes.

Star Wars: Episode V - The Empire Strikes Back 4K Blu-ray Release Date March 31, 2020

 

About how big would your 4k TV have to be before you start seeing "edges" (I am probably using the wrong word) of 1080p content? Is this even an issue? Can 1080p content still look amazing on a 4k TV? My coworker says you can absolutely tell, and that 4k is almost a night/day difference. He was raving about Star Wars 4k streaming on Disney+. He has yet to view a proper 4k UHD disc.

 

You should have a look at these videos by Cinematographer Steve Yedlin Strictly speaking I think even at IMAX sizes you won't notice a huge difference between good 1080p and 4k though you will probably notice it in stuff like the text in opening titles. I remember watching Superman Returns on IMAX back when that was in cinemas and I didn't notice anything wrong with the image. I wonder if what your coworker is seeing isn't the difference in resolution but rather the difference in color grading and mastering.

 

I don't see "edges" at all. You mean pixels? Can't see 'em at 6 feet for sure. I do see HD uprezzed to UHD at work, and it's fine.

I'd like to know how the reviewer's display is setup first. There could be compression issues causing frozen grain, because the DVD/Blu-ray MPEG algorithms determine the film images are moving but the grain is essentially stationary and random. The Lowry Process did not involve frozen grain -- the grain was always moving.

 

I watched the first video, which was interesting, but the 2nd one is too long for me. Thank you for the info, nonetheless.

It's possible he's talking about color grading and mastering, but he has not done a direct comparison, to my knowledge. He said he noted some kind of deficiency in one of the blurays he put on.

I wouldn't have expected Lowry to do a solid grain field, but figured you would know better than most. Thanks for your reply.

 

Target also.

 

Sorry for the late reply, was away for a while ..

No. You are mostly seeing lumps of grain in the scan, not single particles. There is also interference between the scanner matrix and the particle patterns, therefore a lower OR higher resolution scan can improve results. Both cases are possible, though a "microscopic" scan at 10K+ would be best for most types of film.
What is done in the film industry seems to be a "good enough" type scan, clearly there is no attempt to extract everything the film could have to offer. Thus, such discussions as here are mostly academic.

There is always more resolution in the source than matrix sampling will bring out, due to constant misalignment of details with the matrix.
Real world example: Test at opticallimits.com I recently read, Canon 11-24 f/4 delivers >3700 lines per image height at MTF50 on an 21 MP sensor (3744 lines), dropping to 3000 at f/4. Looks like a high res sensor is lost on that lens at small apertures? After all, it can resolve more than 3700 lines and the second result is only 3000.
Wrong, same lens on a 50 MP sensor (8712 x 5808): ca. 5600 at f/4, 3800 at f/11. There is no hard limit, a higher res sensor still shows improvements. That is all I was saying. The same applies to scans. These improvements gradually get smaller as the scan resolution increases or the source quality decreases, e.g. you would see very little improvement when feeding these two sensors images from Super8 film at over 1:1 reproduction size.

If you are seeing a scanned result near your sensor resolution, it indicates the source has much finer detail than the scanner (2x and more). The lens above at f/4 probably delivers detail at the aperture's diffraction limit which is about double the linear resolution of the 50 MP sensor.
Similarly, if you scan something at 2K and it is does reach almost 2K resolution in the digital file at 50% contrast response, the source was more like 4K.

 

This just makes me sad.

I just watched the Mos Eisely sequence from the new 4K copy of Star Wars. and special edition changes and some DNR withstanding. It looks amazing. And the shot that has made a staggering jump in quality is the panning wide shot of the land-speeder pulling up to the Cantina.

In the film prints that scene is always a blurry, super grainy mess. I don’t know what the team who restored it did but it is just night and day the image quality difference.

And just when you think that this is the ultimate version of this scene. The special edition Dino pops into shot and ruins everything.

I pray one day we will get the OOT in this quality, because the definition revealed on the new 4K is just amazing and it’s let down by the SE changes.

 

I'm just guessing here, but even JJ can't get it released. I suspect part of the Disney sale involves a clause preventing the original 1977 version ever being re-released. https://mashable.com/article/jj-abrams-star-wars-theatrical-cut/

 

Does JJ really have enough clout? He has been a part of some big movies, but it isn't like he has multiple Oscars.