logo Sign In

Laserman

User Group
Members
Join date
11-May-2004
Last activity
6-Sep-2007
Posts
903

Post History

Post
#241602
Topic
Info Wanted: Anyone Planning on making Anamorphic versions using 2006 OUT DVDs?
Time
ESHBG - It depends what you mean by does it make the picture clearer.
If you have a widescreen TV then the picture will almost definately look better if you create an anamorphic version 'offline' and play that back on your widescreen TV.

The reason is that you can scale a picture well or poorly, or anywhere in between. A widescreen set has to scale a letterbox image to display it properly. This means it has to do it in realtime so it uses a (usually cheap) chipset to do that. The result is usually not that great.
If you use software to make an anamorphic version first, then you can use really good scaling algorithms that would be too complex to do in real time. This means that instead of the image being poorly scaled by the TV, you send it the already (properly) scaled image and so get a better picture.

This is why a lot of people with high end TVs, projectors etc. buy a lumagen, terranex or other high quality 'outboard scaler' because it does a much better job of scaling the input to the display's native resolution and Aspect Ratio than the TV itself is capable of.

If however you were watching in letterbox mode on a standard TV, then an anamorphic version created from the letterbox version may look about the same or actually worse - unless the original letterbox version was overly soft and gained some improvement from any other video processing you did to it during the conversion to anamorphic.

But in absolute terms straight scaling an image from letterbox to anamorphic doesn't automagically add in any lost detail, so doesn't make the picture clearer - but if you need to watch it on a widescreen TV then the image will almost certainly be clearer than watching the letterbox original, unless you use an equally as awful scaling algorithm as the one in your TV.
Post
#241583
Topic
In defence of the 2004 DVDs
Time
*THUMP*

Hmm, funny how the training mode was never mentioned in any of the books/novels/games before...

Crushed whites and blacks, and oversaturated colour have unfortunately become the current trend for Blockbusters. The general public likes the impression of extra 'punch' or 3D look that is gives the image. It also makes plasmas and LCDs that have poor shadow detail and elevated blacks not look so bad in comparison the movie. The overuse of colour to intone mood is also a current trend - it treats the cinema goer like an idiot. Instead of a slightly cold (blue) grade to help convey menace you get a blue tinted scene to force the idea down your throat. Love scenes have insanely red sunsets and colour grades to match.

So it looks like Lucas just wants to make Star Wars fit in with the current 'look'. It weirds me out because you have gritty realist 70s film styling with a current colourful/smooth/slick look washed over the top, it just doesn't sit right.

It is a shame because it is kind of like listening to music with the treble turned right up, for a few minutes you think wow I am hearing more detail, but after a while it grates on the ear and becomes unlistenable. Crushed movies feel that way to me when I watch them, it gives them a cartoony look which is kind of OK if that is part of the feel (like a superhero/comicbook movie), but I think detracts from the cinematic look of a film. 70s movies were perhaps a little too flat, but the realism of that look fit well with science fiction, it gave it a grounding which helped you suspend your disbelief. The gritty texture of the sets in movies like Alien, the text based industrial look to the computer screens and readouts in Star Wars, Alien, 2001 made all the crazy stuff easier to believe. The look of the 70s cinematography also helped in that it had that same 'every day' quality to it (it would be considered almost a 'documentary' look now).

The smooth slick look kind of feels hyper real (or fake) which when combined with a story that is unrealistic/futuristic just makes it harder to believe IMHO.
Post
#241569
Topic
Info Wanted: Anyone Planning on making Anamorphic versions using 2006 OUT DVDs?
Time
Well, the problem is going to be that because it will be on DVD for the US anyway it falls under the draconian copy protection legislation.

For other countries where maing your own backup is legal, then it may be worth it if you have a widescreen TV that doesn't have a good scaler built in.

Then you could do it for free using avisynth/virtualdub. You could just do a resize and perhaps a limitedsharpen
Post
#241565
Topic
.: The XØ Project - Laserdisc on Steroids :. (SEE FIRST POST FOR UPDATES) (* unfinished project *)
Time
It might, the film problems will definately be cleaned up. I expect it to be better on some fronts for sure, I expect Moth3r's will come out better in some aspects than the OUT as well, (and there are likely to be some scenes from PAL transfers that have more detail than both the OUT and the X0 project) but there are likely to be scenes that have more detail on the OUT unless Lucasfilm really screws it up.

But the X0 will be the only version so far that fixes the film artifacts and that alone will make an more enjoyable viewing experience.
Post
#241269
Topic
.: Moth3r's PAL DVD project :.
Time
Yes I mean picking it up before it gets split into Y/C and before it gets digitised in the case on the 925.

The 925 gives some interesting possibilities, it is easy to modify to get an RGB output, and in theory you could pick up the digital signal and buffer it and do a digital capture from its internal frame buffer.
The 2950 could be modified to get the composite feed before it is split also.
Worst case you might have to put a drive transistor in to lift it to 75 ohms, but we are talking about just a few dollars.

Of course it then means that you need a capture device with a better comb filter than the one in the laserdisc player to see any benefit.
Post
#241101
Topic
.: Moth3r's PAL DVD project :.
Time
We used to build rise time accelerators, but a quick patent search shows that faroudja and Crystal Vision have both used a similar process in some of their standalone comb filters.
Anyone with a videophile mate got a Faroudja VS50 or Crystal Vision VPS-1?

They probably turn up on ebay or similar these days as they are not much use in the digital age to most people, and would probably be cheap. Keep yer eyes out for one.
Post
#241083
Topic
.: The XØ Project - Laserdisc on Steroids :. (SEE FIRST POST FOR UPDATES) (* unfinished project *)
Time
Yep, that is why they look better.
Although if you are in a PAL country and your player isn't setup right US NTSC discs can look really really bad as setup/pedastal whatever you want to call it can end up being added twice, resulting in a really washed out picture.

Basically the Japanese discs have an extra 7% or so of range to play with, so more detail in the shadows etc. Also, the Japanese laserdisc owner was considered a more discerning viewer, so extra effort was usually put into Japanese releases to squeeze the best quality out of a transfer.
Post
#241078
Topic
.: Moth3r's PAL DVD project :.
Time
Thanks for the PM Moth3r.
By that I meant the best way to fix the problem is to do it electronically rather than trying to filter it later.
Basically the problem with Laserdisc is that even the best players crap out at around 5 to 5.5MHz. This leaves laserdisc with a relatively slow rise time (the time taken for a signal on the video path to change from one colour to another) and can cause the 'ringing' or halos as well as awful transitions between colours.
To solve the problem somewhat you can map the Chroma signal to the Luma signal. The Luma signal is relatively quick, around 150nanoseconds, whereas the chroma signal lags with a rise time of around 800nanoseconds. By devising a circuit to lock the chroma signal rise time to the Luma's faster rise time the effect can be pretty much eliminated, resulting in a DVD like picture from laserdisc.

If you have a laserdisc with a set of colourbars on it you can see the problem between the Magenta and Green vertical bars quite clearly.
If you were smarter than me you could then work out a way to lower the rise time of the Luma signal with some sort of rise time acceleration and nail the halos forever.

Post
#240863
Topic
Making our own 35mm preservation--my crazy proposal
Time
Some of the super8 prints were done a lot better than the 16mm reductions. Derran and others managed to get Super8 prints mastered from the same source used to strike the theatrical prints, which means although you get the lower resolution of Super8, you get all of the detail not eaten up by the grain - and the full exposure latitude of the film.
If you can find a 16mm print taken from that then it would be truly amazing, a super8 print could make a great DVD transfer, but it wouldn't be worth trying a HD transfer, the grain eats up the LOD.
Post
#240552
Topic
Making our own 35mm preservation--my crazy proposal
Time
Damn, just typed a massive reply and it disappeared!?

OK short version.

Colour on film VS video.
HD, DVD and DV all compress the hell out of colour. In the PAL world they are 4:2:0 instead of 4:4:4 (in NTSC they are the same except DV which is 4:1:1, so if you take NTSC DV and make a DVD from it you end up with 4:1:0!!), *and* are only effectively a little less than 8bit colour (256 shades of Red Green and Blue).
To wrap one's head around this, do the following. Create a 720x480 image in photoshop. Let's say it is a pattern of 720 x 480 individually coloured pixels, with the first pixel being red, the next green, then next blue and then repeated.). If you looked at a black and white version of the image you created, you could see each individual pixel. If you looked at just the colour information you would also see each discrete pixel clearly.

Now convert that image to DV or DVD. In B&W you can still see each pixel clearly (mpeg2 compression notwithstanding) but suddenly there is not a pixel for each piece of colour information - there might be as little as one piece of colour information for every *four* pixels of black and shite information. If you looked at just the colour information it would look really blocky and low rez. (Studio cameras may be 4:4:4 or 4:2:2, but nothing in the consumer realm is.)

Now even if DVD etc. were 4:4:4, (i.e. keeping the colour and luminance informatio at the same resolution) film hold a lot more than 8bit colour can. To capture film properly you need at least 12bit colour, and many prefer to work in 16bit.
This is why CG effects that look great in the cinema often look hokey on DVD - a lot of the detailed colour information that fooled your brain into seeing the CG as real has been thrown away in the transfer to DVD.

As for exposure/latitude/dynamic range, the sensor in your video camera cannot handle the wide range that film can. So you either end up crushing the blacks or blowing out the whites.
This means that in a scene like the trash compactor you may lose a lot of the detail as that scen is dark, or that you end up with lost detail on the white stormtroopers armour. it is especially bad in scenes where you have a mixture of dark and light parts of the one image.

To get around this, you can make multiple captures - in layman's terms one dark capture, one middle of the range capture and one bright capture.
You can then merge them together intelligently into a single HDR file that keeps all of the information from the really dark detail to the fine lighter details.
Thankfully there is an open source way to store this, the openexr format. You can do more reading here. http://www.openexr.com/samples.html


Take a look at the stained glass image on that page , if doing a single pass to capture that from film, you would have to choose one of those three outcomes, by using HDR you can map it all down to keep the dark and light details, retaining as much information from the film as possible.
Below the left image is Paul Debevec's shot of a Stanford cathedral interior, by doing multiple exposures you can get all of the information and tone map it down to 8bit like the image on the right. It works the same for transferring film to video.

http://www.cybergrain.com/tech/hdr/images1/tone_1.jpg http://www.cybergrain.com/tech/hdr/images1/tone_2.jpg

There are some more examples here http://www.hdrsoft.com/examples.html
and a nice introduction to HDR here.
http://www.cambridgeincolour.com/tutorials/high-dynamic-range.htm
Post
#240494
Topic
Making our own 35mm preservation--my crazy proposal
Time
Then of course you realise that you can't even work out how to drive the damn car, and need to spend a fortune and a lot of time training to be able to even do a lap of the track.

Simple test for anyone wanting to get into doing a transfer from 35mm, or who believes they know someone who would.
Buy one of the 35mm Star Wars or related trailers from ebay, (There are 350 35mm trailers on there right now) they crop up there all the time and don't sell for very much - a Star Trek Insurrection trailer is on there with a buy it now price of $5.
The trailer is guaranteed to be in far better shape than any SW print out there.
Get that transferred to HD and post back here with the results. Preferably pick up an older trailer like the gremlin's one for about $40 which will be closer in condition to a SW print.


If you can't get a 2 minute trailer transferred, then it doesn't bode well for getting 120 minutes done. If you can get it done then it will give you an idea of wether the quality of transfer is going to be worth it or not.

Buying a cheap trailer and trying to get it transferred is an almost no risk way to see if you could get a SW transfer done, and could be a lot of fun.
Post
#239654
Topic
Making our own 35mm preservation--my crazy proposal
Time
Originally posted by: mcfly89
Good ol' fashioned telecine:
Get a Sony Z1 HDV camera ($5,000), a good computer with an HD card, RAID, and a couple terabytes (several thousand $). Set up the camera on a tripod, project the print on a small, clear screen, run component HD out of the camera (uncompressed) into the computer, and then do cleanup in post.

Cheaper version:
If you can't do uncompressed, you can record in HDV mode (25 Mb/s MPEG2 in HD) and it'll only take up less than 30 GB and still look great. This doesn't need to cost a lot--I'm sure we could find someone with a 35mm print and coerce him or her into letting us shoot it.

I must have overlooked something, so feel free to put your two cents in. I don't think we'll get the OOT in HD, and the longer we wait, the longer the prints will deteriorate. I think a solid bootleg with a 3-chip HD camera and some fixes in post would look fantastic. Especially compared to laserdisc.


Couple of problems.

1) Finiding a 35mm projector that you can fire onto a small screen.
2) Your ANSI Contrast Ratio drops through the floor by filming the projected image, so a *lot* of detail is completely lost.
3) The Sony records an interlaced image, the film moves in the gate between fields so every frame will have interlace 'stepping' problems. You will need to re-align all of the fields in post. Not impossible but problematic. You will also need to stabilise the footage unless using a pin registered projector.

4) You could use a Decklink HD card - does the sony actually put out 4:4:4 uncompressed via RGB or component? I really didn't think it did.

5) You will get a turdload of frames with the shutter in shot - you would have to replace or remove the shutter in the projector - you would still get thousands of frames that are 'inbetween' frames.

6) "Do a cleanup in post." Sounds easy when it is put that way, but cleaning up 170,000 frames of HD takes a bit of work. Film is *dirty* and even if professionally cleaned is still *dirty* and pretty much unwatchable on TV, especially in HD. Once again, can be done, but is a hell of a lot of work.

7) Exposure. This is the biggy, there is no way to capture anywhere near the full exposure range in a single pass using a projector and a camera, multiple passes would have to be shot, aligned and merged (similar to the HDR process used by photographers) to get an even remotely acceptable exposure.

8) The projector will also need the gate widened if it isn't already.

You could end up with something usable, it may be better than the OUT release even, but this method would be unlikely to even come close to the official DVD release.

But hey, if you can find someone with the projection equipment and a good print, give it a go with HDV, there is nothing to lose except buckets of time if you already have the camera.


Post
#238561
Topic
.: The XØ Project - Laserdisc on Steroids :. (SEE FIRST POST FOR UPDATES) (* unfinished project *)
Time
Originally posted by: netgurucr
Zion & MeBeJedi, If you had to choose your two favorite softwares for this proyect, which ones would they be and why? Thanks

Slam Dunk 23


Personally I couldn't do without Shake. The nodal nature of the product allows you to layer stuff up in insanse ways, feed various results back into other routines and then recombine with ease. Plus you can dig straight into the scripts.

It makes things simple that would be a bear in other products.

As for these official pictures, I wouldn't take them as an indication of quality. I am pretty sure the official release will have more detail and colour saturation than people are expecting.

As for how it will compare withe the X0 release, who knows, I am expecting some stuff to be better in the OUT and some to be better in the X0 project. I expect the X0 project to have a more coherent look overall.

A lot will depend if Lucasfilm are using the master that would have been created *before* they did the shoddy dirt/noise removal algorithm that threw all the detail away on the THX LD releases. If it is the pressing master then unfortunately all the trails and smearing will be in there and I would expect to get a better experience on the X0 release.


Post
#238403
Topic
Info: Original Trilogy in HD screening this November!!
Time
Ahh, but that is the difference - cracking a code, deciphering a language is a lot easier when you can directly read the characters, it is then just a matter of working out the code itself.
These old computer systems are more difficult in a way - you have a big magnetic sheet, you have no idea how the information is even physically stored, let alone what algorithm is used to store it. You don't even know what direction it is to be read in, what is data and what is flags, what encoding method (it predates EBCDIC and ASCII so even if you get the numbers off it, and work out which number actually represent the data - only then are you at the rosetta stone stage). Plus a lot of 50s scientific data was also encrypted on top of that because of the rampant paranoia at the time.
It makes an interesting challenge, tapes are easier as at least you know the direction and that the data is linear.
It is probably not impossible, but the CSIRO has been at it for nearly 10 years and haven't been able to get any data off them.

So there is a valid argument that stuff that is directly human readable/viewable is safer in the long long run as it removes the problem of having to have a device to read the data as you have it built in.

Of course the upside with digital is you can make a perfect copy, and if dilligent transfer it to newer media as it comes along, like the X0 project transferring laserdisc onto DVD - from that point onwards it could be pushed onto tape, HD-DVD, holographic storage etc. But if that wasn't done, would there be any laserdisc players still operating in 300 years time? If a digital file is released into the public domain then it is likely to still be around after a long period of time and if a digital copy of a digital original will be the same as the day it was struck. This is the first time in history that such a thing is possible.

Film has its own problems, countless important films have been lost completely (or partially) due to lack of interest by the studios themselves, it is easily damaged and degrades over time.

Last century Disney re-used most of their cels so the original artwork was literally washed away.
Lucas has supposedly cut up his original negs, making the OT non existent in its original theatrical form as far as a complete negative goes.

Companies cannot be counted on to maintain archives, the moment they lose commercial interest in a work, it is likely to be dumped.
Post
#238223
Topic
Info: Original Trilogy in HD screening this November!!
Time
Originally posted by: lordjedi
It's not analoguos at all. If NASA really wanted what was on those tapes, they could get it. There are companies in existence today that can read all tapes of backup tapes, even dating back 20 or 30 years. If they really wanted to spend the money, they could contract it out to a company to have all the media cataloged and labeled properly. And besides all that, I'm sure they've got most of the information on servers spread throughout the agency. No one I know makes backups and then deletes the original data (except maybe LFL )


Well no they couldn't actually. Australia also has archives of astronomy data that no known machine exists to read, and they are desperate to find a solution. My favourite are these huge flat magnetic sheets about 2 foot wide and and 3 foot long. They are being kept, but there is little hope that they will ever be able to be read again.

Whippersnappers may not realise how weird and primitive the early days of computing were. Companies developed and built their own storage solutions, there were really no standards to speak of at all. It wasn't unusual for a computer company to only make one or two of a particular device for a given application.
When that equipment was either destroyed, lost or ceased to function, if the data had not already been transferred then it became less likely with each passing year that it ever could. As the original companies that made the equipment ceased to exist and the people died off that created them, then it became impossible.
Go back 30 years and there is not too much of a problem, standards had come into being - go back 50 years and you are pretty much stuffed unless you were dealing with IBM. A lot of it is not even something people would recognise as storage media to look at.

Unlike consumer gear that was mass produced, the devices just literally no longer exist. For mass produced stuff you can usually find a player *somewhere*, it is easy enough to still find a betamax VCR, or an elcasette player, a VHD or ced player, a reel to reel VTR etc. simply because they made a lot of them. When a company only made two items which no longer exist and went out of business 50 years ago and all the techs are now dead - it gets a bit harder.



Post
#226454
Topic
Making our own 35mm preservation--my crazy proposal
Time
My last 2c worth.

Can a theatrical print hold more resolution than HD?
Yes it can but doesn't always. Anyone can make a crappy print, and it depends what you consider HD to be.
Almost any theatrical print that was shot well and printed reasonably holds more than 1080P does, that is easily proved by scanning a frame of any given feature at above 1080P and looking at the detail.
You could however find some features that have around 1080P or less detail wise if you looked hard enough.
Film resolution is dependant on lots of things. Weave reduces the resolution markedly, so depending on the camera used the resolution could be much lower than its theoretical limit.
Loss through the printing process - this also reduces the resolution because of light transfer loss and the grain issue - grain boundaries are different on each frame of film, so detail is lost with each optical process you do.
The prints I have messed with on Star Wars do have better than 1080P resolution, I can state that with certainty.
If you consider HD to be 2K or 4K then it comes down to a case by case basis.

Colour resolution is another thing. Once again, if you consider HD to be 8bit colour in 4:2:0 then any film print will have far better colour resolution. If you consider HD to be 12bit 4:4:4 then you come out about square either way.

Latitude, dynamic range, ANSI CR whatever you want to call it, film outshines current HD in most iterations of HD. However before I left we were working with some prototype HDR digital cine cameras that *far* outstripped the dynamic range of film (and recorded in OpenEXR format), they were incredible and should make it to market in the next 5 years.

Depth of field. This is probably the thing that makes the biggest difference between film and video as far as 'look' goes. It is also the easiest to solve. It simply means making the chip the same size as the film frame. Then you get the same depth of field, can use the same lenses etc. Why isn't the 'better' depth of field on small chip digitals better than films shallow depth of field?
Well, for dramas, comedies etc. a shallow depth of field is preferable as it leads the viewers eye to the subject and makes the backgrounds less distracting. That is a big reason why up until recently video always looked like video, even to the total layman.
For some things though a deep depth of field can be preferable, like 3D movies or wide shots etc. or even just for a particular look.

Film is a funny thing, a lot of its attraction is that we grew up with it, and we are comfortable with its 'look'. Had we grown up with a grain free recording medium (i.e. if film had never existed) and then film was introduced *now* people would probably want to know what is with all of the 'noise' in the picture. They would probably also want to know why the pans are all so blurry as well. (low framerate), and why it can't do true blacks. (OK perhaps an IB print could, but in a cinema a black frame from a theatrical print will always let some light through and give you grey)

We all grew up with film, as did the artists who use it, so we have come to find inventive ways to exploit its limitations and they have become 'features' to many of us, and we have grown to love the look.

There is no reason film cannot be emulated digitally, if the resolution was high enough (say 16K), the HDR system employed, then the rest can be simulated. Organic grain can be emulated exactly if you have enough rez to play with (you can either do it algorithmically or you can scan the grain from a reel of stock and transpose it into the picture). Weave through the gate can also be easily achieved, and DoF is the same if you use the right camera. So it could be done even today if you wanted to throw the time and money at it - but it would be easier to just use film anyway.

In practice though, not enough people in the general public (i.e. the people who pay to go to the movies) care about the intracacies of film and will be happy if the DoF is right and the picture looks good and they can't see any artefacts. A lot of people prefer a 'punchier' picture, (i.e. crushed blacks and high contrast) so even the reduced latitude of current digital doesn't bother them. I think the move away from film will be sudden once the new cameras come through. Film's death has been foretold falsely many times, but never before has there been systems available to do uncompressed 12bit with standard lenses and identical DoF - it will be too attractive to the mainstream studio stuff. It may never die completely, but I think it will disappear from the mainstream suddenly when it does go.

I agree that if you want the film look, you might as well shoot film, and that digital is a great hope for bringing something new to film making. Better framerates, HDR and as yet un-thought of possibilities. There doesn't seem to be much point in reproducing the limitations of film - you would be better off trying to exploit the new things that digital offers.

Short answers are though that:

1) In general 35mm feature films, even the prints exceed 1080P 4:2:0 in resolution and colour depth in nearly all cases. Once you get into uncompressed 2K or better with 4:4:4 colour then you could find examples in both camps that would exceed the other.

2) Distribution prints are nowhere near as good as negatives or even the archive prints, but are still very high quality, and still maintain well over 1080 lines of measurable resolution.

3) DVD is far below the quality of any film stock 720 x 576, 720x480, compressed with colour at 4:2:0 is laughably low resolution and doesn't even look great on 42" televisions let alone projection systems. Any argument that film only measures up to DVD is ludicrous. Even the T2 pics I posted show that easily. DVD is the MP3 of the visual world. i.e. It is great for what is is designed for (TV size viewing/ listening on poratble music players) but isn't great when pushed (watching on a large screen/listening on a great sound system in a good room). I'll say it again - any commercial feature shot even half assed on 35mm film exceeds DVD's capabilities end of story.

4) Generational loss. Optical transfers of film suffer generational loss pretty badly. This is why Lucasfilm went and bought up all the 70mm vistavision cameras as they had to do lots of optical composites for the OT which meant lots of generational loss with each composite. for non effects work though the generational loss between the negative and the print isn't severe at all.
HD can also suffer generational loss if it is captured using a lossy compression method and you have to do composites. If the frames have effects added then they have to be *recompressed* causing compression artefacts and detail loss. This isn't a problem if using a lossless/uncompressed workflow however. It is then usually recompressed with a different codec for digital projection which can cause another 'generational' loss. Once again however, if it gets to an uncompressed workflow from start to finish then digital will be lossless and generational issues will become a thing of the past.

5) Digital projection also currently can't do blacks and has a limited dynamic range, and limited resolution (in some cinemas) compared to film (or even CRT in some cases)

When it comes to the whole analogue vs digital debate to say one is "better" than another is to say that Jet is better than The Strokes, or oil painings are better than watercolours, or that Heavy Metal is better than Punk...


For our purposes here, if you could find a mint technicolour print, and could have a professional run it through a high end scanner like the arri with digital ice, then you would get a better result than the current DVD.

A standard theatrical print transferred by an untrained individual on lower end equipment would probably fall short of the DVD, but would be interesting.

As I said before, first step would be to find a blinder of a print, until you find that, the rest is just fantasy.


Post
#225615
Topic
Making our own 35mm preservation--my crazy proposal
Time
Originally posted by: boris
Just to nitpick, when you say film holds as much colour as picture information (which is true at the time it's filmed), and is comparable to 4:4:4 - what we're talking about is the original star wars negatives, which have deteriated so much that they need to be colour corrected scene by scene (and I might add that they didn't do a perfect job in 1993 or in 2004 with this correcting). With this in mind, wouldn't it be fair to say that the master star wars film reels do not hold colour as well anymore? Wouldn't this be more comparable to 4:2:0 then to 4:4:4?

No. Maybe you'd better read up on what 4:4:4, 4:2:2 and 4:2:0 actually means. It has nothing to do with faded or washed out colour. Even faded film still has a direct 1:1 colour to 'detail' ratio.

Originally posted by: boris

Also, keep in mind when comparing the quality of detail in the 2004 release, that a lot of that detail is in newly created digital elements introduced to those scenes - and there are so many alterations that are of course going to be better and more crisp then the original film, so that when you watch the entire SE movies an illusion is created making you believe it's more crisp with more detail then the film itself had.

No.

Take a look at any of the 'untouched' scenes, like closeups of Leia's face etc. The level of detail is stunning, and this is *after* the film image is downsized and had 3/4 of the colour information thrown away, and then compressed through a lossy codec.

Originally posted by: boris
Maybe I'm wrong, and the original SW negatives do hold picture information to the equilivant of 1080p - but I still really doubt it, I think a good quality interpolation upscale of DVD resolution would look pretty close.

Yeah, I'd have to say you are wrong on this one - I'll scan a super8 frame and post it when I get myself a new computer and you can see how much information even *that* holds compared to laserdisc or DVD. The SW negs definately exceed 1080p in detail and colour.

Really, I used to have to scan and put frames back out to film as part of my job, the difference between 2K and 4K and original film is instantly obvious when viewed side by side. Go down to 1080P and 8 bit colour and 4:2:0 and there is an even bigger difference.

I know Cameron and Lucas have opinions on HD, but then again Cameron prefers Pan and Scan over letterbox, and they are Directors, not DOPs or Cinematographers. HD has enormous cost and workflow advantages for a Director and *looks* very sharp. The older HD cameras like the Sony cinealtas have a massive depth of field which gives the *impression* of more detail in a scene as more of it is in focus. (Which makes it more suitable for 3D which was Cameron's focus in that article - you want a very deep depth of field for 3D)

There is absolutely not more information in a 1080P image though, you can test it scientifically and prove it - it isn't a matter of opinion any more than that a CD audio track holds more information than an MP3 file. You may not personally be able to hear the difference but it doesn't mean that the MP3 file holds the same fidelity.
We have shot HD and film side by side and scanned it and there is just a lot more information on the film negative.


I also disagree with what you're saying about laserdisc quality - I've watched Laserdiscs projected by professional-grade mounted movie projectors (thanks to friends who are complete movie geeks - and it sounds like you've watched them too) and the quality is good. It's not fantastic, of course, but it's still good enough to enjoy on a big screen. By the way, many independent films are filmed digitally at DVD resolution and are still more then acceptable theatrically.


Until I can get set back up and scan you a Star Wars frame myself, here is a grab from another old film (T2) comparing the DVD (ultimate edition) to a 816P scan from the film (The WMV 'HD' release - keep in mind the HD image is *heavily* compressed). If someone has digitised the laserdisc *please* post this portion of same image - it will be HALF the detail again of the bottom image!

http://mudgee.net/ot/t2.jpg

If you really can't see the difference between upscaled lasedisc, DVD and film (the above is just DVD vs low bitrate HD - both sourced from film - laserdisc would look much worse again) All I can surmise is that it was not a very good projector, not very well setup or your eye is *very* forgiving.
No amount of interpolation would make the 'subway' logo on her cup readable from the DVD version.

I've watched DVD upscaled with a Terranex on a G90 (possibly the best projector other than a Cine9) and it just doesn't compare to the same films on a native HD transfer - there is literally no comparison.
Laserdisc going through the same process just looks awful. It kind of looks OK until you play the same move in HD, then you can hardly stand to see the laserdisc. (and I am a BIG laserdisc fan, but it just doesn't cut it for big screens)
There are only 270 or so lines in an NTSC widescreen laserdisc. No matter what you do to it. It will never look in the same ballpark as 1080 lines or even 720 lines. Seriously there is no comparison - you can watch it, but it isn't pretty.
Where do you live Boris? I'd like to send you round to someone where you can see LD compared to HD properly on a decent setup. On a dodgy DLP or LCD projector with a bad scaler, or a bad setup DVD could look almost as bad as laserdisc I guess.

You are welcome to your opinion, and I have no problem if you can't tell the difference between a widescreen laserdisc and film, or think that the difference isn't important - but to say "I'm certain that any 35MM print will not have as much picture information as 720p, let alone 1080p" isn't correct, and is easily shown to be so. Even the image above shows how much more detail there is on film than on DVD, and it doesn't even begin to capture the level of detail of the negative.

It always good for people to discuss this stuff though, as it can help clear up a lot of misconceptions and cut through some of the marketing spin put out there.
Post
#225456
Topic
Making our own 35mm preservation--my crazy proposal
Time
Boris, i think you are a bit misguided in the belief that "I'm certain that any 35MM print will not have as much picture information as 720p, let alone 1080p.". Unfortunatley you are dead wrong on that score.
A few quick points on 'HD', DVD and film.
Film holds a far greater colour depth and a far greater resolution than DVD - this is true of 16mm and exponentially more so for 35mm. Even Super8 holds far more colour resolution than DVD.

For a start, DVD is 4:2:0, even consumer HDTV (1080i or 720p) is only 4:2:0 so your colour resolution is horribly, horribly compromised compared to film where there is effectively a 1:1 resolution match for colour to picture information.

A good quality theatrical print can easily resolve greater than 4000 lines of detail. You can point a camera at a 4K test pattern and actually resolve 4000 lines on the film, you certainly cannot do that with DVD. Weave reduces the resolution of film, but it is still way higher than 2000 lines.

Lucas didn't shoot on digital because he believed it was 'better quality' than 35mm film, but because it saved an estimated 1-2 million dollars in film stock costs (compared to digital tape) and also was 'ready to go' into post production, which saved a fortune on scanning costs and time, which is a big deal on a effects heavy feature like a Star Wars movie.

Also, when you say Lucas shot the prequels on HD, we are not talking conumer HD (i.e. 4:2:0 1920 x 1080 interlaced and heavily compressed.)
Even the aging Sony cinealta 950 camera captures colour in 4:4:4 RGB, and it is really old hat these days. The newer arri rigs shoot uncompressed 4:4:4 at over 3000x2200 (I can't remember the exact rez) in 12 bit colour (actually colour is better than that via the algorithms they use).

The newer SXRD Sony projectors used in digital cinemas run at 4096x2160 , although the early projectors were 2K (and IMHO look pretty ordinary).

In short, even 16mm film leaves DVD in the dust quality wise.

"If they transferred it again, and only removed large, visible, obvious deformities - it wouldn't be that different to the 1993 master, even if it was scanned at 720p or 1080p."
This is something a lot of people incorrectly assume.
If they fed the OT through one of the new arriscan machines, it would look immeasurably better than the 1993 transfers.
Just look at the extra detail in the newer scans of the OT used for the 'official DVD' vs the laserdisc transfers - even if you scale them down to 277 lines they have a lot more detail than the laserdisc or the D2 tapes. This is because of advances in scanning technology, and because the film holds more resolution and colour than any DVD of HD-DVD/BluRay format is capable of holding.

That is one of the beauties of shooting on film vs on digital. When attack of the clones was shot, 1080P was as good as you could get with a digital cine camera. Because they used that format, all of the elements shot on it are stuck at that point in technology, forever.They will never be able to get a 4K or 8K image with 16 bit colour out of it, because it is a 1080P master.
If they had shot it on film, then as scanning technology improved, so could a digital transfer. Had they shot AOTC on pin registered 35mm they could scan it today and get a 8K master with 4:4:4 12 bit colour, and it would have a lot more detail than the 1080P master they have now.

"Even at "laserdisc resolution" (roughly equal to non-anamorphic DVD) you can show a movie theatrically."

Um no - you really really can't. Even on a 10foot wide screen the 277 lines or so of a scope laserdisc looks appalling, you have to run it through a good scaler to make it even watchable. On a theater screen it would be a total joke.
Full frame DVD resolution (720x480 NTSC) is also a block-fest on a theatre screen, each pixel on a 24 ft wide screen is around half an inch wide (1.2cm or so) . You can scale it and so forth, but it still doesn't look any good.

Even for a good home theatre projector like a Sony G90 standard DVD is almost unwatchable unless you scale it up first, and even then DVD through a great scaler looks crapola compared to a decent HD-DVD or DTheater title at 1080i.
(BluRay doesn't look much better than DVD at the moment as it is currently only single layer and MPEGII - what on earth is Sony thinking?!?!). So no, a non-anamorphic DVD transfer could not be shown theatrically - it will look awful.

Moving back onto the thread topic

The problems for a preservationist looking at a film transfer however are a multi headed hydra.

As pointed out the sheer number of frames and storage is daunting.
172,800 frames for a two hour film. Scanning, processing and saving a frame every minute (which seems reasonable including stabilisation and super basic cleanup and reassembly into a 'movie' file) equates to 360 days of working 8 hours a day, 7 days a week.

Then there is the quality of a surviving print. 35mm prints of Star Wars were *never* offered for sale. Ever. This means any surviving print was stolen or *ahem* obtained. It also means that unless it was stolen extremely early in the piece, it went through the projector a *lot* of times. This was a popular movie folks!
It also means the print will be well over 20 years old. Any Eastman prints will now be pink. There will be surviving prints on other stock with the colour in good shape, and perhaps even ones that have not been through a projector more than 20 or 30 times, but they will be beyond rare.

Even if you got hold of one of those ultra rare prints, they will still be covered in an unimaginable amount of dirt, scratches and crud. What a collector would consider a 'mint' print will look very dirty under a scanner, and will require extensive cleanup. A 'good' quality print would need nearly every frame touched.
I don't own a 35mm print of SW, but have had access to some in the past, and while they were useful as a colour reference, none would have been good enought for a high quality scan.

Legality. I've been in the industry a lot of years, and I don't know *anyone* who would scan a 35mm print of Star Wars that is working professionally. They wouldn't just be risking their job, they would be risking their entire career.
They would also be risking a gaol term. If the print got into the wild, I'm pretty sure Lucasfilm would put a lot of resources into tracking down where it was done. They are quite forgiving of the 'home' user making his own versions and playing with laserdiscs, but if a film facility scanned a print I'm pretty sure all hell would break loose. (Speculation on my part of course)

It isn't point and shoot. Scanning a film isn't a 'put it an and press the button' proposition. Even a straight telecine requires tweaking almost on a scene by scene basis to get a good result from a print and not a neg. Scanning is even worse.
Newer scanners like the arriscan out at Weta make the job a lot easier, but there is still a lot of work to do. Even just getting the LUTs right is an artform.
It also means that it would be difficult for an employee to just 'run it through the scanner' during downtime (even ignoring that most scanners keep a frame count of frames scanned too, which is used for costings, maintenance etc. and a few hundred thousand extra frames gets noticed). It takes a lot of setup and work and is hard to do 'on the quiet'. You can get away with scanning your 5 minute short film project, but 6 or more reels of 35mm is hard to stuff up your jumper when people look in your direction. It is also hard to use up that much disk storage without being noticed, and then smuggle the files and film reels back out again.

None of the above means it is impossible. I reckon it is impractical and improbable, but I've been wrong many times in the past.

I still can't believe that Lucasfilm didn't scan all of the original camera negs, or that they cut up the original negs with new footage - it would be an act of vandalism to do so, but if he believes it is his film and he can do what he likes with it, there would be nothing stopping him just setting it on fire if he wanted to I guess.

So in short, if you are still keen then locating a print is the first step, and checking out the quality.You really can't get anywhere without a decent print or two to work from.

Step two would be to keep it to yourself lest Lucasfilm come after you to get their print back (it will still belong to them under the law I would think, and it wouldn't be the first time privately held 35mm features where 'recovered' by a studio).
Then look at what would be involved with getting a transfer done.

(BTW. a huge thanks to everyone here, I'm not officially back, and promise not to take this thread any further off topic - but I borrowed a computer and thought I'd drop in and say hi. Hopefully I'll be back properly in a couple of months and will rejoin the fray.)