Digital Intermediate; Does anyone not use it?

[K Borowski]

Hi. I have been hearing more and more about DI these days, and it bothers me that no one seems to have an issue with the loss of quality and digital artifacts that this process adds. Are all movies being digitized and titled this way now? Are only segments put through computers, or is it the whole movie that will go through? Why can't people just stick with analog? Sorry for sounding so old fashioned.

 

Regards.

[GeorgeSelinsky]

Personally I don't know how some people do HD digital intermediates, unless they shot originally in 16mm. I saw a few 2K digital intermediate jobs and I wasn't too impressed, especially on the wide shots.

 

The fact is that the price/performance ratio is not yet feasible, where DI's can become a replacement for the optical process. But when this is fine tuned in a few years, I guarantee that it will be more accessible and allow you more flexibility. However, I'm sure it will always cost MORE than the optical process, that is for sure.

 

I'll never forget how I was talking to one animator who was doing SFX work. He invented a low cost device which made his work go much faster. I said to him "Your clients must have been very happy, it cost them less I gather". He said "Are you kidding me? I charge them more for the job now because they get their work back faster!". Happens a lot from what I hear.

 

- G.

[Max Jacoby]

I don't think you are sounding old fashioned.

 

Personally I absolutely loathe the look of DIs. It seems to suck the three-dimensionality out of faces, making them look flat and plasticky. Also the blacks become kind of muddy, there is no real snap in the print anymore, even if printed on Vision Premier. I can always spot a DI and I have yet to see one that I think even looks halfway decent.

 

Of course most people don't share this opinion, just read any American Cinematographer Magazine, with seemingly every Dop raving about DIs. Which actually is kind of worrying I must say, since it appears that they themselves don't seem to be aware how it alters the look of their film...

[K Borowski]

I wish I had gotten into film earlier. I can't tell the difference, having seen Passion for instance and not being able to tell that it was a DI. It appears that there were many bad prints of this film too. I noticed color shifts between the different 1000 foot rolls. ARE all films done DI for things such as titles nowadays? Does anyone do true optical special effects? I am seeing less and less concern for quality in movies. I didn't see the Incredible Hulk when it was in theatres for instance, but seeing the video, the whole thing must have been done with DI and I'm sure there was an attrocious loss of quality in the prints in the theatres. Of course, I don't think there is any analog way to do what that movie did. I don't know. Digital has its place, but it bothers me that there is less and less room for models and visual effects. I think movies such as Star Wars I and II and Matrix (II in particular, the last I saw before becoming sickened) are all dreadfully fake-looking. It's pretty sad when George Lucas put out better poop with baked potatoes, battleship kits, and a pickup truck than he's done with all of those banks of computers and computer programmers.

 

Regards.

[David Mullen ASC]

I just saw "Van Helsing" and thought it looked pretty good -- pretty fine-grained for a 1.85 film and very few artifacts from the D.I.

 

Considering that some very talented DP's have used D.I.s and some have gotten Oscar nominations for that work, I'd say it is an exaggeration to say that all D.I.'s look bad. "Amelie" was a beautiful film, as was "O Brother Where Art Thou?" And there have been a number of films that have mixed D.I. work with normal work and it's not noticeable. People forget that the first "Lord of the Rings" was about 70% a D.I. and the rest normal.

 

People also seem to forget that optical printing as with a conventional blow-up from Super-35 also produces "artifacts". It's a bit unfair to say that all photo-chemical and optical artifacts are good but all digital artifacts are bad. Personally, I prefer the look of using a D.I. to blow-up a Super-35 movie to anamorphic than using an optical printer. It's almost the only way I'd be talked into shooting in Super-35...

 

On the other hand, I'd probably only want to use a D.I. if it allowed me to do things I couldn't do photochemically.

 

No, most feature films in theaters still don't use a D.I.

 

Anyone who has spent time in a DaVinci session using Power Windows can tell you why they wish they could do this for theatrical work. On the HD transfer for "Northfork", for example, I was able to bring out all sorts of interesting cloud formations in the hottest parts of the sky using Power Windows, sort of like dodging and burning work in printing stills.

 

As for the fakeness of digital effects in movies, that's a whole other issue. I still think miniatures have more of a weight and presence compared to a digital model.

 

Actually, I don't think "The Incredible Hulk" used a D.I.

[Max Jacoby]

  I can't tell the difference, having seen Passion for instance and not being able to tell that it was a DI. 

'Passion' was one of the worst DIs I have seen recently. Especially when faces were lit with two sources of different color temperatures. The way these different lights touch each other on skin, how they smear into each other, it just looks plain wrong. And that was done by E-Film, one of the best respected DI places in LA...

 

I don't mind the grain of an optical Super35 blowup. Since grain isn't the same thing as sharpness, I'd rather have a bit of grain, but a sharp film where the colors are nice and there is a a pleasant texture, instead of something which looks like a Dvd.

 

I think the concept of DIs itself is great, but unfortunately in my opinion it just isn't ready yet. When I colortimed a film I made recently, I spoke to the people at the lab and they feel the same way.

 

I realize that many Dops whose work I respect are using DIs. But if someone like Steven Poster tells me over at CML that often he can't tell wether a film is DI or not, than I find that worrying.

[Max Jacoby]

Actually, I don't think "The Incredible Hulk" used a D.I.

You're right David, it isn't.

 

As Frederick Elmes says in the July 2003 AC article:

 

"Our tests told us that the digital-intermediate process isn't completely transparent yet. The work done at ILM is very good and I have complete confidence in them, but we didn't want to add this additional layer over the rest of the film as well"

[K Borowski]

If they didn't use DI for those "comic book frame" shots where you see several different camera angles at the same time and the frames actually move around in the frame (does the way I said it make sense?), then how did they? I know they have some sophisticated optical printers still hanging around and I understand the basics of it, but adding multiple moving elements to the picture? Surely they had to do some sort of digital manipulation and compilation for that to work. Maybe it wasn't a true DI. In the AC article, did Holmes mention how that effect was accomplished?

 

Regards.

[Mitch Gross]

That "comic book" effect is an old one, and you can look back to old films such as the original "Thomas Crown Affair" to see them done on optical printers. That said, anytime an effect is seen on screne such as the Hulk himself, you can rest assured that this is material that was at least partially generated in a computer. The difference here is that the individual shot or sequence is burned out to film and then spliced into the cut for traditional printing. But for individual effect shots, a "background plate" of 35mm film is scanned in to the digital environment, manipulated electronically, and then burned back to 35mm film.

[Filip Plesha]

It is not in every case than bad DI is used. For example, Stuart little 2 used

4K digital scanning (on a real scanner) for almost the entire film, you could consider this a DI job as well.

I think a lot of DI problems come from the use of machines like spirit datacine.

Real scanners have a much better output. Plus, compression is used often.

 

There are fast and good scanners like northligh, but they are still a lot

more expensive (and slower) than hybrid telecine machines.

 

I think that if you use no compression and at least 4K resolution from a good scanner like gensis or northlight you can't tell any difference in the prints (even prints made direct of the original negative). That all goes if you use a good digital printer. Just look at the projects of digital restoration of some films, like the ones cinesite did. They scanned entire films into resolutions as big as 6K (for vistavision)

I belive their digital to film outputs are very very close to the originals.

 

 

All in all i see it this way:

 

as long as filmmakers do not consider their DI film outputs as being archival

it doesn't really matter that much. I mean prints are prints, they come and go,

and usually they are low in qualitty compared to the originals.

If original negatives are stored as the archival material, films can be rereleased

in better qualitty on some future home standard or in cinema.

 

But when someone puts a negative printed from a compressed HD tape

into the vault, then i think it is a ruined film.

 

But cheer up. Some day 4K pin register scanners (and storage and output) will scan fast and cheap, and it will be silly to use telecines.

[David Mullen ASC]

I think the comic book panels in "The Hulk" were combined digitally (although I swear I read elsewhere that it was done the old-fashioned way in an optical printer!) but that's different than saying the whole movie went through a D.I.

[Max Jacoby]

I think a lot of DI problems come from the use of machines like spirit datacine. Real scanners have a much better output. Plus, compression is used often.

 

There are fast and good scanners like northligh, but they are still a lot

more expensive (and slower) than hybrid telecine machines.

'Cold Mountain' was scanned on a Northlight and I found it had the same problems as any other DI out there

[Filip Plesha]

Yes but at 2K, and arrilaser was used.

From what i have seen Celco printers have a better output than arrilasers

(stronger blacks,sharper image and cleaner colors) and as for laser recorders,

i think kodak lightning recorders are better (judging by the work cinesite has been doing in the past)

 

I think that the biggest problems in DI are resolution and output machine.

 

By the way,what were those problems audiris (i havent seen the film)?

[David Mullen ASC]

My personal experience has been that Celco CRT transfers are softer and grainier than Arrilaser outputs. All the low-budget efx companies in Los Angeles use Celcos and I've never been very happy with the results.

 

There's a reason why Celco transfers are half the price of an Arrilaser transfer...

[Filip Plesha]

I was refering to the new xctr technology (not conventional CTR) that is in

the new furry. They won an academy award for it.

 

Traditional CTR technology is indeed inferior to laser technology.

[Max Jacoby]

By the way,what were those problems audiris (i havent seen the film)?

It had the usualy DI Problems: wrong, plasticky skintones, weak blacks. On top of that it was really grainy. Using Super35 with 500 Asa on wide, exterior sunlit shots is not a really good idea. I am not a big fan of John Seales work in general, but in this film his approach of using only 500 Asa, even if it is the 5218, hurt the look of the film. Especially one shot during the battle in the beginning looked like blown-up Super8.

[Filip Plesha]

Well you can't blaim DI for grain, but as for the skin tones.. i just don't get it.

There are some projects that have almost invisible scanning/recording done at 4K.

But the resolution should not affect the skin-tones, it's apples and oranges. Perhaps they used compression or something.

The film recorder could be blaimed for blacks, saturation and contrast perhaps,

but i doubt it would distort skintone colors like that.

 

Does anyone know for sure what things in the whole DI process are guilty for

bad image qualitty?

 

p.s. audiris, there is allso one more thing to consider. Since the whole idea of DI is to manipulate the image, you can't really judge any of the image properties without knowing the filmmakers intentions

[Max Jacoby]

That's the whole point:

 

Although skanning and recording should be invisible, it isn't. Because it usually takes me less than 2 minutes to recognize if a film is DI or not. And I ALWAYS recognize it (and I don't like it, as you might have guessed by now).

 

I think the difference would be much more obvious to most people if you had a side by side comparison between a regular print and a DI.

 

All I can say is that when I explained the problems I had with DIs to a friend of mine, who is an sound engineer in the music industry, he knew exactely what I meant. He said the music indusrty went (and is still going) through something similar: they have digital recorders that theoretically are good enough (they cover the whole latitude of audible frequencies), but people still prefer to use analogue recorders, who record 'more' than one can hear. They can still hear a difference, although there shouldn't be any anymore.

 

Funnily enough, for all my DI bashing, in a film that I made recently on 35mm anamorphic, there is a DI shot in it: we had to extend the lenght of a shot and it got done digitally. I didn't mind, because neither I nor the Dop (who hates DIs as well) didn't notice any difference. But that was because it was the close-up of an old teddybear :D

 

But for another shot (fade out of a close-up of a boy) the lab did it digitally first, but we asked them to do it optically, because we didn't like the way the DI shot looked.

[John Pytlak RIP]

Here is the latest from Kodak Cinesite:

 

http://www.cinesite.com/?1611&0&3537

 

http://www.cinesite.com/?122

 

http://www.cinesite.com/?1223&0&3994

 

http://www.cinesite.com/?1441&0&880

[K Borowski]

So all of those movies have NO optical prints but rather some digitized representations transferred to the final prints??? I'm glad I haven't seen any of them in theatres. Those must be dissappointing compared to true optical prints. I guess this is a sign of things to come. I don't see what's the use of even shooting film if the middle steps are run through a computer anyway. There are a lot of people interested in those new 6 mp cameras they have out. No one gives a poop about quality anymore it seems :-( I guess it's only a matter of time before cine goes the way that still photography went. I'd better start stocking up silver and acetate rollers :-( John, don't get me wrong, I love your company and it's devotion to film, but sometimes EK makes some really poor decisions. I think this process of digital intermediates is doomed to failure because it detracts from all of the advantages that film has. If the final output is either digitally recorded film or digitally projected (makes my eyes sore) video-whatever, I think a lot of producers are gonna start going the way of Lucas. When I found out that most of television is shot on film a couple years ago, I was amazed and thrilled that people still cared. Now I don't know anymore. Most people thought Star Wars II looked great and had a great story line (I thought it looked like poop and was written like poop, but I guess I'm alone here). This trend towards having to do EVERYTHING on computers is honestly a source of great confusion and depression for me. I thought that I might have some sort of future as a filmmaker or photochemist; now that future isn't going to be possible because people are so lazy that they won't take time to light because they can just use DI instead of having guys with light banks or reflectors. No more prints, no more labs, no more optical printers, no more new stocks. Or why wait for dailies when digital is "good enough"? I don't know.

 

Regards.

[John Pytlak RIP]

Digital Intermediates are a TOOL available to filmmakers. I'm sorry you haven't chosen to see any films using digital intermediate. (Although I suspect if you have seen ANY films in the last five years, you may have seen some scenes that used digital intermediate, and not known it).

 

And Kodak continues to advance the "state of the art". We've always tried to "raise the bar" on quality, while reducing the cost by improving productivity of scanning and recording:

 

http://www.kodak.com/country/US/en/motion/...v2/sehlin.shtml

[K Borowski]

John, I have seen a film or two with DI used for titles and selected scenes. Apparantly Passion was one such film. I merely mean that using DI for a whole movie is overkill, especially in light of the fact that there is a SIGNIFICANT loss of information from the master neg's. I applaud Kodak's EXR, Vision, and Vision@ product lines and hope to see Vision3 one of these days :D I just don't think that DI is an alternative for proper lighting and reflector use. I don't think that titles should be DI either, since optical titling is a fine alternative, even if the titles themselves are burned from a computer. My point in making this post is that DI is starting to become over-used. I have heard all of this talk about how 16mm should all be done on DI now since there's no signifcant loss in quality or something. I disagree. When DI improves (maybe 8K???) then I will re-evaluate my position.

 

Regards,

Karl Borowski

[Filip Plesha]

My god, and I thought I was a film purist...

 

You are overreacting. If you don't mind me saying so, the problem is in your subconsciousness. Your brain automatically connects "digital" with bad qualitty.

Even if they made a perfect 8K DI, you would probably be unsatisfied.

I think you need to relax, explore every case of DI and see how good can it be.

 

Like i said, a digital restoration is allso a DI process in some way. There are such fine examples of 4K digital restoration work done by companies like Cinesite.

If that kind of qualitty standard was used in your typical DI, it would probably be

undetectable.

 

Not every DI work is bad. Like John said, you probably saw thousands of

scanned and recorded shots that you were not aware of.

 

Now i'll tell you something from a psychological point of view (since i am studying psychology). You go into cinema with preprogramed views on the subject of DI , you sit, you see a tiny artifact which trigers your belief that the film went through DI, and sudenly the power of suggestion comes in and you start precieving the image as bad qualitty. Perhaps you have seen worse optical prints in your life, but to your brain they were "better" because you knew they were optical. I am not saying you are seeing things. I am saying that your brain exaggerates certain facts because of your beliefs.

(if this was not true, any kind of advertisment would be in vain, and hypnosis would not work)

[David Mullen ASC]

I've gone to see split-screen projected demos of D.I.'s compared to a 35mm contact print of the SAME footage and the differences are minor at best, NOT a "signficant loss" by any means.

 

And blowing something up optically through an IN and IP step also entails some loss of information -- it is not an artifact-free process either. The real issue is whether what you gain from the D.I. offsets what you lose. Super-35 blow-ups, for example -- in a 2K D.I., the end result is softer but less grainy, while an optical with dupes is sharper but grainier. Which result is preferable is a matter of taste and the particular visual style of the movie. For "O Brother Where Art Thou?" for example, the slight softening was fine because it was a period film shot without filters; had no D.I. been used, another DP might have used a light diffusion filter (not Deakins though since he doesn't like diffusion filters.) For example, for "Amelie" the DP used Zeiss Ultra Primes unfiltered because he wanted to counteract any softening from the D.I. but the end result was actually a little sharper than he expected.

 

As for compression artifacts and chroma noise, these are less acceptable in a D.I. was sometimes are the result of deliberately pushing the image more than the system can handle, which is not really the fault of the D.I. process itself. I've seen many D.I.'s which are pretty clean of these artifacts.

 

Using an optical printer and rephotographing an image through a lens tends to "edge enhance" the grain particles, creating the typically sharpened-but-gritty look which is not always desirable (for something like "Seven" it's great but for a romantic comedy, less than ideal...)

 

Take a look at the 65mm restoration of "My Fair Lady". There are only TWO digitally restored shots: the title sequence of flowers and a close-up of Audrey Hepburn that had a big emulsion chip painted out. In particular, on a large screen at a 70mm screening that I went to, that close-up intercut seamlessly.

[Filip Plesha]

Allso i belive that this problem with softening (in the super35--->scope conversion)

would not be there if 4K resolution was used. In that case there would be nothing to loose by using the DI. Plus, not every DI work is compressed, and color noise

varies from scanner to scanner. With datacine you get a lot more noise and unprecise color than with a kodak lightning scanner.

(and the new arriscan uses a CMOS sensor for even lower noise)

 

By the way, David, who did the restoration of "My fair lady" ? And what resolution did they use?