Home of Professional Cinematography since 1996

class="Paragraph" Film-Digital Projection

Published : 13th April 2004


I saw the StEM DCI/ASC presentation yesterday and found the general message it promoted to be exactly as I have experienced with the D-Cinema test and projects I have been involved in - that 2K data images can match the best possible from a 35mm film source; a direct OCN answer print.

Throughout the test there were shots that looked creatively better in the film image or the D-Cinema image, depending on the creative decisions taken during the grading process, but overall the comparison was very close. It would have been nice to have seen a general release print in comparison...

What was very noticeable was the instability and 48fps flicker in the film projection, as well as the increased grain (neg and print grain in the film projection, neg only in the D-Cinema projection). This alone made the D-Cinema projection preferable to me.

The data was scanned 6K and down sampled to 4K for the DI process using proxy images for the creative process, which is a worry for creative decision making, and as the final projection images were further downsampled to 2K it really begs the question is 4K necessary? It would have been very interesting to see 4K scans immediately nyquist sampled to 2K for 2K DI post-production. My experience says the results would have been identical, making 4K scanning for 2K nyquist image post-production the ideal method

Dology for a business that needs to be profitable.

It would also have been very interesting to see how much detail a true 2K direct scan would have lost by comparison - especially from a Spirit, DSX or Millennium. Again, my experience suggests that if the scanning is performed correctly the results are not too dissimilar. But setting up telecines to perform accurate data log scans is not as easy as with a true scanner. However, obviously the speed improvement and cost saving have a bearing on the profitability of any DI operation so cant be dismissed.

I was a little unsure of the post process used for this project. taking a 14 bit linear scan to a 16 bit log file and generating a 12 bit linear projection image, although the disc system used could only show a 10 bit linea image for projection, seems a tad unnecessary.

Surely a 14 bit linear scan has a shadow detail granularity matched by a 10 bit log (CIN/DPX) file, as per Kodak's density mapping specification? If so, why not stay 10 bit log throughout, as all post-production companies tend to at present, as this prevents unnecessary data overheads and additional image processing that will introduce the possibility of unwanted artefacts?

And someone has got to ask what would a HD RGB (Viper/F950) version have looked like in comparison...?

I'm also unsure of the need to move to a new colour space for projection, especially as it has to be reconverted back to standard RGB for projection. Is this not making yet more complexity when there is no need?

And finally, what of a true D-Cinema delivery system using compression, etc.? No D-Cinema business model can afford to use uncompressed disc playback.

Anyway, it was nice to see general presentation showing that 2K is a match for a 35mm answer print. Can someone please now show a true 2K DI post-production process as described above to complete the learning curve?

Steve

Steve Shaw
Digital Praxis Ltd.
www.digitalpraxis.net



If Anyone interested in looking at or presenting...

The Film & HD Footage...I shot of the Line Charge explosions is welcome to it for looking at the same shot done in 5218 and in F900/3

Same FOV same action.

You are welcome to it.

B. Sean Fairburn SOC
Director of Photography
Castaic Ca



Sean wrote :

>If Anyone interested in looking at or presenting the Film & HD >Footage...I shot of the Line Charge explosions

Let us at it...is it a down convert? film? HD what? How would we project a comparison...or would we?

Nick Hoffman NYC DP



Nick Hoffman writes :

>Let us at it...is it a down convert? film? HD what? How would we project a >comparison...or would we?

Its 120 FPS Film Processed OCN and HDCAM 1080 Interlace F900/3 Original Camera Tape.

What you do with it is dependant on what you want to see
and how you want to see it.

Your creativity will be your only limit.

You Could :


a) Transfer the Film to HD and project both in HD
b) Transfer the HD to Film and project both in Film
c) Transfer the Film and HD to Digital Intermediate and do a split screen...then project both in Film and HD
d) All of the above

I hope to Transfer the Film to D5 then do a tape to tape HDCAM to D5 then do a Split screen of both all on the D5 Tape.

The shoot was not done purely for a comparison But instead to get the Shot So one problem I will have is making the HD at 60I shot for slow-mo keep up with the right tool for the job Film at 120 Frames per second.

In that regard a comparison split screen will be difficult at best to pull off.

> What was your impression of the results?

I haven't seen the film yet but the HD looks incredible.

/3 did a wonderful job of holding the highlights of the explosion.

I hope to see the Film soon $$$ always in the way between me and what I want to do.

B. Sean Fairburn SOC
Director of Photography
Castaic Ca



Frame rate was 60 I presume? de-interlaced somehow?

Nick Hoffman NYC DP



Steve Shaw wrote:

>I saw the StEM DCI/ASC presentation yesterday and found the general >message it promoted to be exactly as I have experienced with the D->Cinema test and projects I have been involved in...

Where did you sit (number of image heights from the screen) when you made those judgements?

I was not able to attend the screenings at the ETC in Hollywood this week. But last week I did attend the 2K Digital Cinema screenings at the ShoWest theatre owners convention in Las Vegas. Certainly the screenings looked very acceptable when I sat in the middle rows of the Les Theatre Des Arts at the Paris Hotel in Las Vegas (>2 image heights), or in the ballroom back bleachers for the 20th Century Fox product reel. But for two screenings, I sat in the first few rows of the theatre (about 1.5 image heights from the screen), and still found the 2K pixel structure to be noticeable. I also found the opening scenes of the "Shrek 2" footage that was screened on Thursday afternoon to have obvious contouring issues, with the pages of the tan coloured "storybook" looking like a topographic map, rather than the uniform shading you might expect.

John Pytlak
EI Customer Technical Services
Research Labs, Building 69
Eastman Kodak Company
Rochester, New York 14650-1922 USA
http://www.kodak.com/go/motion



John Pytlak writes :

>But for two screenings, I sat in the first few rows of the theatre (about 1.5 >image heights from the screen)

Does anything look good in the first couple rows?

Noting where you're sitting is a very good question to ask, but frankly, practically every movie theatre I've been to and have been forced to sit in the first couple rows has been a very poor viewing experience-and that was with film. So 2K or film, I'd rather not sit in the first couple rows-not that people won't sit there, but I don't think it's fair to assume film is without fault at that viewing distance-at least with the theatres I've attended in the Dallas/Tulsa/Norfolk area. No, it's not LA/NY, but we watch movies here too...

Jason Rodriguez
Post Production Artist
Virginia Beach, VA



John P. Pytlak wrote :

>But for two screenings, I sat in the first few rows of the theatre (about 1.5 >image heights from the screen), and still found the 2K pixel structure to >be noticeable.

You can easily see sharpness differences between the film and digital versions at 1.5 image heights from the screen. Film having a bit more detail.

Film appears to have enough headroom that (in a good theatre with a good print) you can be fine sitting in the front row. Digital does not appear to have the same headroom for close viewing.

I find it annoying enough that I make sure I don't sit in front for digital screenings.

Jim Houston
Pacific Title Imaging



>Does anything look good in the first couple rows?

Being rather interested in image quality I always sit 1 1/2 to 2 screen heights from the screen.

If you sit in the back of the theatre you may just as well be in the lobby when it comes to any serious assessment of image quality.

Regards

John Lowry
Lowry Digital Images



I don't normally like sitting much closer than 2 image heights, but many modern theatre designs force some of the audience to sit that close, especially during busy weekend screenings of popular movies.

Yes, film grain can be an issue that close, especially with older high speed films, underexposed scenes, or smaller formats (Super-35 vs. anamorphic). But for me, random "noise" is much less objectionable than fixed pattern "noise". And jaggies, aliasing, and contouring always catch my eye, since they seem so unnatural in an image.

John Pytlak
EI Customer Technical Services
Research Labs, Building 69
Eastman Kodak Company
Rochester, New York 14650-1922 USA

http://www.kodak.com/go/motion



John,

>Where did you sit (number of image heights from the screen) when you >made those judgements?

The Pacific Theatre floor is marked in screen heights. I chose my seat at 2.

Howard invited the audience to get up during the screening to get as close to the screen as they wished. I took it to the extreme of viewing the screen for part of the film at a distance of about 12 inches. With my nose practically on the screen what I noticed most was noise that looked like film grain. Pixelation did not seem significant, at least in the scene I was viewing.

I asked how much of that noise was film grain and was told that it depended on brightness, that in the blacks dithering was adding to the effect. All the footage shown was live, not CG.

> I also found the opening scenes of the "Shrek 2" footage....

Shrek 2 wouldn't have film grain to help hide banding.

Robin Rowe
Editor
Hollywood, California
www.DigitalFilmLibrary.org



I agree with Robin and also varied my viewing distance. I've also been able to perform these test in other cinemas and with HD sourced material too. I have not seen any fixed pattern noise, jaggies or other 'digital' artefacts when using 2K projectors.

Digital, when used with experience, can match film in all respects in my view.

The problem is that many tests to date have not been performed correctly with the different requirements for film/digital not properly understood.

When film sourced, the 2K digital image can match print film in resolution and dynamic range without any artefacts, if processed correctly - just as bad capture/processing can make film look bad.

To my mind the benefits of low grain, stability, lack of flicker alone make digital preferable. And with 2K 10bit images matching film, as the DCI test showed, things are getting interesting.

On the digital praxis website there are split screen shots of film vs. Viper images too, showing what today's digital capture is capable of...

Steve Shaw
Digital Praxis Ltd.



Jaggies depend on the projector and the image content. There need to be no jaggies with 1K projectors. Of course the price is softness.

I find the amount of aliasing to be quite varying on different HD material I see over the air. The DCI material scanned at 6K and downsampled to 2K should be practically aliasing free if high quality filters were used.

For some reason, and I would love to know why, HD transfers of Warner Brothers films have more aliasing issues than other studios', like, Columbia Tristar or DreamWorks. If this is on the master tapes or for some reason only on the broadcast version I have no idea, but it's definitely the case.

Examples are : The Majestic, JFK, Michael Collins, Liberty Heights, Assassins, Angel Eyes, Pay It Forward, Miss Congeniality, Wild Wild West.

Michel Hafner
www.imdb.com