This is topic Image resolution of 35mm cinema film in forum Film Handlers' Forum at Film-Tech Forum ARCHIVE.


To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=005886

---

Posted by Richard Fowler (Member # 893) on 02-23-2004, 11:55 AM:
 
The SMPTE Journal, February 2004 issue, has an article regarding film resolution in a theatrical setting.To paraphrase the seven page article, they shot a custom resolution chart using industry standard cameras with Kodak stock; made wet gate answer prints and normal negative, IP, dupe negative and release prints on different printers and release print stock.
In the negative stage the resolution was 2100L/Picture height, answer print = 1400L/PH, release print 1000L/PH. They projected the release prints in selected cinemas In the USA / Canada / France / Italy, the highest resolution was 875L/PH and the average was 715L/PH. In fairness they noted that higher resolution in the cinema is possible but they where looking for what today's filmgoers receive when they go to a typical, state of art, cinema.
 

---

Posted by Floyd Justin Newton (Member # 1316) on 02-23-2004, 12:22 PM:
 
Richard--

NOT very much, at least here When you have wannabe operators
(children) in the booth who don't even know how to focus a
picture on the screen. I don't worry about resolution or any
thing else in the cinema today, considered the way they are
run (ruined).

fjn
 

---

Posted by Christian Appelt (Member # 1067) on 02-23-2004, 01:30 PM:
 
Do you mean LINES or LINE PAIRS, which I know from many measuring procedures? That makes quite a difference, and 2100 LP from original negative comes close to the practical 4K resolution for 35mm scanning/recording.

I believe that most of the original quality is lost in high speed mass printing AND improper duping (John P. has reminded us in numerous threads about the proper procedure which requires step printing at least once, and looking at the release patterns, I doubt that the studios really care about that extra image quality. So most of what is preserved during post production will be visible only for those few theatres that get EK or "grade A" prints, the rest of the world gets whatever can be produced cheaply.
Many good films, i.e. THE LAST SAMURAI have been totally destroyed by improper release printing, that level of quality is enough to make people wait for DVD. An no amount of care on the exhibitor's side can make these prints look better or even good.
 

---

Posted by Larry Myers (Member # 738) on 02-23-2004, 01:37 PM:
 
Hi

I am not sure what they mean by L/PH. Most film image resolutions are measured in Line Pairs per Millimeter or LP/MM or L/MM for short.

A perfect f-1 lens in noon summer sun runs about 2000 l/mm center resolution. Most name brand still camera lens will do about 500 l/mm in noon summer sun center resolution. Most professional still images from slides with ASA speeds around 100 are about 75 l/mm. Typical 35mm release print resolutions I have seen run between 10 to 30 l/mm although some can go as high as 40.

Old projection lenses (Kollmorgen) can actually be half the resolution as a new ISCO HD lens. Now add someone not focusing the projector and you have a real mix and match of presentation quality.

Larry
 

---

Posted by Manny Knowles (Member # 1171) on 02-23-2004, 02:33 PM:
 
How do these findings affect the argument that dCinema's 2K resolution doesn't match 35mm?
 

---

Posted by Scott Norwood (Member # 30) on 02-23-2004, 02:39 PM:
 
What format is this? Academy or scope will have greater resolution than 1.85, given the same screen height.
 

---

Posted by Richard Fowler (Member # 893) on 02-23-2004, 02:52 PM:
 
Lines per picture height based on a custom chart using sine wave patterns to permit more accurate MTF reading from each stage of the film process, from negative to projected image. The chart had a mid-gray background which also allow for more accurate evaluation of the image. Based on the chart set-up the maximum could equal 2400L/Picture Height; the chart was shot to resolve 2100L/PH to allow the more target wave patterns to be spread out on the image area to help in evaluations. Line pairs to check lenses, especially on a collimator is fine but the lines per picture height is a good guage to compare 35mm to a digital image. They noted that they tested the camera lenses and found that they dropped 25 - 30% from their maximum ( ideal resolution based on the focus / stop used. It appears that Academy was used ( there are a couple of sentences in the paper which are confusing ) with the 1.85 chart apparently cropped which yield the 2100L/PH. The projected screen evaluation was a 1.85 image......so true anamorphic would of course be better.
 

---

Posted by Larry Myers (Member # 738) on 02-23-2004, 03:09 PM:
 
Hi

Image resolution has nothing to do with format.

Really, the only imputs to image resolution or the ability to see detail, are the focal length of the camera, the distance to the subject and the type of film being used.

This is a very hard concept to understand so I expect to get a little flack on this.

Larry
 

---

Posted by Richard Fowler (Member # 893) on 02-23-2004, 03:42 PM:
 
Larry...on the same page. The article is of course much more detailed regarding lenses, equipment, type of film stock, microdensitometer tests, use of different labs, original test film shot in L.A. and France to distill into a couple of paragraphs....
 

---

Posted by Michael Schaffer (Member # 1204) on 02-23-2004, 03:58 PM:
 
Call me stupid, but I still haven`t understood if you are talking about lines or line pairs. Could you clear that up for me please?
 

---

Posted by Larry Myers (Member # 738) on 02-23-2004, 04:22 PM:
 
Hi

Lines per millimeter is a unit of measurment much like dots per inch, miles per gallon and so on. It's a way of telling just how sharp an image can be. The eye sees about 8 lines per millimeter. At 4 lines per millimeter things start to look soft. At 16 lines per millimeter, things look very sharp. It's impossible for the unaided eye to see the difference in sharpness between 20 l/mm vs 40 l/mm unprojected motion picture film strip. Both will look very sharp to the eye yet when projected, they will look very different.

Larry

[ 02-23-2004, 05:52 PM: Message edited by: Larry Myers ]
 

---

Posted by Bobby Henderson (Member # 840) on 02-23-2004, 05:47 PM:
 
Just to hairsplit, digital projection is not up to 2K standards --at least 2K is not representative of the installed base of digital projector systems. The 2K chips are just getting to market. And they apparently require entirely new projection systems for install to replace the more common 1.3K (1280 X 1024) DLP format.

Anyway, the line pair figures would be easier to understand if they were broken down into numerical terms similar to pixel values for the frame, such as the 1280 X 1024 figure I quote for DLP.
 

---

Posted by Steve Kraus (Member # 476) on 02-23-2004, 06:23 PM:
 
I'm with Michael. "Lines" has no meaning (except in the video sense). They have to be talking line pairs--that is, a black line next to a white line. You can't have just lines. Black lines next to black lines? That's laughable. But it's important to get the terms right because if you're talking about a fixed number of pixels like with digital I believe the line pair resolution is going to be something less than half the number of pixels in that direction.

Besides the terminology, I have two objections:

Line pairs across the image or line pairs per mm of film is a function of what contrast level you are looking at, even when starting with pure black and pure white. Look at a test chart under a microscope you'll see that rez goes on and on at less and less contrast until it disappears into muck. If you are being given a fixed number like so many per mm then it means someone has decided for you that they are going to count this far and no farther. Yet in reality it may go on somewhat farther but at lesser contrast (less ability to distinguish between what was once a black line and a while line). I am guessing that digital goes to a certain point then just ends abruptly. But with film there is no abrupt end. That is why resolution is best described as a function where line pairs/mm (or across an image) is related to contrast level...you can see how it goes on and on but peters out into muck. Remember, that "function" means a graph...if you aren't shown a graph you aren't being told the whole story.

The second objection is how the measurement is done. I tend to think that there is an integration over time which will penalize film projection for jitter in the projected image. It is my firm belief that human vision handles jitter without it penalizing the perceived resolution the way it does would if a camera photographs the screen and multiple frames in slightly different positions are combined into one for measuring purposes.
 

---

Posted by Gordon McLeod (Member # 33) on 02-23-2004, 08:59 PM:
 
Image resolution is a factor of all of the above but magnific ation is the and stability is the bigest
Native resolution ie what can be recorded on the film is one thing but what can be reproduced on the screen is the only number that realy counts
 

---

Posted by John Pytlak (Member # 331) on 02-23-2004, 09:23 PM:
 
Steve Kraus wrote:

quote:

---

I tend to think that there is an integration over time which will penalize film projection for jitter in the projected image. It is my firm belief that human vision handles jitter without it penalizing the perceived resolution the way it does would if a camera photographs the screen and multiple frames in slightly different positions are combined into one for measuring purposes.

---

quote:

---

Did the article state the motivation for carrying out this research?

---

quote:

---

It appears printing was all done using continuous contact printers, so proper procedures to handle printing onto a short-pitch raw stock during duplicating were not used. Assessments of print resolution were made subjectively, from a distance of 1.5X the picture height. It was not reported if higher resolution could be seen from closer to the screen.

---

quote:

---

That means that some trailers are first strikes using original negative outtakes plus special takes that were made just for the trailer. This way, the normal production isn't held up and assures very high quality trailers.

That was the whole point of the writeup. Newsweek called this a bait and switch. Seems that some scenes in the trailer are not in the movie and the image quality of the trailer far exceeds the image quality of the final production.

---

This strays from the image quality issue, but trailers have been done that way since the beginning of trailers. Lots of classic films have scenes in trailers which aren't in the movie, due to the scene not being used, an alternate take, etc. The oldest, easily viewed example would be on Milestone's new "Phantom of the Opera" DVD, which includes an original trailer. The trailer contains a brief, cool shot (unused in the film) of Lon Chaney riding upwards on the counterweight for a hidden door mechanism. Often it's due to trailers being edited simultaneously & independently of editing the film, & just sending stuff over to be used. With production & release windows even shorter, it's not surprising it still happens.
 

---

Posted by Larry Myers (Member # 738) on 02-25-2004, 09:33 AM:
 
Hi

I am not sure my comments about high quality trailers strays form this thread. The subject is Image Resolution of 35mm Cinema Film. My comment is that this image resolution varies widely. It ranges from poorly done flats with lots of slip in the printers(less then 20 lp/mm)to first strike anamorphic trailers done on slow speed printers by a special print crew (75+ lp/mm).

Larry
 

---

Posted by Manny Knowles (Member # 1171) on 02-26-2004, 08:59 AM:
 
The one that stands out most in my mind is the trailer for THE TRUMAN SHOW where there were *quite a few* deleted scenes represented in the trailer.
 

---

Posted by Daryl C. W. O'Shea (Member # 1303) on 02-26-2004, 03:25 PM:
 
*quite a few* -- are you inferring there were actually scenes from the movie in that trailer?
 

---

Posted by Manny Knowles (Member # 1171) on 02-26-2004, 04:30 PM:
 
You've seen the trailer!
 

---

Posted by Wolff King Morrow (Member # 2226) on 03-20-2004, 02:06 AM:
 
What I'd like to know (if someone here can tell me) is what the resolution of a single frame on an average print would be. Leave lens, lighting, and focus out of the equation and lets go with strictly the frame itself. I'd like this figure to be given in pixels if possible too (as in 1280x1024 for example).

I'm also curious what resulotions digital effects studios work in when they do frame-by-frame alterations of scenes. Often times you see them play back a scene they created on a computer during a documentary. They obviously had to scan in each frame of live action, add digital effects, and then finally output the image back onto negative film stock. I'm just wondering what resolution they do this process in.
 

---

Posted by Steve Kraus (Member # 476) on 03-20-2004, 09:09 AM:
 
I'd like this figure to be given in pixels if possible too (as in 1280x1024 for example).

*sigh* You cannot measure nor adequately describe film resolution as a simple number of pixels. Please see the earlier discussions in this thread and others about modulation tranfer function (MTF). Quick...how many bytes does one second of LP phonograph sound represent? It's the same sort of problem. The most you can say is that so many pixels is kinda sorta a similar resolution just as you might say for audio that a certain sampling rate and bits/sample is similar in quality to the LP. But look how 44.1 kHz/16-bit (CD quality) is in some ways better than vinyl yet in other ways not as good.

Most movie special effects and digital intermediates are being done at 2K (2048 pixels horizontal, vertical in proportion to aspect ratio of the on-film image). Many say it takes at least 4K or greater to emmulate 35mm film. IMAGICA has announced a 10K film scanner (the data would be down sampled to a lower number which would be more accurate than a scan done at the lesser rez).
 

---

Posted by Wolff King Morrow (Member # 2226) on 03-20-2004, 03:37 PM:
 
No need to get crabby about it man. I already know 35mm film is not measured in pixels. I was just asking for a rough equivalent. And yes, LP resolution can be approximated too. The sound wave is visually "blurry" looking, while CD waves are visually "blocky". All one need do is appoximate the wave of an LP to what it would look like in digital format. I am one of those who simply does not buy the argument that CD is not as good as Vinyl. Everything I have ever listened to sounded better on CD, and I grew up on vinyl.

Anyway, I figure the 2048 horizontal digital effects resolution you mentioned would have a virtical resolution of 871 for scope and 1107 for flat.
 

---

Posted by Bobby Henderson (Member # 840) on 03-20-2004, 06:00 PM:
 
A full 2K frame, sans room for optical track, would be 2048 X 1536. Many CGI elements are rendered at vertical resolutions that proportionately fit with 2048 being the bigger, horizontal number. And that method makes the 2K standard even worse.

The 4K standard would allow an effects studio to render visuals for 'scope and anamorphically compress to 2K width while still using the whole frame in the same fashion as real 'scope photography.

But with the whole film production industry geared more and more to 480p DVD release and HDTV standards at best, it will be tough for any effects house to have the balls to produce visuals at the better 4K standard. It quadruples the rendering time. With so many effects studios battling for contracts the low bidder and one with the lowest standard often can win.
 

---

Posted by Steve Kraus (Member # 476) on 03-20-2004, 06:53 PM:
 
It's beginning to happen though.
 

---

Posted by Larry Myers (Member # 738) on 03-21-2004, 06:49 PM:
 
Hi

I will take a stab at it. A 35 film frame opening of 24mm by 18mm might have a "pixel" or "line space" if you will range of 700x500 to about 2480x2160 depending on film type and printing methods.

Larry
 

---

Posted by John Pytlak (Member # 331) on 03-21-2004, 08:27 PM:
 
The on-line Kodak motion picture film data contains full MTF and granularity curves for each film. Here is the data for Kodak VISION2 100T Color Negative Film 5212, the sharpest Kodak color negative film on the market today:

Kodak VISION2 100T 5212 data

Remember, MTF data for film is given as a function of cycles per millimeter on the film. Usable visual information is usually obtainable even when the response is well below 20%, thus averaging higher than 100 cycles per mm for this film.

Note that on a color negative film, the top imaging layer (blue) is the sharpest, the middle (green sensitive)layer close to it, and the bottom (red sensitive) layer the least sharp. This is because light scatters somewhat as it goes through the emulsion layers of the film. Kodak's T-Grain and 2-electron sensitization technologies have helped improve sharpness and reduce graininess tremendously in recent years:

Kodak R&D
 

---

Posted by Gordon McLeod (Member # 33) on 03-21-2004, 09:08 PM:
 
Larry is fixated on resolution of the film itself which in the long run is a meaningless number
The number that actually counts is what is measured on the screen in a theatre and that is the comulation of the negative the print stock the printer the processing the projector lens the ambient room light the lamphouse output
Numbers get rather meaningless at this point
The bottom line is that Kodak as other film manufacturers have consistently raised the bar of the raw source of the data
regrettably many cinematographers have chosen to degrade that to suite the artisitic statement of the final film that is all well and good as it is the end product that counts but in raw numberts it doesn't
current 35mm due to its contrast ratio and resolution is still a better medium than electronic means and 65mm/70mm is still better
 

---

Posted by Larry Myers (Member # 738) on 03-22-2004, 04:32 PM:
 
Hi

Yes it is a mix and match problem when you talk total system resolution. I think with Kodak's best films along with using a good $100,000 camera, 100lp/mm or more us possible. Although, as you say, all this is wasted if the projector shutter is not in sync and your using a old lens.

I do believe high quality showprints from Kodak's latest high resolution film stocks projected with the latest ISCO HD scope lenses, should equal or better screen image resolution of the 65/70mm presentations done in the 50's and early 60's.

Larry
 

---