Chuck Derry raised the issue in a query to Screen-L, the film/TV-studies discussion list. Responses from Leo Enticknap, Dave Trautman, Don Larsson, and myself provided some interesting answers...
Date: Wed, 8 Oct 1997 15:08:25 +0100
From: Jeremy Butler <jeremy@tcf.ua.edu>
Subject: Re: Reply: Film/Video Stills
On Thu, 2 Oct 1997 12:42:28 -0500 Charles Derry said:
>Re: The continuing discussion of film stills.
>
>What about the use of frame enlargements, or, more pertinent to what
I am
>now working on, the use of video frames from a television series (in
order
>to do extensive visual analysis)? Incidentally, although I have all the
>shows I'm writing about taped, has anyone already "invented the wheel"
>and found the best way to extract specific images from a videotaped show
>for subsequent publication as stills?
Hi Chuck,
This is something I've been grappling with since beginning work in 1990 on a television criticism textbook (shameless plug: TELEVISION: CRITICAL METHODS AND APPLICATIONS, Wadsworth) and I'm happy to report that there are now many computer-based options for nabbing video images.
'Course, there's still the old-fashioned method of setting a camera's shutter speed below 1/30th of a sec and photographing the screen, but I've found that that often results in scanline trouble.
But, if you decide to use a computer, then what you want to do is a "video capture." The images you nab can then be displayed on your computer and tweaked to look just right. Later, you send the image files directly to the printer of the book/journal article--just as you do the word processed files. The resulting images look just as good as, say, frame enlargements from 16/35mm film (cf. FILM ART's images).
When TV:CM&A was printed in late 1993, it included dozens of captured images. This technology was still in a semi-primitive state at the time, however, and I'm not entirely pleased with the results.
Today, even with a fairly rudimentary computer, you can get much better results. You can grab frames with:
1. a high-end video capture board--or even a non-linear video editing system (e.g., Avid, Media 100, Adobe Premiere)--to grab the frames, or
2. Snappy, an inexpensive (less that $200) device that plugs into the parallel port of your Windows-based computer.
I've become quite a fan of Snappy. It's cheap and easy to install and you just plug the NTSC signal directly into it using a regular RCA connector. Thus, it'll accept output from a VCR or a camcorder. The only drawback I've found is that it operates on a 9v *battery* (which, of course, has to be replaced periodically) instead of using a transformer, but I reckon a trip to Radio Shack could solve that.
In fact, I was just using it yesterday to capture some images from RULES OF THE GAME for a class of mine on Bazinian realism. Check it out online at:
http://www.tcf.ua.edu/classes/Jbutler/T340/Bazin02.htm
Even with the scuzzy nth generation print on video that I was using, the images turned out pretty decent, I think. (The BEST YEARS OF OUR LIVES still on that site is *not* a frame grab.)
Thus far, this is the best option I've come across. If anyone knows of a simpler, more effective, and/or cheaper system, I'd sure like to hear about it.
Take care,
----
Jeremy Butler
Associate Professor
jeremy@tcf.ua.edu
ScreenSite http://www.sa.ua.edu/ScreenSite
Telecommunication & Film/University of Alabama/Tuscaloosa
From: Leo Enticknap <L.D.G.Enticknap@EXETER.AC.UK>
Subject: Reply: Film/Video Stills
On Fri, 3 Oct 1997 10:10:47 -0500 Jeremy Butler <jeremy@TCF.UA.EDU> wrote:
> But, if you decide to use a computer, then what you want to do is a
"video
> capture." The images you nab can then be displayed on your computer
and
> tweaked to look just right. Later, you send the image files directly
to
> the printer of the book/journal article--just as you do the word
processed
> files. The resulting images look just as good as, say, frame
enlargements
> from 16/35mm film (cf. FILM ART's images).
I once tried using a laserdisc player linked via an S-video wire (and then I swapped it for a Euroconnector, to little effect) to a video card (origin unknown: the computer was one from the University IT department) and Microsoft Video for Windows in order to get stills off a PAL CAV disc. Even after trying to enhance the frames with Paint Shop Pro, the result was still unacceptable compared with a 10 x 8 reversal printed from a 35mm positive release print and scanned at 300dpi. Scanning lines fragmented the picture, the colour bled all over the place, contrast and definition was lousy. If the plates in Bordwell & Thompson's "Film Art" (I have the 4th edition) came from domestic videotapes then clearly I was doing something very wrong.
Leo
__________________________________
Leo Enticknap
Postgraduate Common Room
School of English and American Studies
University of Exeter
Queen's Building, The Queen's Drive
Exeter
Devon EX4 4QJ
United Kingdom
email: l.d.g.enticknap@exeter.ac.uk
Date: Tue, 7 Oct 1997 09:17:02 -0500
From: Jeremy Butler <jeremy@TCF.UA.EDU>
Subject: Re: Reply: Film/Video Stills
On Fri, 3 Oct 1997 21:24:44 +0100 Leo Enticknap L.D.G.Enticknap@EXETER.AC.UK> said:
>On Fri, 3 Oct 1997 10:10:47 -0500 Jeremy Butler
<jeremy@TCF.UA.EDU> wrote:
>
>> But, if you decide to use a computer, then what you want to do is
a "video
>> capture." The images you nab can then be displayed on your computer
and
>> tweaked to look just right. Later, you send the image files directly
to
>> the printer of the book/journal article--just as you do the word
processed
>> files. The resulting images look just as good as, say, frame
enlargements
>> from 16/35mm film (cf. FILM ART's images).
>
>I once tried using a laserdisc player linked via an S-video wire
(and
>then I swapped it for a Euroconnector, to little effect) to a video
card
>(origin unknown: the computer was one from the University IT department)
and
>Microsoft Video for Windows in order to get stills off a PAL CAV disc.
Even
>after trying to enhance the frames with Paint Shop Pro, the result was
still
>unacceptable compared with a 10 x 8 reversal printed from a 35mm
positive
>release print and scanned at 300dpi. Scanning lines fragmented the
picture,
>the colour bled all over the place, contrast and definition was lousy.
Yeah, you're absolutely right, Leo. Images captured from video (to digital) are unacceptable compared to those taken from a 35mm film print.
What I should have said is that *digitally* captured images from *video-originating* material are superior to analog photographs-of-the-monitor images of the same. What Chuck Derry and I have been trying to figure out is a good way to get video images (not film images) into print.
My experience has been that digitally captured images more closely approach the quality of film-based frame enlargements (e.g., those in FILM ART) than photographs of the monitor do. But I'm certainly open to suggestions for better methods of nabbing material from video.
>If the
>plates in Bordwell & Thompson's "Film Art" (I have the 4th edition)
came from
>domestic videotapes then clearly I was doing something very wrong.
Interestingly, the FILM ART Instructor's Manual (at least for the 5th edition) details the method Kristin Thompson uses to "Make Slides from Film and Video Images." She uses a Canon Duplicator 35 and 35mm prints for most of the stills.
As she explains, the Duplicator is a macro lens with a device for holding a 35mm film strip close to the camera body. This allows you to photograph a single frame and get strikingly good results--as FILM ART illustrates.
However, I don't think that the few video-based images in FILM ART look as good as they could. E.g., the frames from the video version of ADVISE AND CONSENT (fig. 6.69-70 in the 4th edition; 7.68-69 in the 5th) are basically okay, but you can see a diagonal scanning aberration in 6.70/7.69. In the Instructor's Manual, she explains that one may nab video frames by photographing a monitor. I suspect that she could get better results by digitally capturing the image.
Regards,
----
Jeremy Butler
Associate Professor
jeremy@tcf.ua.edu
ScreenSite http://www.sa.ua.edu/ScreenSite
Telecommunication & Film/University of Alabama/Tuscaloosa
Date: Wed, 8 Oct 1997 15:08:25 +0100
From: Leo Enticknap <L.D.G.Enticknap@EXETER.AC.UK>
Subject: Re: Reply: Film/Video Stills
On Tue, 7 Oct 1997 09:17:02 -0500 Jeremy Butler <jeremy@TCF.UA.EDU> wrote:
> My experience has been that digitally captured images more closely approach
> the quality of film-based frame enlargements (e.g., those in FILM ART)
than
> photographs of the monitor do. But I'm certainly open to suggestions
for
> better methods of nabbing material from video.
I would have thought that the type of monitor and camera being used in a photographic transfer or the computer hardware in a video transfer are all variables. From the few telerecordings (pre-video TV recordings made by pointing a (usually) 16mm camera at a high-definition monitor) I've seen, the quality varies significantly. Getting back to stills, I wonder what sort of result you'd get by using a 35mm stills camera on a video image produced by an LCD projector?
> Interestingly, the FILM ART Instructor's Manual (at least for the 5th
> edition) details the method Kristin Thompson uses to "Make Slides from
Film
> and Video Images." She uses a Canon Duplicator 35 and 35mm prints for
most
> of the stills.
What I've usually done is to place the 35mm cinema film in a stills enlarger and make a Cibachrome paper print (which I then scanned). It has sometimes taken me two or three test strips to get the exposure and gamma levels right, but the results are usually acceptable. I've never tried a slide duplicator attachment on an SLR camera, though. I'll have to get the new edition of B & T...
Best wishes
Leo
__________________________________
Leo Enticknap
Postgraduate Common Room
School of English and American Studies
University of Exeter
Queen's Building, The Queen's Drive
Exeter
Devon EX4 4QJ
United Kingdom
email: l.d.g.enticknap@exeter.ac.uk
Date: Wed, 8 Oct 1997 09:59:51 -0600
From: Dave Trautman <dave.trautman@UALBERTA.CA>
Subject: Taking video stills to print
Jeremy Butler wrote,
>What I should have said is that *digitally* captured images from
>*video-originating* material are superior to
analog
>photographs-of-the-monitor images of the same. What Chuck Derry and I
have
>been trying to figure out is a good way to get video images (not film
>images) into print.
Ah, finally something I know something about.
I was going to come into the thread and castigate Jerry for misleading the group about video images but he has redeemed himself through his clarification. However, I think the list would benefit from a little tech talk about this in case you're preparing to take the plunge into the muddy waters of video capture for printing.
Let's start with video itself. I often describe video in my seminars as looking at life through a screen door. It's fuzzy. Really clean, original video is fuzzy. Even NHK's 1125 HDTV is fuzzy (in a manner of speaking) closer than 3 feet. If you were to have a TV screen in a book in your lap you'd be able to see the lines and the fuzzy edges. We normally watch TV screens from about 6 feet away and the specs for NTSC, with respect to clarity, are interpreted for a minimum of 6 feet viewing distance.
The second barrier to scramble over is the player machinery. A paused video image is not the whole signal. An image running at play speed has about twice the signal information as a paused image. A captured frame ought to be running at full speed when it arrives in the computer then it can be selected from the "stream" as a single interval.
The third barrier to haul yourself through is the quality of the capture encoding hardware. The highest quality capture systems cost plenty. They capture excellent images and convert to digital without adding any artifacts to the signal. Capture systems for non-linear computer editing are top notch and top priced. They capture each frame individually, which is necessary for editing as you can well imagine. Some capture "streams", are not frame-based, and need to have more than a single frame's information to reproduce a still. But that's another essay.
The next obstacle on the course is the compression of the captured file. If you have the real estate on your hard drive (or server system) you can capture the image stream without any compression and save it to a disk directly. We're talking big big big files here. I have a one gigabyte file which represents a 720 pixel by 480 pixel image size capture of a Betacam original tape segment lasting only about 5 minutes. It's beautiful and clear, but it's trapped on my AV speed drive because I can barely squeeze it onto a JAZ removable cartridge.
Most software which operates capture systems offers you a pallette of compression schemes. If you compress the file when saving, you're going to lose image. Fuzzy originals become worse when compressed. Getting that one frame from the stream (the uncompressed one you make) means having software which will "clip" that frame and make an image file from it. Some software hasn't a clue what you need. Other stuff is helpful or will handle plug in software to help with the transitions. However, most plug in technology is developed to get graphic files into video and not video files into graphic formats.
Long before you play the tape into the computer you should have had the kind of meaningful conversation with your graphics professional you would normally have with your spouse concerning family planning. This intimate exchange should include their preferred file type, their need for "screen" parameters, and a discussion of the final use of the image in the printed document. They may ask that you capture as big and deep a picture as you can afford so that they later can reduce it, enhance it, and then compress it.
It's only one frame after all. How big can it be?
The final hurdle before the finish line is the graphic output. My graphics professional has found a process in which he "shrinks the image" by some magic percentage and increases the "perceived" clarity of the image, then he weaves his enhancement magic on the file and compensates for much of the contrast, colour edges, and in-frame motion artifacts using all those "curves" and spikey image charts. He shrinks the picture in order to create a more dense set of pixels. This improves the apparent sharpness of the image.
He handles all the friction between himself and printer without my help. They speak that foreign tongue of screens, dots, ink bleed, and paper quality which I never learned in school.
If anyone wants details on the hardware a person normally needs for this (or maybe just the stuff I use) they can mail me for it. Have a cheque book handy because these things don't come cheap. If you have an inexpensive, consumer level capture system you won't be happy with the results of video captures becoming graphic files. You can take that to the bank.
I haven't found a store-front service in my area which either understands enough of the process or has the hardware or software to do an adequate job. These are rare skills.
Sincerely,
Dave Trautman
Media Specialist for ATL
University of Alberta
Edmonton, Alberta, Canada
Date: Wed, 8 Oct 1997 15:00:12 -0400
From: David P Kintsfather Jr <kintsfat@KUTZTOWN.EDU>
Subject: Re: Taking video stills to print
I use the "Snappy" video capture device and working with SVHS tapes I have gotten amazing clarity. The Snappy defaults to a 300 dpi capture at a variety of different image sizes that correspond to different monitor resolutions.
Images can be edited in Photoshop or similar software, and the unsharp mask filter can go a long way toward compensating for the lack of resolution in video. I print images on an Epson Stylus Photo at 720 dpi, and have found that a 240 dpi file prints best, so I can increase the image size a little over the Snappy default. I get images that come close to a photographic print from a 35mm negative, all from a $200 toy!! Pretty neat! A 900K file yields a wallet sized print, and 5 megs (NOT gigabytes) is sufficient for an 8x10 image (I may be exagerating slightly, but a 5 meg image is very large).
With plain VHS the image is a lot softer and looks a little posterized when you crank in sharpening, but it still is usable if you need a still image and the only source is VHS.
As a postscript, I have no financial interest in Snappy, and I'm sure some of the other video capture boards will work as well. I've just been amazed at how well this one has done.
David
------------------------------------------------------------------------------
David Kintsfather | PHONE: (610) 683-4497 Dept. of Telecommunications | FAX: (610) 683-4659 Kutztown University | INTERNET: kintsfat@kutztown.edu PO Box 730 | "We need not think alike to love alike." Kutztown, PA 19530-0730 | Francis David ------------------------------------------------------------------------------
Comments: Jeremy Butler,
jeremy@tcf.ua.edu
Last modified: October 10, 1997