Testing the Limits
Experiments in emulsion manipulation
By Mark Woods
Some of the various film and modified lab processes cropping up on current movie and music video productions have intrigued me-specifically, skip-bleach with color negative film stock, and reversal film stock cross-processed in a negative film developer. A recent trip to Rochester, New York as an Eastman Kodak guest, along with directors of photography like Newton Thomas Sigel, ASC who shot Three Kings and X-Men, piqued my interest even more. During casual conversations on cinematography, Sigel offered interesting insights on the cross processing of reversal film stock. On video shoots, he would expose 100-foot bulk loads of still film and have it cross-processed in a negative film developer. While in skip bleach processing, the negative is developed normally except that it forgoes the final bleach bath to remove the processed silver. This silver-effectively a mask-adds density to the negative, resulting in increased speed as it desaturates colors. Contemplating a skip-bleach test of my own, I decided to test both procedures and shoot side-by-side images of what could be a commercial.
Applying the "LAD System of Testing" to Unusual Processes
In the reversal cross-processed to a negative, Red, Green and Blue emulsions do not express anything close to a "normal" tone. However, I believe my LAD System of Testing is the most accurate means of evaluating a film's emulsion for speed and color temperature balance, and of providing a quicker determinant of accurate printer lights. As FotoKem's Stacy Gill worked on zeroing out the printer lights, he mentioned how much easier it is to accurately make a truly neutral print using LAD Patches. Cross-process reversal yields a skewed color palette to a print, and a neutral gray card-to determine the printer lights-won't guarantee neutral colors in the tonalities above and below 18 percent neutral gray. FotoKem and other labs always use density aims to make an IP and, on all other steps, to control against any subsequent shift in color or density in the duplication or printing processes. (See "A Simplified Motion-Picture Laboratory Control Method for Improved Color Duplication," in the SMPTE Journal, Oct. '76 for a comprehensive discussion of Lab Aim Densities.)
The basis of the "LAD System of Testing" is to shoot a correctly lighted gray card (including color temperature) and read the resulting densities with a densitometer. When the density approaches the Lab Aim Density, note the EI (exposure index) as well as the light's color temperature: this indicates the optimum exposure with which to achieve the negative's optimum density in its designed manner. To determine printer lights, print the gray card exposure with the LAD densities, and read the print densities. When print density approaches the print LAD, the printer lights used to get that density are then applied to later shots, and become the suggested light on a production. As a side note, for artistic or other reasons, the "optimum density" may not be the density chosen by a director of photography - it's only a starting point. (For a complete discussion of the test, see "L.A.D. Testing Procedure" in ICG Magazine, May '96. The article can be downloaded from the technical area of the this web site (http://www.cameraguild.com/technology/lad_testing.htm).
Films and Processes
Though only Kodak emulsions were used, the same methodology can be applied to Fuji stocks. To determine all of the exposure indexes, each stock was exposed for a normal and off-process LAD Test. Once each EI had been determined, then each LAD gray card was printed (those cards are reproduced here from the LAD workprint). The normal batch became a "control" to accurately determine the density shift and the resulting "look." On both the cross-processed 5285 reversals to yield a negative, I averaged the density to approach what might be an LAD negative's average density. I also placed the EI as close as possible to the center of the gamma curve.
|Film Stock ||Normal Process ||Off Process |
|5285 VS 100D Reversal ||E-6 Reversal ||ECN2, C-41 |
|5277 Vision 320 T Negative ||ECN2 Color Negative ||Skip Bleach |
|5298 EXR 500 T Negative ||ECN2 Color Negative ||Skip Bleach |
The "LAD Patches"
Although the gray cards all rank equally on the densitometer, discrepancies do surface on the card, particularly with the 5285 processed in C-41 and ECN2: when shooting the gray card, a .1 stop variance had been let go. Usually, it doesn't matter but it clearly factored into these cross-processed images, because of the stock's new gamma. For example, Tom Sigel told me that he placed shiny boards and huge fixtures on his subjects to "flatten" the light-not exactly an accepted tactic for a "normal" production. Yet his images on Three Kings turned out quite wonderful. As can be seen from these tests, venturing into this area requires that one think "outside the box." The other gray cards come pretty close, but most interesting is what occurs on either side of the middle 18% gray. On the density data, the "normal" negative Dmin and Dmax is shaded. Although in this process density is still sometimes being added over 4.00, the numbers quickly go beyond what a film printer can normally handle. For the optimum printer range, the densities were placed as close to the center as possible. When +0 +13+ 0 Trim on Printer appears, add the trim amount to the appropriate printer light-in this case the Green light. For example, a 27-39-12 printer light with +0 +13+0 trim would come out to a 27-52-12 printer light-this zeroed-out the light as best as possible for this system. Since all the LAD Patches had been assembled on a 16-foot piece of film, a decision was made to do an overall correction on all the film tests so that no problems would arise in the mechanical issues of timing.
5285 VS 100D Reversal: Normal
LAD for 5285 VS 100D Reversal = 1.02-1.00-.87
LAD Patch EI 100 = .64-.65-.57
LAD Patch EI 200 = 1.06-1.09-1.00
Test Exposed at EI 125
Intermediate from Master Positive Printed at: 30-30-30
Each Scene Timed From InterNeg 5272
The Kodak 5285 VS 100D is a still film slit and perforated for motion picture use. The E-6 procedure is the normal reversal process that yields a brilliant reversal image. The C-41 developer is utilized in still photography to process color negative. ECN2 is the developing process designed for motion picture color negative film. In this test, we shot our normal test (see the sensitometric curve titled "L.A.D. 5285 100D Reversal Film Normal E-6"). The actual EI appears to be about 160. (This test did not include the normal one-third stop increment exposure adjustment.) Rated at EI 125-halfway between Kodak and the test results-the reversal image is spectacular. What's been reproduced is a print of a dupe negative made so that all the results could be projected-i.e. a film finish. Of course, in a telecine, one would use the positive camera original. Examine the shadow depth and how the highlights retain detail, as well as the wood around the gray card to see the different textures and the yellow's response. This stock's useable range is much less than on negative. Also, since it's reversal, most directors of photography would expose it about one-third stop higher than its rated EI for more density-just the opposite concept of negative where directors of photography generally decrease the EI by one-third to add more density. Again, there is no substitute for doing one's own tests.
5285 VS 100D Reversal: C-41
Cross Processed With C-41 Color Negative Developer
LAD Patch EI 400 = .99-1.75-1.36
Printer Lights: 27 39 12 (+0+13+ 0 Trim on Printer)
Each Scene Timed
Upon first approaching Steve Willard at Spectra American Color Lab, he mentioned that the C-41 developer is very "vigorous." When the 5285 (and the Fuji 8540 Velvia) is processed in this manner, the image emerges as very contrasty with huge color shifts. Because of the contrast, it has been mentioned to appear as a "color" Kodalith. Though unsure of the new EI, Willard remarked that cinematographers generally rated it faster by one-third to two-thirds of a stop. (Since the 5285 is rated 100 daylight, the rating would be EI 125 or EI 160.) Our tests indicate almost a two to three stop speed increase. With more testing related to a specific project, I would probably use EI 500 or EI 650: thus placing the middle of the exposure in the center of about a two-stop range. The gamma on the sensitometric curves is very steep, indicating very high contrast. Also, the Blue emulsion is in the center with the Green emulsion on top (i.e. most density) until merging close to Dmax. Color choice, overall set contrast and the desired look would be a major consideration in deciding to shoot 5285 cross-processed in the C-41 developer. On the color chart, the blue chip approaches Dmin at EI 800, and appears on the color chart as virtually black. The blue chip can actually be printed as indigo, but other trade-offs arise in terms of color shifts. While the blue has almost turned black (Dmin), the yellow is almost transparent (as it's approaching Dmax). The cyan has shifted blue while the green, magenta and reds appear very saturated. Such information can be used to help the lab and production understand the sought after look. It does appear bizarre, but it's quite wonderful-as seen in the clips. By the way, I read densities up to 4.00 (Status M) on the densitometer, but these densities go beyond most lab printers' practical range; this density also runs past the range of many of the older telecine machines. Finally, if a print is made at the LAD Patch printer light, the mid-tones will shift red. (This happened in early prints of "Woman with Orange Stained Glass." Timing out the red in her face made the orange in the glass approach Dmax, becoming white, as it appears here.) A timer must be involved because the Red, Green and Blue emulsions aren't the same speed and don't track parallel. I am considering a follow-up test that uses a magenta and blue combination filter pack. Mark Van Horn, at FotoKem, suggested using a CC 30 magenta filter to hold back some of the green emulsion's density. To increase the Blue emulsion's density, I would add, say, an 80B Blue Filter. Such a correction would allow the resulting negative to be printed in a lab's normal printer range, and also facilitate a telecine transfer.
5285 VS 100D Reversal: ECN2
Processed ECN2 Color Negative Developer
LAD Patch EI 200 = .94-1.79-1.29
Printer Lights: 27-43-12 (+0 +13+0 Trim on Printer)
Each Scene Timed
Partly due to the wide availability of the ECN2 negative developer process, this stock and developer will probably be the most popular cross-process. With this particular method, the yellow actually photographs closer to normal, the cyan still shifts blue, and the blue still approaches black-yet some blue/purple detail does remain in the color chart. With skillful art direction, a paler blue would reproduce much darker. Reexamine the characteristic curves and note that at just below "normal" exposure, the blue disappears into Dmin, which, for print, is black. Some of the blue might be brought out on a telecine, particularly one with a Power Windows type application. The yellow is still less dense than on a normal control exposure. Check out the grain definition of the wood around the chart: this shift is interesting on all the images. Useable contrast depends on the emulsion being looked at: the Red emulsion has eight stops of range; the Green emulsion has three stops of range; and the Blue emulsion holds a two-stop range. The characteristic curves clearly indicate what is visible on the color charts. In studying the black-and-white chip chart, one may notice whites shifting to cyan. When the Red, Green and Blue emulsions aren't parallel, the gray chips indicate the color shifts indicated on the chart from toe to head. In this process, the gray could be timed neutral at any given gray density, but that color timing would shift the colors with greater, and less, density. Also, where color tonality exists-like in skintones -the shift can be more radical. In "Woman with Orange Stained Glass," the skintone changes with the process, from rich and full-toned (5277 Normal) to loss of detail as the density approaches Dmax (5285 X Processed C-41). Filmstocks and lab off-processing truly expand opportunities for personal expression with images.
Vision 5277 (320T): Normal and Skip-Bleach
Normal LAD Patch EI 320 = .87-1.34-1.66
Normal Printer Lights: 31-24-28 (+0+13+0 Trim on Printer)
Skip Bleach LAD Patch EI 640 = 1.00-1.52-1.82
Skip Bleach Printer Lights: 34-30-33 (+0+13+0 Trim on Printer)
Characteristic curves for 5277 should be familiar to most-it's tested to be accurate at its EI rating of 320 tungsten. When skip-bleach processed, the film's speed effectively doubles to EI 640. I suspect that an exposure of EI 800 would appear fine, perhaps with a little more grain. (Again, the normal one-third stop increment exposures was not undertaken for a more accurate appraisal.) At two stops over EI 640, the emulsion has effectively reached Dmax. Though some latitude persists to print some highlight detail, it's not much. The three emulsions track parallel, resulting in an image that doesn't shift color in shadows or highlights. In comparing similar densities of the two processes and T-stop being used, one aspect should be noted in particular: at N-4, for example, the shadows become texture without much detail. Observe "normal" density (.32-.73-1.01) and check for a similar density in skip-bleach. With the skip-bleach process, the same textured shadow would occur at about N-3. Highlight detail can be evaluated with the same approach. In "normal" density, N+3 is 1.29-1.85-2.19 while a similar density with the skip-bleach process is about N+2/3. Clearly, the process impacts highlights much more than the shadows. Since this test is empirical, and quantifiable, it allows one to effectively set up a test geared for a final determination of lighting ratios, colors to be used, make-up and so forth.
5298 (EXR 500) Normal and Skip-Bleach
Normal LAD Patch EI 500 = .96-1.47-1.77
Normal Printer Lights: 31-24-28 (+0+13+0 Trim on Printer)
Skip Bleach LAD Patch EI 2000 = .87-1.33-1.64
Skip Bleach Printer Lights: 34-30-33 (+0+13+0 Trim on Printer)
In this test, the 5277 exposure index is increased by only one stop. The 5298 rating is altered by two stops for an effective speed of EI 2000. All the 5298 emulsions are slightly below the LADs, but this EI was chosen for several reasons. (An EI of 800 would probably put the density fairly close to the LADs.) The test was shot with the higher EI to ascertain the difference between the two stocks-the 5277 with about one-third stop added density and 5298 with about one-third stop less density than the LADs. The images turned out quite pretty, and though added grain had been expected, it wasn't objectionable and only enhanced the look. The only concern was whether the blacks would gray out or become muddy. Fortunately, such was not the case, although on an early timing, the scene was printed up and the blacks did become muddy. The selected EI 2000 is a good indication of how one meters, or uses the EI, to achieve a look. Where one places the center of the scale impacts the stock's dynamic range and the presence of grain. 5298 has slightly higher contrast than the 5277, yet under skip-bleach processing both have a total of about seven stops of useable contrast range. When the 5298 is exposed at EI 2000, there are two stops of latitude over N, but only four stops below N. If rating the stock at EI 2000 for a production, I would make sure that the face's keylight always approached N+1, and that no important shadow detail fell much less than N-2.
Woman on a Bicycle
This scene's exposure makes it read a little "hot"-I wanted an ethereal feeling, and was curious how the various processes would handle it. Even after printing 5285 onto an internegative 5285, the reversal qualities remain: highlights lost detail fairly quickly and printing "down" the image only made the highlights less bright with lost detail. The image quality is softer than expected but beautiful, nonetheless. This example presents the wide range of representation possible with just these few off-processes. The 5285 X Processed C-41 was totally blown out at the head of the shot, but as the woman rode on, she became visible and resolved into the closer shot printed here. The cross-processed negatives were timed to skintones, and color negatives printed at the processes' LAD Patch.
Woman by Tombstone
While perusing the images, please evaluate the emulsion/process gamma curve. Understanding the gamma (the characteristic curve's slope) will aid predictions of exposure, color and image. Observe the velvet jacket: in the normal negative, detail appears that is lost in the other processes. Also, note that the loss of overall color shifts from the warmth in the normal negative. The 5285 (printed from the IN) has less color saturation because of contrast in the textures. Also, the tombstone acts like a gray card behind the woman's head. In the different processes, the tombstone becomes progressively darker as a visual indication of the steeper gamma.
Woman with Orange Stained Glass
As the face modulates from N+1.5 to N-2.5, the difference in the stocks and processes becomes more pronounced. Beyond the contrast, major color shifts occur in the orange window. The orange tone appears somewhat natural in the normal 5277, the skip-bleach 5277 and the 5285. But in the 5285 cross-processed C-41, the orange approaches Dmax (white) as red hues were timed out of the girl's face. The darker skintones in the two 5285 cross-processed images probably result from the increased gamma- i.e. the shadows have much greater contrast and approach Dmin more quickly. This sample shows that one would need to flat light the subject to hold enough information in the two to three stop useable contrast range-unless one was purposely after such a look.
Woman in Front of Picture Window
Here, the light spills fairly even over the subject and background, and all exposures have a relatively full range. In the images of 5285 cross-processed, the woman's face becomes more of a graphic image with her dress approaching more two-dimensionality, much like a Fauvist painting. Shifts in yellows are particularly apparent in the yellow stained glass.
Given more time, this experiment would have featured a filter pack for the 5285 cross-processed, including magenta and blue filters; this arrangement would lower the green printer light, increase the blue density, and also remove the need for the +0+13+0 trim light on the printer. I must confess that with all the electronic possibilities, the photochemical process still offers "organic" possibilities that cannot be duplicated, and which help create our artform's "happy accidents." Spectra American's Steve Willard stated that the lab could process E-6 in percentages plus or minus stops-that being push or pull-and has done so for a number of last season's TV programs. Also, although the graphs appear "official," they are suited for the labs that collaborated in this survey. One hopes that all the information is useable at respective labs, but an empirical test should be done for fine-tuning. If the lab is working to Kodak standards, the test results should closely approximate mine. Heartfelt thanks go out to Stacy Gill and the FotoKem team who assisted with "Testing the Limits" along with Steve Willard at Spectra American who surpassed the call of duty to correctly carry out all film processes. It is sheer joy to work with people who, though toiling in the trenches every day, still retain their passion to lend a hand to the film community at large.
Director of photography: Mark Woods
Camera assistants: Michael Bratkowski and Paul Wojiack
Director: Jeff McQueen
Actress: Drea Hoffman Creech
5285 100D Reversal Stock
West Coast Contact:
East Coast Contact:
E-6 & C-41 Processing
11037 Penrose St.
Sun Valley, CA 91352
ECN2 & Skip Bleach
Mark Van Horn & Stacy Gill
2801 W. Alameda Ave.
Burbank, CA 91505
12950 Rayner St.
North Hollywood, CA 91605