How Light Leaks Ruin Scans
If you routinely use autoexposure, one guaranteed way of making for
wasted effort is to
mount the image carelessly and include a light leak.
As demonstrated in the previous page, autoexposure operates by adjusting the
scanner's exposure until it maximizes the range of intensities that are achieved over the
scanner's sensitivity range. It searches for the brightest area of
the image and increases or decreases exposure until that area is close to the
right edge of the histogram. Increasing exposure has the effect of moving
the right edge of the histogram rightward; decreasing exposure has the effect of
moving the right edge leftward. Though the crop box defines the boundaries
of the image to be produced, NikonScan evaluates the full frame, not just the
crop area. Therefore if the brightest area is outside the crop area, as it
will be with negatives, the user may have to add exposure to obtain a full range
of tones. Normally this is modest or zero. A light leak, however,
confounds the exposure system with an area
so bright that NikonScan thinks the image contains the most extreme dynamic range
possible.
|
Fig. 2a Left, raw scan (without color corrections or
points set) of a carelessly mounted transparency with a light leak.
The light leak is so much brighter than any density in the image that
the autoexposure determines that the dynamic range is much greater than
it actually is. As a result the image is noticeably underexposed
and has low contrast |
|
Fig. 2b RGB histogram. Compared with fig.
1b the light leak results in a constricted range. The red arrow
indicates that the brightest highlights (mainly in the plate) have been
misplaced. The blue arrow is the position of the brightest
highlights in fig. 1b. The light leak is represented by the spikes
(for the RGB channels)
at the extreme right |
Effect of Light Leaks on Negatives
(Low-Contrast Films)
Here is a negative scanned as a positive that has been properly masked:
|
Fig. 3a Left, raw scan of a negative |
|
Fig. 3b RGB histogram for main subject of fig. 3. |
|
Fig. 3c Histogram for main subject of fig. 3
(yellow crop box). Range of intensities is typical for negative
film -- less than transparency film. |
|
Fig. 3d Blue channel histogram
(yellow crop box) |
The effect of a light leak on negatives is more serious than with
transparencies:
|
Fig. 4a Left, raw scan of negative with light
leak not masked |
|
Fig. 4b Histogram for raw scan of negative with
leak, fig. 4a |
|
Fig. 4c Histogram for main subject of fig. 4a
(yellow crop box) |
|
Fig. 4d Note the impoverished range of the blue
channel as compared to fig. 3d |
An obvious question is why I didn't use NikonScan's color negative mode to
demonstrate the effect of a light leak. The reason is that much of the
negative mode's underlying functionality, which is shown deconstructed here, is hidden
from the user. By performing an 8+ bit scan, NikonScan is able to return a
histogram with a full-range of tones, the end-product that the user sees.
Return to LS-8000/9000 Notes
|
|