Updated March 2010
The following notes apply to both scanning and camera profiling. More detailed camera profiling instructions are provided in the latest (November 2008) HCT instructions.
Capturing the target
CCD scanner evenness
Cleanup in Photoshop
Default contrast and density
Lightening 'dark' 35mm profiles
Recommended RGB workflow
Scan the target without USM and save it as an RGB TIFF image, with at least 750 pixels in the long dimension. Larger files will not increase quality. For advice on setting up your scanner for RGB work download the free Scanning_Guide. Pay particular attention to disabling any automatic adjustments, setting fixed end points and creating an optimum tone curve.
Ideally, every patch in the target scan should contain slightly different values from its neighbor patches and no patch should have RGB values as high as 255 or as low as 0. Failure to observe these rules may lead to clipping or reversals in the profile.
Flatbed scanners, especially some 35mm slide scanners, often suffer from uneven illumination across the image area. If the HCT (or any scanner target) is scanned with uneven illumination it may produce a flawed profile. To check for evenness, scan a flat tint of evenly exposed slide film with about a 0.30 to 0.60 density and check it for evenness in Photoshop by examining RGB values across the entire area. Variations greater than +/- 002 in either R, G or B across the frame (ignoring dust or scratches) may be cause for alarm.
Most flat bed scanners - especially 35mm slide scanners - suffer from enough optical flare to negatively affect the quality of scanner profiles. The problem occurs when one or more patches are affected by light scattered across from neighboring patches, or from white areas surrounding the target. Profile errors caused by flare may include plugged shadow tones, incorrect gray balance and inaccurate colors, which can negate the whole point of scanner profiling.
The best cure for flare is to clean the optical system. If this is not viable or successful, at least be sure to use the largest target possible. For example never use a 35mm target when a 4x5 target is available. Flare may sometimes be reduced by scanning the target two or four times, rotated in 180 or 90 degree increments, then registering the separate scans as Photoshop layers set to the 'Darken' mode and flattening the assembly.
With great care and patience, flare may be partially canceled - at least for profiling purposes - by applying low amplitude, large radius USM to a duplicate Photoshop layer set to the 'Darken' mode, but this is highly experimental.
If the most obvious problem with your profile is plugged shadow tones, experiment with different degrees of shadow extension - see Extended range profiles. If your profile seems to make everything too dark, see Lightening 'dark' 35mm profiles.
In Photoshop, rotate the scanned HCT until the patch rows are exactly horizontal and the text block is in the lower left corner, with text upright and reading correctly.
Check the target scan for scratches or blemishes and CAREFULLY clean any imperfect patches with the clone brush (rubber stamp). Be careful not to copy from a neighboring patch onto the patch you are cleaning! Dont worry about blemishes outside the central 50% of each patch.
Due to film manufacturing and development variations, some real-world images may have even higher maximum dye densities than the target, and may produce lower RGB values that would not be 'recognized' by the profile. This could lead to loss of extremely dark detail, or 'plugging', in very underexposed originals.
An easy way to avoid this problem is to add an 'infinite density' patch to the target and extend the range of the target scan before making the profile, as explained in the User Guide.
Note that in scanners with high flare factors, adding a small infinite density patch to the target may not be enough, as flare can affect the infinite density patch. The only reliable way to determine the 'infinite density' response of a 35mm scanner is to make a separate scan of a completely black piece of unexposed slide film to which an infinite density patch has been added.
Most high-end color management users scan with fixed end densities set to approximately 0.10 (white) and 4.00 (black), then set highlight and shadow later in Photoshop. This workflow assures the most accurate reproduction from a wide range of originals, but requires some post-scan image adjustment on almost all originals, which is expected by professional users. Under this workflow images may look a little darker or flatter with an HCT profile than with an IT8 profile, because the HCT profile covers a wider range of densities than are present in most originals.
If an original's highlight density is darker than the target's white patch, the scan will look darker when the profile is first assigned, and if the original's shadow is lighter than the target's blackest patch, the scan will look lighter (less black) in shadows. This characteristic is obviously a GOOD thing whenever extremely light or dark originals come along, but on typical transparencies an HCT profile may produce slightly less contrast and require slightly stronger highlight and shadow adjustments in Photoshop than an IT8 profile. This is NORMAL and should not affect productivity or quality.
35mm scanner profiles often produce results that seem too dark, even when great care has been taken in scanning and creating the profile. This is often seen as a failure of either the target or the profiling software, but the explanation is usually much simpler.
Optical flare can be a significant factor, especially in 35mm scanners, because if the darkest patch in the target produces lighter-than-normal RGB values due to flare, the profile software will associate those falsely light RGB values with the true 'darkness' or L* value of what is really a very dark patch. This means the profile will add density to images in which flare was less active or nonexistent.
An 'extended range' profile will often solve this problem.
Another explanation for the apparent 'darkness' of 35mm profiles is that 35mm transparencies submitted for reproduction are usually selected under nonstandard viewing conditions that make them appear lighter than if they were viewed on a normal light box. For example 35mm slides are usually viewed under a magnifying loupe or on a slide projector which isolates the image from the viewing light source. Unless the eye can simultaneously see some of the illuminant around the image (e.g. part of the light box), it has no way of judging relative density, so even quite dark images - up to one whole f-stop underexposed - can look pleasing on a projector or under a loupe, when it would have looked too dark on a viewing table. The same effect happens when 4x5 or even 8x10 inch transparencies are viewed inside a black mask.
For these reasons, a scanner profile that works well for 4x5 or 8x10 inch transparencies viewed on a light box (with some surrounding area of white) will usually SEEM too dark when used on 35mm transparencies, or larger transparencies viewed inside a black mask.
The simple solution is to make a modified profile with a built-in lightening edit, as described next.
If, after setting white and black (e.g. with Photoshop Levels), all scans are consistently too dark, you can make the scanner profile lighter by DARKENING the target scan slightly with a simple curve or gamma adjustment. This fools the profiling software into believing the darkness is a scanner characteristic, which it 'corrects' with a complimentary amount of lightening. If you adjust the darkening curve to suit your average original, most images will look pleasing after assigning the profile and setting white and black.
CAUTION: A lightened scanner profile that works well for 35mm scans may cause large-format transparencies to appear too light.
The following scanning and editing workflow is based on acquiring a raw RGB scan without any image-specific scanner adjustments. This workflow is recommended for all demanding work, especially when the best possible match to the original is required.
NOTE: As with all workflows, accuracy and consistency are dependent on many factors, only one of which is the profile. This workflow does not necessarily guarantee perfect results.
For custom workflow training, contact firstname.lastname@example.org