Fixing banding artifacts in Coolscan scans of dense negatives

When you pull a heavily underexposed roll of Tri-X or Portra 800 from a forgotten plastic canister and feed it into a Coolscan 8000, the densest frames sometimes come out looking like a striped beach towel. Those parallel lines, faint or sharp, repeating across the height of the image, are what most hobbyists call banding. They are especially frustrating because they show up precisely where you need the most data: the deep shadows of a backlit portrait or the dark face of a performer lit by a single stage lamp in a Melbourne theatre. Worse, they often appear intermittently, which makes them feel like software glitches rather than a mechanical reality.

The good news is that most banding in Coolscan output is not a sign of imminent scanner death. It usually points to a handful of well-understood physical causes: LED output drift, misaligned analog gain, dirty optical path, or simply heat and humidity affecting the film transport. With a careful sequence of checks, you can usually track the source of the streaks and remove them without sending the unit off to a specialist.

What banding really looks like in a dense negative

Banding in a Coolscan scan tends to show itself in one of two ways. The first is broad horizontal bands a few millimetres tall, repeating down the image. These are usually linked to the LED moving unevenly across the film or to the sensor picking up small variations in the light source. The second pattern is a finer, repetitive stripe that looks almost like a watermark or a faint venetian blind laid across the slide. That finer pattern is often a moiré effect, but when it appears only in the dense parts of a negative, it is more likely related to quantization at the bottom of the analog-to-digital converter's range.

The reason dense negatives trigger this more than thin ones comes down to how the scanner adjusts its sensitivity. When the Coolscan encounters a very dark frame, it automatically increases the gain of its photomultiplier or, in the LED-based models, boosts the integration time. Both adjustments exaggerate any small fluctuation in the light source, because the signal you are amplifying is itself weaker. If your LED is ageing, or the optical path has the faintest smear of dust, the noise floor rises into the part of the curve the scanner is now sampling, and stripes become obvious in the resulting TIFF.

It helps to remember that a 35mm negative is roughly five stops slower than a typical slide, so anything that worked fine on your last roll of Velvia might fall apart on a roll of HP5 shot indoors in Hobart's winter twilight. That contrast in base density is the trigger that pulls the problem into view.

Calibrating the LED and analog gain

A good starting point is to force a fresh calibration of the scanner before each batch of dense negatives. In Nikon Scan, open the preview, choose "Calibrate" from the scanner menu, and let the unit complete its full sequence rather than cancelling it early. The internal reference strip and the LED warm-up are both parts of the calibration, and skipping either will leave the gain settings based on a cooler, dimmer LED than the one actually doing the scan ten minutes later. Run the calibration twice if your scanner has been idle for more than a day.

If you are using VueScan or another third-party driver, look for the option to disable automatic exposure on dense frames. Many operators in Sydney's film photography community have found that turning off the auto-gain and instead setting a manual exposure compensation of -1.5 to -2 EV gives much more consistent results across a roll. The reasoning is simple. Auto-gain makes decisions frame-to-frame, and that decision is exactly what produces the uneven signal that banding amplifies.

Pay attention to the LED itself. The Coolscan 5000 and 8000 use a white LED module that drifts in output over the first ten to fifteen minutes of operation. If you have just powered the scanner on and your first frame is a dense negative, you are scanning the LED at its coldest and dimmest. Let the unit idle with the lid closed for at least ten minutes, scan a throwaway preview first, and only then run the dense roll. This single habit has fixed more banding complaints in Australian home darkrooms than any software trick.

Managing heat, humidity, and mechanical drift

The Australian climate is not kind to precision optics, and that matters more for film scanning than many hobbyists realise. In a Brisbane summer, an un-air-conditioned room can easily push the scanner's internal temperature above thirty-five degrees, and at that point the silicon photodiode starts to drift in its dark current, producing faint vertical bands that appear and disappear as the unit heats and cools. Air-conditioning the scanning room, or at least running a small fan to keep air moving across the scanner, makes a measurable difference in output stability. Even a corner of a cooler spare room can be enough.

Humidity is the other quiet troublemaker. Film that has been stored in a garage in Perth through a wet winter can be slightly warped, and the warped film does not sit flat against the scanner's film guide. The gap between film and glass, even a fraction of a millimetre, can create uneven illumination across the height of the frame. If you are scanning older negatives, consider letting them acclimatise in the scanning room for at least twenty-four hours before running them. Specular and other Australian scanning suppliers also sell low-cost anti-Newton-ring glass replacements that hold warped film flatter, and these are worth the small investment for dense archival work.

Mechanical drift in the film transport is rarer but worth ruling out. Open the scanner lid, watch the film carrier move through a preview scan, and look for any stuttering or hesitation. If the carrier stops and starts, especially as it crosses the densest part of a frame, the banding you see may not be electrical at all but a symptom of a dry rail or a tired belt. A tiny amount of appropriate lubricant on the guide rails, applied carefully so it does not migrate onto the optical path, often restores smooth travel and removes the bands in the process.

Software workarounds in Nikon Scan and VueScan

When hardware checks do not fully eliminate the streaks, the scanner software still has a few levers left to pull. In Nikon Scan, the most useful is the "Multiple Sample" option in the analogue gain menu. Setting it to 2x or 4x forces the scanner to read each line of pixels two or four times and average the results, which dramatically reduces the random component of banding at the cost of scan time. For a single dense negative that matters to you, the extra minutes are well spent. Digital ICE and the GEM rocking filter should be enabled as well for any colour work on dense slides, since the infrared cleaning pass also acts as a second reading that smooths random noise in the dense shadows. Be aware that it will dramatically soften the grain on black and white film, so it is best left off there.

Bit depth matters too. If you are scanning at 8-bit output, the scanner has only 256 levels to describe a negative that may span a density range of twelve or more stops, and the rounding errors that result look exactly like banding. Switch to 16-bit output for any dense negative, even if you plan to convert to 8-bit later for screen viewing. The intermediate file preserves the data you need to do a clean levels adjustment afterwards, and a future re-edit will not be locked into the banding you saw at the time of capture.

Finally, consider whether the banding you are fighting is actually banding at all. Some dense negatives, especially pushed black and white, develop a grain pattern that, combined with the scanner's pixel grid, can look like bands when viewed at certain magnifications. Running a careful unsharp mask in Photoshop or GIMP at a radius of 0.6 to 1.0 pixel usually clarifies the situation. If the bands vanish with a small radius of sharpening, they were likely grain aliased onto the sensor, not true banding, and the fix is in post rather than in the hardware.

When the streaks point to a deeper hardware issue

If you have worked through the LED warm-up, the calibration, the humidity, the lubrication, and the software workarounds and you still see persistent horizontal or vertical stripes, the problem has likely moved into the scanner's analogue front end. Failing capacitors in the gain stage, a tired photomultiplier tube in the older 4000ED, or a damaged slot connector on the main logic board can all produce exactly this kind of recurring pattern. Diagnosing these requires opening the unit and inspecting with appropriate tools, which is beyond what most hobbyists want to attempt on a working scanner. Recapping the analogue board is a recognised fix in the repair community for the older 4000ED and 5000 units, both of which used capacitor batches that are well past their rated service life by now.

For these deeper issues, the Australian online forums and the small community of Coolscan repair specialists can often point you at the specific board revision that suffers from a known fault. One recurring cause worth investigating before more invasive work is a slot connector issue that has been reported on certain LS-4000 units after extended use, where the ribbon cable from the optical assembly develops a marginal connection and produces stripes that change position as the unit warms. Reseating the ribbon after carefully unlocking the connector is sometimes enough.

Repair or replacement of internal components is rarely economical on a Coolscan these days, which makes prevention all the more important. A clean optical path, a warm LED, a stable room temperature, and a careful calibration routine will keep banding out of your scans for years. A working scanner kept in a stable environment will outlast any repair budget, and a well-kept LS-5000 still produces scans that compare favourably with much newer dedicated units on the second-hand market.

The thing to take away from all of this is that banding in a Coolscan scan is rarely one thing. It is the interaction of LED warm-up, density of the negative, gain settings, and the room the scanner sits in. Work through those in a calm sequence and the stripes almost always give way to a clean, detailed scan of the dark frames you cared enough to keep.