Banding appears as visible stripes or steps in what should be smooth gradients—the sky transitioning from light to dark shows distinct horizontal bands rather than a seamless progression. This posterization effect results from insufficient bit depth to represent smooth tonal transitions, either in the original capture or introduced during post-processing.
The Mathematics of Banding
Digital images divide continuous tones into discrete steps. An 8-bit file offers 256 levels per color channel. When a gradient spans a large area—like an evening sky going from dark blue at the horizon to lighter blue overhead—those 256 steps must be distributed across potentially thousands of pixels.
If the gradient is subtle enough, neighboring pixels require nearly identical values. With only 256 choices, the smooth transition becomes visible steps: many pixels at value 142, then a jump to many pixels at 143, creating a visible boundary. The mathematical limitation becomes perceptually obvious.
Common Causes of Banding
JPEG Compression
JPEG format uses 8-bit color depth and lossy compression that can introduce banding, particularly in areas of smooth tonality. The compression algorithm groups similar pixels into blocks, which can create subtle tonal discontinuities that appear as bands in gradients. Shooting RAW and saving final files as high-quality JPEGs helps minimize compression-related banding.
Aggressive Post-Processing
Every adjustment redistributes available tonal values. Starting with an 8-bit file and making significant exposure corrections, curve adjustments, or color grading can create gaps in the histogram that manifest as visible banding. The adjustment spreads the limited tonal steps across a wider range, making the discrete nature more apparent.
Bit Depth Reduction
Converting from 16-bit editing workspace to 8-bit for export can introduce banding if not handled carefully. Dithering algorithms help by adding slight random noise that breaks up visible banding patterns, trading obvious banding for barely perceptible noise—usually a worthwhile trade.
Display Limitations
Sometimes apparent banding exists in the display rather than the file. An 8-bit monitor showing a 16-bit file might reveal banding that won’t appear when the image is printed or viewed on better displays. Always check problematic gradients on calibrated displays before attempting fixes.
Preventing Banding
Work in 16-Bit
Process images in 16-bit color depth, which offers 65,536 levels per channel—enough to maintain smooth gradients through even aggressive editing. Convert to 8-bit only for final output, and apply dithering during conversion.
Shoot RAW
RAW files capture 12-bit or 14-bit data, providing far more tonal information than 8-bit JPEGs. This extra bit depth gives substantial headroom for post-processing adjustments without introducing banding.
Minimize Adjustments
Getting exposure right in-camera reduces the need for extreme post-processing corrections that exacerbate banding. Use graduated neutral density filters for difficult sky exposures rather than relying on shadow and highlight recovery.
Add Grain
Intentionally adding subtle noise or grain breaks up visible banding by introducing random variation that disguises the stepped transitions. The same principle as dithering, but applied creatively as visible texture.
Fixing Existing Banding
If banding already exists, fixing it requires careful work. Subtle blurring can smooth transitions, but risks making the image look soft. Targeted noise addition in the banded region helps, as does slight localized adjustment to break up the uniform bands. Some editing software includes specific de-banding filters designed for this purpose.
Prevention remains far more effective than cure. Working in proper bit depth from the start avoids quality degradation that’s difficult or impossible to reverse later.
Related Concepts
- Bit Depth – Core cause of banding limitations
- RAW Format – Provides higher bit depth to prevent banding
- Gamma – Affects how tonal steps are distributed
- Color Grading – Can introduce banding if not careful
- Noise – Can be used creatively to mask banding
- Histogram – Shows gaps that indicate banding