Color Banding Test: Monitor Gradient Test Online
Color Banding Test. Click Start Colour Banding Test to run this color banding test and look for hard steps in smooth gradients. Run the touch screen test online for a few seconds under normal room lighting for the most reliable result.
Interactive Tool
The readout in the controls bar shows the ramp range and step size of the current gradient at the chosen brightness. Key grayscale values: #000 black, #333 dark gray, #666 medium gray, #999 gray, #CCC light gray, #FFF white.
This color banding test sweeps a smooth gradient across your entire screen so you can watch for visible stepping — hard bands instead of a smooth blend — in the transition. Click Start Colour Banding Test to cycle through a black-to-white sweep and a couple of subtler tonal gradients, keeping in mind that your browser itself dithers CSS gradients, so this is a visual approximation rather than a lab-grade banding measurement; still, obvious hard bands you can see are a genuine sign of low color-depth rendering.
Stripes in a sunset photo or rings around a lamp in a dark film scene are hard to judge on real content, so the Monitor Color Banding Gradient Test, a simple color banding test, strips the problem down to one smooth gradient. It fills your display with a ramp that should blend without a seam, so any stepped bands you spot belong to your display or its signal, not the picture. Look closely at the dark end: shadows, dark skies, fog and game menus are where your monitor runs out of tones first, and where brightness steps show up most.
What Gradient Banding Looks Like on Your Monitor
A healthy ramp has smooth transitions from one tone to the next. Gradient banding breaks that ramp into flat strips with visible lines between them, like a staircase where there should be a slope. On a horizontal ramp the strips stand upright; on a vertical ramp they run across the screen. A blue sky in a photo shows the same thing as rings of slightly different blue.
On a colour ramp each step can also shift hue, so a neutral fade picks up green, pink or yellow patches. That tint is a separate fault, a color cast, and it hints at processing in one channel rather than missing levels. Rate what you see on a simple scale:
- Excellent: no strips anywhere, even in the dark areas.
- Good: faint steps only near black, invisible in videos and games.
- Fair: clear strips across much of the ramp that also show in normal content.
- Poor: wide, obvious blocks that hurt image quality everywhere.
Why Bit Depth Causes Contouring
Every screen draws a gradient from a fixed number of code values per channel. When a ramp spans more pixels than there are values, the display repeats each value across a strip, and the edge between two strips is the band. More bit depth means more intermediate steps and narrower strips, which is also why it matters for color accuracy in fine tonal work. The monitor gamma test online is free and runs entirely client-side, so nothing about your display is ever uploaded.
The maths is short. A channel with n bits has \(2^{n}\) levels, three channels give \((2^{n})^{3}\) colours, and the width of one band is the ramp width divided by the levels it crosses:
$$\text{band width (px)} = \frac{\text{ramp width (px)}}{\text{levels in the ramp}}$$
Panel Bit Depth: 6-bit, 8-bit and 10-bit
Our worked example is a 34-inch ultrawide showing a black-to-white ramp across its full 3,440-pixel width. The table shows what each signal path gives that ramp. Screen width matters too: at 8 bits the same ramp makes 7.5 px strips on a 1,920-pixel display and 15 px strips on a 3,840-pixel 4K one.
| Signal path | Levels per channel | Total colors | Band width on 3,440 px |
|---|---|---|---|
| 6-bit panel | 64 shades | 262,144 | 53.8 px |
| 8-bit, limited range (16-235) | 220 | 10.6 million | 15.6 px |
| 8-bit, full range | 256 shades | 16.7 million colors | 13.4 px |
| 10 bits per channel | 1024 shades | 1.07 billion colors | 3.4 px |
Those shades per channel follow the panel technology more than the price tag. Budget TN models are often 6-bit, most IPS and VA models use 8 bits, and many OLED TVs and high-end OLED desktop models accept 10 bits.
What an 8-bit Panel Does per Color Channel
A true 8-bit display, meaning an 8-bit panel without dithering, gives each color channel 256 levels, so the full ramp above gets strips 13.4 px wide, too narrow to see at normal distance. Now stretch only the darkest eighth of the ramp (codes 0 to 32) across the same width: 33 levels over 3,440 pixels makes each strip 104 px wide. The same dark stretch gives 430 px strips on a 64-level path, 123 px with limited range and 26.7 px with 1,024 levels. That is why one screen can look perfect on a full ramp and band badly in a night sky.
Dithering and FRC
Dithering hides missing levels by mixing neighbouring ones. Temporal dithering, also called frame rate control, flickers a pixel between two values fast enough that your eye averages them, so a 6-bit panel can pass for 8-bit and an 8-bit panel with A-FRC can pass for 10 bits. A monitor with a high-bit LUT does its internal scaling at higher precision, which keeps calibration from creating new steps. Neither equals a true 10-bit path end to end: for 3D renderings or grading you still need 10-bit source files, an app that outputs them and a driver that passes them through.
Signal Problems That Break a Smooth Gradient
Plenty of bands are added before the picture ever reaches the hardware. Display processing, browser rendering, heavy video compression and compression artifacts in the source content all produce strips. If bands only show in streamed video, blame the stream. If screenshots carry the same bands onto another display, the cause is in the file or the browser, not your screen. Run the hdr display test tool for a few seconds under normal room lighting for the most reliable result.
Full RGB Range vs Limited Range
PCs use codes 0-255; TVs and video use 16-235. When your computer sends limited range and the display expects full RGB range, the picture is stretched from 220 levels to 256, leaving gaps that show as bands. A mismatch the other way gives black crush and clipped highlights, where pure black swallows dark detail and the brightest tones merge into pure white.
DisplayPort, HDMI and the Signal Chain
The cable and ports set a ceiling on colour depth. HDMI 1.4 at 4K only carries 8 bits per channel, while DisplayPort 1.2 and HDMI 2.0 can carry 10. A cheap adapter or a long cable may also force the signal chain down to fewer bits or to 4:2:2 colour when bandwidth runs short. Test SDR and HDR modes separately, since each uses its own path. If the ramp bands over one connection but blends cleanly over another, that link is dropping levels, so rerun the ramp after every cable or port swap.
How to Run the Monitor Color Banding Gradient Test
Set the display to its native resolution, open the gradient test tool and switch it to full screen (or press F11 for fullscreen in your browser) so no window edges break the ramp. Sit at your normal viewing distance. This screen gradient test works like any color gradient test: you judge the steps by eye.
Start With Grayscale, Then Each Channel
Begin with the grayscale ramp, because a neutral gray fade shows both bands and tints. A grayscale gradient test that shows strips in every colour points to bit depth or range; strips that appear in only one of red, green, and blue point to processing or clipping in that channel. Run all four gradient ramps in turn, then repeat with the dark end zoomed if the tool offers it. Pages titled screen grayscale gradient test run the same black-to-white check.
Rotate the Pattern to Rule Out Uniformity
Next, rotate the ramp 90 degrees. Bands caused by the gradient turn with it; a stripe that stays put is a uniformity fault in the hardware, such as uneven backlight or backlight bleed, and needs its own test pattern. Check from straight on, since viewing angles shift tones on TN and VA screens. A single odd dot is one of the dead pixels and a flicker is a refresh rate issue; neither is banding, so don't read them into this ramp.
Counting Strips on a Grayscale Ramp to Catch a Range Mismatch
A night-sky photographer has a 27-inch 2,560 x 1,440 monitor plugged into a laptop over HDMI, and every Milky Way edit shows rings around the horizon glow. The decision on the table: return the monitor inside its 30-day window, or fix something upstream. The answer turns out to be the cable setting, not the panel: the dark ramp shows 28 strips where full-range 8-bit should show 33.
The steps run in this order. First, the full-screen grayscale ramp at native resolution: faint strips, mostly near black. Next, the dark-end view, which stretches codes 0 to 32 across all 2,560 pixels. With a ruler held to the glass, the strips measure roughly 91 px each, and counting edges from left to right gives 28. That number is the clue. Full-range 8-bit has 33 levels between 0 and 32, so the strips should be 2,560 / 33 = 77.6 px. If the laptop sends video levels (the 16-235 range in ITU-R BT.709) and the monitor stretches them back out, codes 0-32 collapse to 28 levels, and 2,560 / 28 = 91.4 px. The measurement matches the second case almost exactly.
| Signal path | Levels in codes 0-32 | Strip width on 2,560 px |
|---|---|---|
| 8-bit, limited range | 28 | 91.4 px |
| 8-bit, full range | 33 | 77.6 px |
| 6-bit panel, no dithering | 8 | 320 px |
| 10 bits per channel | 129 | 19.8 px |
Intel's Quantization Range sits on Limited, the default this laptop picked for an HDMI display. Switching it to Full and rerunning the same dark ramp gives 33 strips at about 78 px, and the horizon rings in the photo fade from obvious to barely there. The monitor stays.
A caveat worth knowing: counting only works on a panel that shows each level as a clean strip. On a 6-bit panel with frame rate control, or with a driver that dithers its output, strip edges blur into noise and a count tells you little, so judge those by whether the steps are visible at normal distance instead. In my view the strip count is the most useful number this color banding check gives, because it points at the stage of the chain that is losing levels, not just at the fact that something is.
How to Fix Contouring on Your Display
Work through the fixes in order and rerun the ramp after each one, so you know which change helped. Treat each rerun as a fresh test for color banding: same ramp, same distance, same lighting.
Set GPU Output and Output Color Depth
The most common fix is the GPU output format. Set RGB, full range and the highest output color depth the monitor accepts:
- NVIDIA Control Panel: Display, Change resolution, Use NVIDIA colour settings, RGB, 8 bpc or 10 bpc, Full dynamic range.
- AMD Software: Display, Pixel Format set to RGB 4:4:4 PC Standard (Full RGB), plus the colour depth setting.
- Intel Graphics Command Center: Display, Color, Quantization Range set to Full.
Update the graphics driver afterwards, because a driver reset can quietly drop the graphics card back to limited range or a lower pixel format.
Where Calibration Helps
Software calibration can ease bands by smoothing gamma curves and the monitor's contrast settings, but it can't add levels the hardware lacks; an 8-bit correction applied on the card can even add steps. A hardware calibrator that writes to the monitor's own LUT is the better route, followed by loading the matching color profile. Before you calibrate, turn off vivid modes, dynamic contrast and shadow enhancement, set a sensible color temperature, and leave gamma at 2.2.
When Stepped Bands Point to the Panel Itself
Suspect the display when the same strips appear in every app, over every connection and from more than one computer, and when their position on the panel never changes. Tints or crushed shadows that survive a factory reset deserve a closer look as well.
Keep a sense of scale before you file a warranty claim. Faint contouring that you only find by hunting on a test pattern rarely shows in films, games or design work. Base return decisions on bands that repeat in real photographs and footage. Photographers and video editors who grade smooth skies all day are the ones for whom a native 10-bits-per-channel panel is worth the extra cost.
Frequently Asked Questions
- What is color banding and why does it happen?
- Color banding is when a smooth gradient shows as separate strips of colour. It happens when the display, its settings or the signal can't show enough in-between tones, usually because of low bit depth, a limited RGB range or heavy compression in the image itself.
- How do I read the result of a gradient test?
- A smooth ramp with no visible lines means your screen is drawing the gradient correctly. Faint steps only near black are normal on many 8-bit monitors. Clear strips across the whole ramp point to a 6-bit panel, a limited-range output setting or a cable that is cutting colour depth.
- Can software calibration fix gradient banding?
- Only partly. Calibration can smooth the gamma curve and reduce visible steps, but it can't add levels the panel doesn't have. Correction applied on the graphics card can even add bands, so a hardware calibrator that writes to the monitor's own LUT is the better choice.
- Why do gradients look different on different displays?
- Displays differ in bit depth, panel type (IPS, VA, TN, OLED), dithering, factory calibration and processing. The cable, the graphics card's output range and the viewing angle also change how smooth the same gradient looks.
- Is gradient banding more noticeable in photos or videos?
- It usually shows more in still images, because your eyes have time to study the tones. In video, motion can hide it, but large dark scenes and skies in compressed streams often band badly, and that comes from the video file rather than the monitor.
- Why do I see bands only in one colour?
- If the grayscale ramp is smooth but red, green or blue alone shows steps or a tint, the problem is usually processing or clipping in that one channel, such as a picture mode, colour temperature setting or a calibration profile.
- How often should I calibrate my monitor?
- For colour-critical photo or video work, calibrate about once a month. For everyday use, every three to six months is enough. Rerun the gradient test after each calibration to make sure no new steps appeared.