Short answer: use a matte chroma-key green close to #00B140, light it evenly, and sample that exact patch in the keyer instead of trusting a preset. Shade matters: in the keyer's own color math, a dark shadow green sits 13.19% from a green shirt, while lit #00B140 sits 15.87% away. The darker the green, the less room you have.
The green is a point; the tolerance is a sphere
This page is about choosing the backdrop, not about tuning the slider after you have one. The distinction matters because every pixel is a point in RGB space. When you click a patch of green, the keyer puts the center of a sphere there, and the Tolerance slider sets its radius as a percentage of the 441.67-point RGB diagonal. Every pixel inside the sphere becomes transparent. Every pixel outside it stays opaque, with a fade band at the edge if you add softness.
So the green you choose is really a choice about which materials sit near the center. The table below measures seven candidate backdrops against six subject colors using the same loop as this site's keyer. The closest test color is the first one the sphere reaches. A bigger distance is more room; a smaller distance is less.
| Backdrop | Closest test color | Distance | What it means |
|---|---|---|---|
#00B140 chroma key green |
Green shirt | 15.87% | The studio standard. A green shirt is the whole margin; neutral clothing and skin are far away. |
#00FF00 digital green |
Green shirt | 34.40% | The largest numerical margin here. It is also the brightest and the least forgiving in a real shot; see the warning below. |
#008000 web green |
Green shirt | 17.81% | A common paint shortcut. Only 1.9 points more room than #00B140, not a free upgrade. |
#00612B dark chroma green |
Green shirt | 13.19% | This is the shadow side of a good screen, not a paint choice. Bad lighting turns #00B140 into this. |
#00310F deep shadow green |
Black hair | 8.06% | Black hair is the first casualty, not a green prop. There is almost no separation left. |
#0047BB chroma key blue |
Teal mug | 21.62% | The blue alternative. A green shirt is 32.42% away, but a teal or cyan prop is now the closest object. |
#5B7F2A olive green |
Green shirt | 11.05% | Grass or olive paint. The tightest useful green tested: less room than a shadowed chroma screen. |
The six test colors are light skin #D9A07A, white shirt #E6ECE6, navy denim #1F2A44, black hair #141414, green shirt #2F7D3F, and teal mug #1F6F6A. Distances were computed on 2026-10-29 with the exact sRGB loop from /green-screen-remover: square root of the three channel differences divided by 441.6729559. A solid swatch is the easy case; mixed edge pixels sit closer to the key, which is why the material page also measures blends.
#00FF00 wins this table and still is not the recommendation. The table measures color separation only. A pure green is the brightest saturated green you can put behind a face, so it bounces more green onto skin, hair and light clothing, and it clips easily on consumer cameras. A clipped channel is no longer the color you sampled. Use #00B140 unless you have tested your own camera and lights; the extra 18.5 points of separation on paper are not worth a spill problem you then have to fix in post.
Does shade matter? Five versions of the same green
Shade is not a cosmetic difference. The same fabric can run from #00B140 in the lit corner to #00310F in a shadow, and the keyer does not know those are the same material. It only knows how far each pixel is from the green you clicked. The table keeps the key fixed at #00B140 and asks how wide the tolerance has to be to clear each shade, then shows what that width does to three materials.
| Backdrop shade | Tolerance needed from #00B140 |
Green shirt α | Teal mug α | Navy denim α | Margin |
|---|---|---|---|---|---|
#00B140 |
0.00% | 255 | 255 | 255 | +15.87 |
#00A73D |
2.36% | 255 | 255 | 255 | +13.51 |
#008936 |
9.34% | 255 | 255 | 255 | +6.54 |
#00612B |
18.73% | 0 | 28 | 255 | −2.86 |
#00310F |
31.03% | 0 | 0 | 29 | −15.16 |
Alpha 255 is fully opaque, 0 is fully transparent, 28 is about 11% opaque. All alpha values were recomputed on 2026-10-29 with the keyer's piecewise formula at 3% edge softness. The five shades are the exposure steps from the 2026-10-28 falloff test; the point here is the color choice, not the zone strategy.
Read the last two rows. A shadowed corner at #00612B needs 18.73% to disappear, and at that setting the green shirt is already at 0 while the teal mug is down to 28. A deeper #00310F needs 31.03%; now the navy denim is at 29 and the teal is gone. Skin, white and black hair stay at 255 in both rows, but that is not much comfort when the wardrobe is disappearing.
This is why “just make the green darker” is bad advice. A darker green is not a safer version of a good key. It is a different key with a wider net, and the wider net reaches the things you wanted to keep.
The five-point margin
Here is a rule you can apply before the shoot instead of after. Subtract the tolerance needed to clear the darkest part of the backdrop from the distance to the closest material you must keep. Call the result the margin. Keep it at five points or more.
| Margin | What it means | What to do |
|---|---|---|
| 5 points or more | With 3% softness, the material is fully opaque and you still have about two points of cushion for light drift. | Shoot. Sample the darkest clean patch on the day, not the patch from the test. |
| 0 to under 5 | The material is in or near the fade band. It may look solid on one frame and translucent on the next. | Add light to the dark end, change the wardrobe, or split the frame into zones. |
| Below 0 | No single tolerance clears the backdrop without touching the material. | Repaint or relight the screen. A bigger slider number cannot fix this. |
The five points are not arbitrary. At 3% edge softness a material starts to fade at the tolerance value and reaches full opacity three points later, because the keyer maps the fade band to alpha = 255 × (distance − tolerance) / softness. A five-point margin means the material is past the fade band with two points to spare. A two-point margin gives 255 × 2 / 3 = 170, which is visibly translucent. If you shoot with zero softness the arithmetic is kinder, but the edge is jagged; do not trade the edge for margin.
- Sample the darkest clean patch of the backdrop, not the average and not the brightest corner.
- Raise tolerance until that patch is 0 and write the number down. This is the demand side of the margin.
- Measure the closest material with the two-click meter on green screen remover settings by material. Write that distance down too.
- Subtract. If the result is under five, change the plan now: add light, swap the green, change the wardrobe, or use the zone pass on the uneven backdrop page.
- Re-check on the day. A test shot on Thursday does not transfer to a different lens, a different white balance, or a phone screen in the frame.
Worked examples from the table above: #00B140 has a margin of +15.87 and is comfortable. #008936 has +6.54 and passes, but only just; any drift past six points starts to show. #00612B is at −2.86 and #00310F at −15.16, so neither can be saved by moving the tolerance slider.
If you already have the wrong green
Most people do not choose a green from a chart; they inherit a wall, a cloth, or a paint can. These are the practical exits, in the order that costs least:
- A dark painted wall. Add light before you touch the key. Two lights at roughly 45 degrees to the wall, one from each side, reduce the spread. If the darkest patch still needs a tolerance that reaches the wardrobe, repaint or hang a flat cloth. The zone pass is the fallback when the wall cannot change.
- An olive or grass-green wall (
#5B7F2A). This is the tightest green in the table: 11.05% to a green shirt. It can work for a neutral wardrobe, but a green or teal prop will fail. Repaint if you shoot people in green; otherwise keep the wardrobe neutral and measure. - A
#008000wall. This is usable. Its closest test color is still a green shirt at 17.81%, so the safe tolerance with a five-point margin is about 12.8. Do not push past that because the corners look clean on a small preview. - A green wardrobe. Switch the screen, not the shirt.
#0047BBputs a green shirt 32.42% away, but brings a teal mug to 21.62%. Check every prop for teal and cyan before committing; the two-color key page covers the mixed case. - A phone or tablet in the shot. Its screen is a light source. When it goes white the green gets more light; when it goes black the green darkens. Lock exposure and white balance, sample a clean patch at the start of each take, and do not reuse one key across takes. The spill page explains the reflected-green part.
- A rough stucco wall. Matte chroma paint helps the color, but texture casts tiny shadows. Test a one-metre square under the actual lights before painting the room; if the spread is more than about five points, a flat cloth in front is cheaper than repainting.
A four-step color test before the shoot
- Put the real wardrobe on the screen. A color chart does not tell you that the shirt is 11 points from the wall; the shirt does.
- Light the screen the way you will shoot it. Include the darkest corner, not just the patch under the meter.
- Take one still and sample the darkest clean patch in the keyer. Raise tolerance until the backdrop is gone and note the number.
- Measure the closest material and subtract. If the margin is under five, fix the light, the green, the wardrobe, or the zone plan before the shoot, not in post.
That is the whole test. It takes one frame and two clicks, and it catches the failure that a tolerance slider cannot: a green that is technically the right hue but too close to something you need to keep.
Green screen remover color FAQ
What color should I use for a green screen, and does shade matter?
Start with a matte chroma-key green near #00B140. It is a compromise, not a magic value: against the six test colors on this page its closest neighbor is a green shirt at 15.87%, so a green wardrobe is the real constraint. Shade matters because the keyer measures distance from the patch you sample. A shadowed #00612B needs 18.73% tolerance to clear, and at that setting the green shirt is already gone. If the screen is uneven, a brighter green does not fix it; it only moves the nearest material further away on paper. Light the screen first, then sample the darkest clean patch.
Which color of cloth reflects least green light?
That question has two readings. If you mean the screen fabric, a matte mid-green reflects less than a glossy one, and a darker green reflects less than a bright one — but a darker green also shrinks the gap to black hair, measured at 8.06% for #00310F. If you mean the subject's clothing, avoid green, teal and light cyan: from #00B140, a green shirt is 15.87% away and a teal mug is 19.05%, while navy denim is 31.37% and black hair is 37.19%. Matte fabric and distance from the screen matter more than a brand name. A Video SE question about cloth reflection describes the same problem from the shooter's side.
Can I paint a rough wall with chromakey paint to make a green screen?
Yes, but roughness is a lighting problem before it is a paint problem. Stucco casts tiny shadows, so the same wall reads as several greens; the keyer then needs a tolerance wide enough to cover the darkest dimple, and that width can reach the wardrobe. Use matte chroma paint, light the wall from both sides at roughly 45 degrees, and measure the spread across at least four patches before painting the whole room. If the spread is more than about five points, hang a flat cloth in front instead. The measured 18.73% falloff test on the uneven backdrop page shows what a 19% tolerance does to a green shirt. The original question is on Video SE.
Why did my green background change when the phone screen changed?
The phone is a light source. When its screen goes white it adds light to the green and the camera exposes the scene slightly differently; when it goes black the green darkens. That is not a keyer bug, and one sampled green cannot cover both ends. Lock exposure and white balance, shoot a clean patch at the start of each take, and key each take from its own patch. If the change happens inside one take, split the take or fix the light. A shooter described exactly this with a painted canvas on Video SE; the same reflection problem is covered on the spill page.
Is pure green #00FF00 better than #00B140?
Only on the distance chart. #00FF00 sits 34.40% from the closest test material versus 15.87% for #00B140, so it gives the key more room on paper. In a real shot it is also the brightest green you can put behind a face: it bounces more green onto skin and hair, clips easily on consumer cameras, and makes spill worse. The chart measures separation, not spill. Use #00B140 unless you have measured your own camera and lights, and treat any “best green” claim that only quotes a hex code as incomplete.
Next step
Take the green you actually have, sample its darkest patch in the browser keyer, and run the five-point check before you shoot. If the margin fails, the fix is light, paint, wardrobe, or zones — not a bigger tolerance number. For the material side of the same decision, read settings by material; for the lighting side, read the uneven backdrop test.