RGB Channel Editor
Drive the red, green and blue of a picture yourself: three sliders with a live preview, and moving one channel never touches the other two.
or browse filesdrop an image here or paste
.png
Drive the red, green and blue of a picture yourself: three sliders with a live preview, and moving one channel never touches the other two.
or browse filesdrop an image here or paste
Every digital picture is three greyscale images stacked on top of each other: one for red, one for green, one for blue. This is the RGB editor for them. Drop a photo on the page and you get a slider per channel, live preview, and nothing between you and the pixels. Push red up for warmth, pull blue down for a golden look, drag green to kill a fluorescent cast, or take one channel out completely.
The RGB, R, G and B buttons above the sliders switch what is on screen. RGB is the edited picture. R, G or B shows that one channel on its own, as the greyscale image it really is, and the download follows the view: pick R and the file you save is the red channel. Flip the switch under the buttons to see it in its own colour instead of grey.
This is not a canned RGB filter or a one-click effect preset. You set the three numbers and you can see exactly what each one is worth.
Two more controls sit on the image itself. Auto reads the whole photo and sets all three sliders to neutralise its colour cast. Hold to compare shows the original for as long as you press it. The i button opens the numbers: the exact gain on each channel, what mid grey turns into, and how much of the picture a strong setting is burning out to white.
Drop a JPG, PNG or WebP onto the page, or paste one from the clipboard. Drag Red, Green and Blue, or press Auto and adjust from there. Switch to a single channel if that is what you came for. Download as PNG for a lossless file, or JPG or WebP for a smaller one. Everything runs in your browser: the image is never uploaded, there is no signup, and no watermark is added.
A pixel is stored as three numbers from 0 to 255. Take the first number from every pixel and you have the red channel, a greyscale image where bright means "a lot of red here" and dark means "almost none". Do the same with the second and third and you have green and blue. Stack the three back up and the picture returns.
That is why the single-channel view is grey rather than red-tinted: the red channel is a quantity, not a colour. Looking at it is genuinely useful. Skin and skies separate cleanly in one channel and not in the others, the blue channel is where noise lives in most phone photos, and a scanned document is usually cleanest in one channel alone. Showing it in colour is the other convention and the switch is there for it, but grey is the one every image editor uses.
Adjusting a channel means changing every one of those numbers by the same rule. Because the rule for red only ever looks at red, moving the Red slider leaves the green and blue bytes exactly as they were, not approximately. That is the whole promise of a channel editor and this page keeps it literally: a channel sitting at zero is passed through untouched.
Most online channel tools add a number to the bytes: red plus 40, and every pixel gets forty more red. It looks like it works until you look at the dark parts. Pure black is (0, 0, 0), so adding forty red makes it (40, 0, 0) and every shadow in the frame turns to dark red mud. At the other end everything above 215 is flattened into 255 and the highlights lose their detail.
Here each slider is a gain: the light in that channel is multiplied, in linear light, not in the compressed sRGB bytes the file stores. Black has no light in it, so anything times zero is still zero and the shadows stay clean. The slider runs one stop each way, so +100 is exactly twice as much of that channel and -100 is exactly half, and the readout beside each slider says the multiplier rather than a meaningless number out of a hundred. At +50 you get 1.41 times, because that is what half a stop is.
A gain can still clip at the top, and it would be dishonest to hide that: doubling the red in a picture that already has bright red in it pushes those pixels to 255 and they stop being distinguishable. The info card counts them for you and tells you what percentage of the picture that is, so you can back off before you lose anything you cared about.
What this page deliberately does not do is add an offset on top of the gain. Lifting the black point of a single channel is a levels operation and it belongs with the shadow lift in the color grading tool, which has three tonal zones to put it in. It also does not swap channels around: a swap would break the one-slider-one-channel rule that makes everything else here predictable.
A colour cast is one channel being consistently stronger than the other two, and the direct fix is to pull it back. Yellow indoor light means too much red and green, so drop them or lift blue. An underwater shot has almost no red left in it, so lift red. Old prints go magenta as the dye fades, so lift green.
Auto does it in one press. It works on the grey world assumption, that the average of a whole scene should come out neutral, so it measures the average amount of red, green and blue in the visible pixels and sets each slider to the gain that makes those three averages match. It ignores blown-out and near-black pixels, because neither carries any cast worth reading. Crucially it sets the sliders rather than applying a hidden correction, so you can see what it decided and pull it back or push it further, and running it twice does nothing the second time because it always reads the original picture.
Auto reads a scene wrong when the scene genuinely is one colour, a lawn, a sunset, a photo of a blue wall. When it does, just drag. If the cast you are fighting is the colour of the light rather than one channel misbehaving, the color temperature tool is the better instrument: it works out the same three gains for you from real black-body physics and gives you one warm-to-cool slider instead of three.
This page is the manual one, and that is the point. The color temperature editor derives the three channel gains from an illuminant in Kelvin; here you set them yourself. The color grading tool puts separate colour into shadows, midtones and highlights; here one setting acts on the whole picture. The hue shifter rotates every colour around the wheel and the image tinter casts the picture toward one colour you pick; neither lets you touch a single channel. For a plain greyscale conversion of all three at once there is the grayscale converter, and for brightness rather than colour there is exposure and contrast.
The preview and the download run the same three tables over the pixels, so what you see is what you save, at the full resolution of the file you loaded. With all three sliders at zero and the view on RGB the image passes through untouched, byte for byte. Transparency is kept, and the alpha channel is never touched by any of this.
All of it happens on your device through the canvas API, so no image is transmitted anywhere and the page keeps working offline once it has loaded. Large photos are previewed at a reduced size for speed, but the file you download is always rendered from the original.
Open the picture here and drag the Red, Green and Blue sliders. Each one multiplies the light in that channel and nothing else, and the preview updates as you drag. Press Auto for a one-click colour balance, hold the compare button to check against the original, then download as PNG, JPG or WebP at the original resolution. No account, no upload, no watermark.
They are the three greyscale layers a colour image is made of. Each pixel stores how much red, how much green and how much blue it has, from 0 to 255, and the screen adds the three together to make the colour you see. Editing an image channel by channel is more precise than editing the colour as a whole, because you can change one component without disturbing the other two.
Press R, G or B above the sliders. The picture is replaced by that single channel, drawn as a greyscale image, which is what the channel actually contains. The Download button then saves that channel, so this is also how you extract one: pick the channel, pick a format, save. Turn off the grey switch if you would rather see it drawn in its own colour.
Not on this page. Swapping channels is a different operation from adjusting them, and mixing the two would break the rule that each slider touches one channel and nothing else. For deliberately strange colour there is the hue shifter, which rotates every colour at once, or you can pull two channels down hard here to get most of the way to the same look.
It is the gain, the amount that channel is multiplied by. 1.00 is untouched, 2.00 is double, 0.50 is half. The slider runs one photographic stop in each direction, which is why the ends read 2.00 and 0.50 rather than something rounder. The i button on the image shows all three at once, plus what a mid grey pixel of 128 turns into, which is the clearest way to see what a setting is really doing.
Press Auto, which sets all three sliders so that the average colour of the picture comes out neutral. If the result is close but not right, keep adjusting from where Auto left it, since it moves the sliders rather than hiding a correction behind them. For a cast that comes from the colour of the light itself, indoor bulbs or open shade, the color temperature tool will get there in one slider.
Yes. JPG, PNG and WebP all load, including photos straight off a phone, and you can paste from the clipboard. Transparency is preserved and the alpha channel is never altered. The download is full resolution and can be a different format from the file you loaded.
No. The image is decoded, adjusted and encoded in your browser, at preview and at full download resolution alike. Nothing is sent to any server, and the tool keeps working with the network disconnected once the page is open.