How to make a photo black and white (without it looking flat)

A practical guide · Updated June 2026

Making a photo black and white takes one click. Making a photo that is better in black and white takes judgement, because the conversion deletes an entire dimension of information. A colour photo separates things two ways: by brightness and by hue. Grayscale keeps the first and throws the second away, so an image leaning on hue collapses into grey mush. This guide covers what each method does to your pixels, which photos survive, the one edit nearly every conversion needs afterwards, what happens to file size, and how to batch-convert privately.

Grayscale, desaturation and true monochrome are not the same thing

Three terms get used interchangeably and produce visibly different files. Knowing which one your tool does explains most "why does this look wrong" moments.

MethodWhat it does to each pixelTypical resultWhere you meet it
Grayscale (luminance-weighted)0.299 × red + 0.587 × green + 0.114 × blueNatural tones matching how the eye reads brightnessImagloo and most one-click B&W tools
Desaturation(brightest channel + darkest channel) ÷ 2Muddier; unrelated colours land on identical greysThe saturation slider in basic photo apps
Channel mixing / colour filtersA custom recipe, e.g. 100% of the red channelDramatic and controllable — darkens skiesB&W mix panels; film red filters
True black and white (1-bit)Every pixel forced to black or white at a thresholdStark and graphic, no greys at allFax, screen printing, dithered art
Toned monochrome (sepia)Grey, then one hue re-applied across the rangeA warm or cool one-colour imageSepia presets, darkroom toning

The weighting in the first row matters more than it looks. Green counts for 58.7% because human vision is most sensitive to it, red for 29.9%, blue for only 11.4%. Proper luminance turns pure red into grey level 76, green into 150 and blue into 29 — a huge spread. Desaturation turns all three into level 128, because (255 + 0) ÷ 2 is 128 every time. That is why a saturation slider dragged to zero looks lifeless while real grayscale has bite.

One vocabulary fix: monochrome means "one colour", so grayscale and sepia both qualify. "True black and white" properly means the 1-bit case, which is almost never what people mean.

What survives the conversion, and what dies

Learn to look at a scene the way the converter will. Four qualities carry through and get stronger:

Three things vanish: hue contrast, separation that existed only because two things were different colours of similar brightness; colour-based isolation, the red jacket in a grey crowd; and colour mood. The tones remain, the emotional cue does not.

A worked example: the apple that disappears

Here is the whole problem in two arithmetic lines. Picture a red apple on green grass — a photo that pops because red and green sit opposite on the colour wheel.

The apple is roughly RGB (200, 30, 40). Run the luminance formula: (0.299 × 200) + (0.587 × 30) + (0.114 × 40) = 59.8 + 17.6 + 4.6 = grey level 82. The grass is roughly RGB (60, 120, 50): (0.299 × 60) + (0.587 × 120) + (0.114 × 50) = 17.9 + 70.4 + 5.7 = grey level 94.

Twelve levels apart out of 255 — under 5%, well inside the range where the eye reads two areas as "the same". The apple does not just look duller; it stops existing as a distinct object. Every flat conversion is some version of this calculation.

The same maths explains two other classics. A red-orange autumn leaf at RGB (180, 60, 40) lands on level 94 and a green one at RGB (80, 130, 60) on 107 — 13 levels apart, so a blazing hillside becomes a uniform grey slope. A sunset's orange-to-magenta gradient, RGB (240, 120, 40) to (170, 90, 150), collapses from 147 to 121.

Film photographers solved this with coloured filters — a red filter darkens blue sky to near-black while leaving red objects bright, which is what a channel-mixing slider does digitally. A one-click converter has no such control, so predict the collapse instead: if you can only explain a photo in colour words, it is a colour photo.

Which photos actually work in black and white

Use this to triage a folder before you convert it.

Photo typeWorks in B&W?Why — and what to do
PortraitsUsually excellentSkin becomes tone, blotchiness disappears, eyes dominate. Needs directional light, not flat flash.
Street & documentaryAlmost alwaysGesture, shadow and geometry carry the frame, and it reads faster without mixed street colour.
ArchitectureReliably yesLine, repetition and cast shadows are all tonal. Buildings are near-neutral anyway, so little is lost.
LandscapesDepends on the skyWorks with cloud structure, mist or terrain contrast. A flat blue sky becomes featureless grey.
Fog, snow & minimal scenesYes — often betterAlready near-monochrome. Converting removes the colour cast and lets shapes breathe.
Product shotsOnly if texture-ledLeather, steel and glassware do well. Anything chosen by colour must stay in colour.
FoodRarelyAppetite is a colour response; grey meat looks inedible. Exceptions: baking, coffee, still life.
Autumn foliageAlmost neverPure hue contrast at near-identical brightness — the 13-level example above. It becomes a grey hillside.
SunsetsAlmost neverThe gradient is entirely hue-based — without a silhouette, just a bright blob in a dim sky.

A fast pre-flight test: shrink the photo to thumbnail size and squint. If subject and background still separate as light and dark shapes, the conversion will work. If the subject only emerges through its colour, it will not.

Portraits and skin tones — the easiest win

Portraits improve in black and white more often than not, for technical rather than nostalgic reasons. Skin is not one colour: it carries red from irritation, blue-green from veins, yellow in the highlights, and those variations are what we read as blotchiness. Luminance conversion flattens them into a narrow band of grey — an under-eye shadow and a red cheek land within a few levels of each other, a smoothing effect that costs no real detail.

What to watch:

Raise the contrast after converting, not before

Almost every straight conversion needs one follow-up edit, and skipping it is the most common cause of a dull result. Once hue separation is gone, tonal separation works twice as hard — yet the converted file typically sits bunched between level 25 and level 230. Nothing is truly black or white, so it reads as grey-on-grey.

The fix, in order:

  1. Set the black point and white point first. Pull the shadow end up until the darkest 0.1% of pixels clip to black and the highlight end down until the brightest 0.1% reach white — this alone is most of the improvement.
  2. Add a gentle S-curve. Lift the highlights, drop the shadows, leave the mid-tones alone. Roughly 10–20% more contrast suits most images; 30%+ destroys texture.
  3. Aim for a full range. A good black-and-white image holds a patch of near-black (levels 0–10) and one of near-white (245–255), even if the bulk sits mid-frame.
  4. Sharpen last, and lightly. Without colour, sharpening halos are far more visible.

Order matters: crop before converting, so you judge contrast on the final composition — see how to crop an image.

Does black and white make the file smaller?

Yes, but far less than people assume. JPEG does not store red, green and blue: it stores one brightness channel plus two colour channels, kept at half resolution because the eye barely notices. Colour therefore accounts for only 10–20% of the bytes — the brightness detail is the file. Grayscale it and those two channels flatten to a constant that compresses to almost nothing, so a 2.4 MB colour JPEG lands near 2.0 MB.

PNG does better. There, every pixel genuinely stores three bytes, and making them identical creates exactly the repetition PNG's compression exploits — expect 25–50% smaller.

So convert for how it looks and compress for how big it is. Running that 2.0 MB grayscale JPEG through the image compressor typically brings it under 600 KB with no visible loss — a far bigger win than the conversion. Our guide on compressing images for web covers the quality settings.

Batch converting a whole set consistently

Converting one photo is easy; converting forty into a coherent set is the real task. Four habits:

How to convert privately in your browser

You do not need an editor or a subscription, and you should not upload portraits or client work to a stranger's server just to remove their colour. Imagloo's black and white converter runs on your own device, so nothing is transmitted and sensitive images stay private.

  1. Crop and straighten first using the cropping workflow, so you convert the final composition.
  2. Open the black and white converter and drop in a batch of JPG, PNG or WebP. Each file is redrawn with proper luminance weighting — not a saturation slider — and saved in its original format with a "-bw" suffix.
  3. Judge honestly. If the subject no longer separates from the background, that frame belongs in colour.
  4. Add contrast, then compress. Set a real black and white point, then send the files through the compressor before publishing.

Convert for tone, contrast for punch, compress for size — in that order, and only on the photos that earn it.

Frequently asked questions

Is grayscale the same as black and white?

In everyday use, yes — when people say black and white they almost always mean grayscale, an image made of 256 shades from black through grey to white. True black and white strictly means only two values with no greys at all, which is what a fax or a screen-printed stencil produces. Monochrome is the wider word: any image built from one colour, so grayscale and sepia both qualify.

Does converting a photo to black and white make the file smaller?

A little, but less than most people expect. In a JPEG the colour information is already stored at reduced resolution and typically accounts for only 10 to 20 percent of the bytes, so removing it shrinks the file by about that much — the brightness detail, which is the bulk of the file, is untouched. PNG does better, because a grey pixel is genuinely repetitive data: savings of 25 to 50 percent are common. If you need a smaller file, compress it rather than relying on the conversion.

Why does my photo look flat after converting it to black and white?

Because you removed one of the two ways the image separated subject from background, and it was probably leaning on colour rather than brightness. Straight conversions also leave the histogram bunched in the middle, with no true black and no clean white. Set a proper black point and white point, then add a gentle contrast curve. If the subject still merges into the background, that photo is simply a colour photo.

Try the free tool — private and in your browser

Convert one photo or a hundred in seconds — every Imagloo tool runs locally, so images are never uploaded: