What makes a beauty effect look natural

Every beauty preset eventually faces the same question: does the person still look like themselves, only better? Two presets can use identical features and land on opposite sides of that line. The difference is rarely which effects are on — it is how hard each one is pushed.

Keep the texture

Smoothing is where presets go wrong most often. The tempting move is to raise the strength until blemishes disappear, and that is exactly when the skin stops reading as skin. Pores and fine shading are what the eye uses to decide “this is a face, not a sticker”.

A workable rule: pick the lowest strength that evens the tone, then let the other effects carry the rest.

• Cheek and forehead highlights can take more; they are already smooth.
• Under-eye and nose-wing areas need the least — that is where flatness shows first.
• A preset tuned on a single face looks wrong on the ninth. Test across skin tones, ages and camera quality.

Give every parameter a range

The API takes a float, but your UI should not expose the whole 0–1 span to an end user. Wide ranges push people into the part of the curve where the effect is obvious, and obvious is what users report as “fake”.

EngineConfig config;
config.app_id = "your_app_id";
config.app_key = "your_app_key";

auto engine = BeautyEffectEngine::Create(config);

engine->SetSmoothing(0.8f);
engine->SetReshape(Reshape::EyeSize, 0.5f);
engine->SetLipstick(0.4f);

auto result = engine->ProcessImage(inputFrame);

Narrow the slider, then name the stops. “Soft / Balanced / Strong” survives user error; a raw 0–100 does not.

Preview before you ship

The cheapest fix in the whole pipeline is looking at the result. A preset that only ever ran on your dev machine will meet lighting you never planned for: a window behind the subject, a webcam with heavy noise reduction, a phone that over-sharpens.

Natural is not a look you find once. It is a range you keep presets inside, and the range is defined by the worst lighting your users have, not the best you have.

Want to see how the parameters behave on your own footage? The free tier runs every effect with a watermark, which is the fastest way to check a preset before writing any code.