What gets measured
| Measure | What it detects | | --- | --- | | Sharpness | Missed focus, shake, or an upscale from something smaller | | Contrast | A flat, washed-out or hazy image | | Noise | Grain from high ISO or lifted shadows | | Clipped highlights | Blown areas where detail is gone for good | | Clipped shadows | Crushed blacks, same problem at the other end | | Bits per pixel | Whether the file is over- or under-compressed | | Distinct colours | Posterisation, or flat artwork mistaken for a photo |
The findings, not the numbers
Raw measurements are hard to act on, so anything outside a normal band is written out as a sentence. Nothing outside those bands means nothing is reported — which is a real answer, not an empty one.
Reading two measures together
The interesting conclusions come from pairs:
- Low sharpness with high bits per pixel — a blurred image saved at high quality. The file is large and there is nothing in it to preserve.
- Low noise with low sharpness — over-aggressive noise reduction. The grain went and the detail went with it.
- High clipping with low contrast — the highlights blew out and the rest is still flat, which usually means the exposure was pushed after the fact.
Everything stays local
The image is decoded and measured in this tab. Nothing is uploaded, so a client photograph can be checked before you send it anywhere.
Questions
What is bits per pixel and why does it matter?+
File size in bits divided by pixel count. It is the only compression measure that compares fairly across dimensions and formats — a 200 KB thumbnail may be far more compressed than a 2 MB poster. Under 0.4 expect visible blocking; over 8 the file is much larger than it needs to be.
What counts as clipping?+
Pixels at the very top or bottom of the range: pure white or pure black. Those carry no detail and none can be recovered, because the sensor or the encoder saturated. A few percent is normal in a contrasty scene; a lot means the exposure was wrong.
Is a low noise number always better?+
No. Heavy noise reduction produces a very low number and a waxy, detail-free image. Read it alongside sharpness: low noise with low sharpness usually means over-processing rather than a clean capture.
Does this replace looking at the image?+
No, and it does not try to. These are measurements; whether a value is a problem depends on what the image is for. A deliberately grainy photograph and a badly denoised one produce different numbers and neither is wrong.