How to Detect Blurry Photos & Check Focus Quality
Every photographer has felt it: the event's over, the card is full, and somewhere in those 2,000 frames are soft, out-of-focus shots that only get discovered on a big screen. Here's how to find them in seconds — and why you should never have to upload your photos to do it.
Why Photos Come Out Blurry
| Culprit | What it looks like | Fix |
|---|---|---|
| Camera shake | Smear in one direction | Shutter ≥ 1/focal length, stabilization, tripod |
| Missed autofocus | Sharp background, soft subject | Back-button focus, single-point AF, burst |
| Slow shutter, moving subject | Subject motion trails | 1/250s+ for people, pan with the subject |
| Wide aperture + shallow DoF | Edges soft, subject sharp | Stop down, or embrace it deliberately |
| Heat haze / long lens | Even softness, shimmer | Shoot earlier/later, get closer |
Whatever the cause, the symptom is the same: missing high-frequency detail. That's what a sharpness detector measures — not intent, not artistic choice, just how much fine detail the pixels actually contain.
How Sharpness Detection Actually Works
- 1 Grayscale conversion. The photo is drawn onto a canvas and reduced to an analysis size, so the math runs fast on any device.
- 2 Variance of the Laplacian. A 3×3 kernel measures contrast between each pixel and its neighbors. Sharp edges produce large positive/negative swings; blur flattens them.
- 3 Tenengrad edge energy. Sobel filters measure gradient strength in x and y; the summed energy is high when edges are crisp.
- 4 Normalize to 0–100. The combined metrics map onto a score, and the same math runs per-region to build the 4×4 heatmap.
These are the same focus metrics camera engineers have used for decades — they're deterministic, explainable, and they run in a fraction of a second on a modern laptop. No AI model, no cloud round-trip.
Reading the Score and the Heatmap
| Score (standard) | Verdict | What it usually means |
|---|---|---|
| 70 – 100 | Sharp | Strong fine detail — safe for print and large crops |
| 25 – 50 | Soft | Detail present but weak — fine for web/social, risky for print |
| 0 – 25 | Blurry | Missing detail — likely shake or missed focus, reshoot if it matters |
The region heatmap is the part photographers find most useful. It splits the frame into 16 regions and colors each one: green = sharp, yellow = slightly soft, orange = soft, red = blurry. A portrait with tack-sharp eyes but soft ears shows exactly that pattern, telling you the lens was focused on the eyes but depth of field ran out — not that the whole shot is bad.
One important subtlety: a plain sky or a clean wall has almost no detail even when perfectly focused. The detector marks very low-texture regions as neutral rather than blurry, so a dreamy sky doesn't tank an otherwise sharp landscape.
Batch-Scanning a Whole Shoot
The real workflow win is batch mode: drop an entire folder of photos and every frame is scored, then listed worst-first. The blurry frames rise to the top instantly, so you can:
- Flag the 5% of shots that need re-shooting or a retouch budget.
- Screen an entire catalog before it goes to print or a client gallery.
- Export the scores as CSV to keep with the shoot's notes or hand to a retoucher.
- Flip sensitivity to Strict for product shots (every frame must be tack-sharp) or Lenient for casual galleries.
For lens testing, the compare mode is the tool: load the same scene at two apertures and read both scores side by side — no more squinting at 100% crops.
Why Blur Checking Should Be Local
- Upload-based detectors send your entire photo to a server just to return a number.
- AI 'sharpening' services process your images remotely — and many keep them for training.
- A local detector reads pixels on a canvas in your browser: no upload, no account, no limits, works offline.
Frequently Asked Questions
What causes blurry photos?
The three big causes are camera shake (motion blur), missed autofocus (the lens locked onto the background), and a shutter speed too slow to freeze your subject. Each looks slightly different to a detector: motion blur smears detail in one direction, while missed focus softens everything evenly. A sharpness score flags the result either way — detail is missing.
How do sharpness detectors work?
Most use focus metrics computed directly on pixels: variance of the Laplacian measures how much contrast exists between neighboring pixels, and Tenengrad (Sobel gradient energy) measures edge strength. Sharp photos have high-frequency detail everywhere, so these values are high; blur smooths the image and collapses them. The Prescosoft Blur Detector normalizes these to a 0-100 score and also scores a 4x4 grid of regions for the heatmap.
Is it safe to check photos for blur online?
Only if the analysis happens locally. Upload-based tools send your photos to a server, which matters for client work and personal images. The Prescosoft Blur Detector analyzes every photo on a canvas in your own browser — nothing is uploaded, stored, or transmitted, and it works offline after the page loads.
What is a good sharpness score?
Scores are normalized from 0 (heavily blurred) to 100 (very sharp). Under standard sensitivity, below 25 is Blurry, 25-50 is Soft, and above 50 is Sharp. A photo can be 'Sharp' overall yet have a soft region — which is exactly what the region heatmap is for, since depth of field often leaves the edges of a frame softer than the subject.
Can blur detection be wrong?
A detector measures detail, not intent. A deliberately soft portrait or a long-exposure shot of water will score low even though it's exactly what the photographer wanted, and a flat, out-of-focus background scores low too — which is why the tool marks low-detail regions as neutral rather than calling them blurry. Use the score as a screening tool, and check the heatmap before judging a photo.
How do I avoid blurry photos in the future?
Match shutter speed to your focal length (roughly 1/focal length as a minimum), use image stabilization or a tripod for slow shutter speeds, pre-focus and use back-button focus for moving subjects, and shoot burst mode so at least one frame in a sequence is sharp. For events and run-and-gun work, a fast aperture and a shutter speed of 1/250s or faster avoids most motion blur.
Find your blurry photos in seconds — privately
Whole-folder scans, region heatmaps, and side-by-side comparisons. Your photos never leave your device.