Case Study · Failure Analysis
Blurry Photo LEGO Mosaic Failure: Why Low-Quality Input Fails at 48x48
This is a documented failure case — a blurry, low-resolution photo that produces an unrecognizable mosaic output. The goal is to help you identify these warning signs before you order 2,304 pieces.
A blurry or low-resolution photo fails at 48x48 because color quantization can only work with the tonal information that exists in the source — and a blurry photo has almost none. The output is a mosaic of undifferentiated color blobs with no recognizable facial structure. The fix is simple: use a sharper photo. Never order parts based on a BMBrick preview that doesn't clearly show the subject you intend to build.
Quick Facts
| Photo type | Blurry portrait — motion blur + low resolution |
|---|---|
| Photo resolution | Approximately 280x320 pixels at source |
| Mosaic size | 48x48 — would use 2,304 pieces |
| Preview result | Unrecognizable — subject not identifiable |
| Parts ordered? | No — failure detected at preview stage |
| Money lost | $0 — the preview system prevented wasted spend |
| Resolution | Found a sharper photo of the same subject |
Original Photo Analysis
Subject Clarity
Severe motion blur — the subject was moving during the shot and the camera was set to a slow shutter speed. Eyes, nose, and mouth are completely indistinguishable as separate shapes.
Background Complexity
Background complexity is irrelevant when the subject itself is unresolvable. No amount of background removal improves a photo where the face has no sharp edges.
Lighting Quality
Indoor mixed lighting — part tungsten, part flash. Even with good lighting, the motion blur would have caused failure. Poor lighting compounds the problem further.
Contrast & Tonal Range
Blur eliminates edge contrast — the photographic mechanism that creates tonal planes in a portrait. Without sharp edges there are no tonal transitions for the color quantizer to work with.
Crop Suitability
Crop was reasonable — face-first with appropriate framing. The crop decision is irrelevant when the face content is unrecoverable. A good crop cannot save a blurry photo.
What Was Attempted
Attempt 1: Standard generation
Generated at 48x48 with default settings. Preview showed a mosaic of vague color patches with no recognizable face. The blur had averaged all facial features into a uniform mid-tone blob.
Attempt 2: Background removal
Applied Magic Cut hoping to improve subject contrast. Magic Cut isolated a blurry blob — because there were no sharp edges for the tool to detect, the subject boundary was poorly defined. No improvement.
Attempt 3: Higher color count
Increased palette to 28 colors hoping more colors would resolve detail. This added color noise to the already unrecognizable output — more colors on a blurry source only creates more confusion, not more clarity.
Resolution: Better photo found
Located a sharper photo of the same subject — a well-lit portrait taken with a stationary camera. That photo generated a clean, recognizable 48x48 mosaic on the first attempt. The mosaic was built successfully.
Result Gallery
Comparative analysis: Blurry input vs. Sharp input at the same 48×48 resolution (Authentic BMBrick Pipeline Output).
Blurry source photo — motion blur, ~280×320px
BMBrick preview — unrecognizable color patches
Sharp replacement photo of the same subject (Pipeline-verified)
Successful preview with sharp replacement photo (Authentic Output)
Always review the BMBrick preview before ordering parts — if the preview is unrecognizable, the build will be too, and no settings adjustment can fix a blurry or low-resolution source photo. The fix is always to find a better photo, not to adjust the tool settings.