How to Improve Scanned Photo Quality: Fix Scanner Artifacts the Right Way

Tools referenced in this guide: Photo Denoiser · Photo Deblurrer · JPEG Artifact Remover · Photo Enhancer · Old Photo Restoration · Photo Colorizer

Quick path: If your scanned photos have compression artifacts, grain, or fading, the ArtImageHub pipeline — JPEG artifact removal → denoising → restoration — costs $4.99 one-time, processes in under 90 seconds, and requires no photo editing experience.

Scanning old photos feels like it should be simple: place the photo on the glass, press the button, done. In practice, the output often looks disappointingly flat, grainy, or marked with strange lines and patterns that were never in the original print.

Most of these problems have predictable causes — and most of them are fixable, either by adjusting how you scan or by running the output through the right AI tools in the right order. This guide covers both.

What Are the Most Common Scanning Quality Problems?

Scanner Dust and Scratches

The scanner's glass surface is optically focused — meaning any particle sitting on it is in perfect, razor-sharp focus. A single hair or speck of dust that is invisible to the naked eye will render as a crisp bright line or blob across your photo. This is different from physical damage on the photo itself: the artifact is on the glass, not the original, so cleaning the glass removes it completely before scanning. If you discover it after the fact, ArtImageHub's photo denoiser can suppress many small dust artifacts, though large bright streaks may require the old photo restoration tool which is better equipped to reconstruct missing or overexposed regions.

JPEG Compression from Scanner Software

Most flatbed scanner software defaults to JPEG output, and many compress aggressively. JPEG compression works by grouping the image into 8x8 pixel blocks and discarding fine detail it judges unnecessary. On a photo of a face, this creates the familiar "blockiness" around transitions — especially visible in skin tones and sky gradients. The fix on the front end is switching your scanner output to TIFF. If the scan is already compressed, ArtImageHub's JPEG artifact remover uses SwinIR to detect and suppress the block grid pattern while preserving underlying photographic detail.

Color Shift from Scanner Lamp Aging

Scanner lamps age with use. As a lamp ages, its color temperature shifts — typically toward yellow or green tones. A scan made with a well-used lamp can show a distinct color cast across the entire image that was not present in the physical print. This is different from fading in the original photo: an aged lamp creates a uniform shift while fading from photo chemistry tends to cluster in specific tones (yellowing shadows, faded highlights). Running the output through old photo restoration handles both issues, since the model is trained on color-degraded photos and has learned to distinguish original tonality from lamp-induced drift.

Moire Patterns from Halftone Originals

Moire is the interference pattern that appears when you scan a photo that was printed using halftone dots — newspapers, magazines, photobooks, and some older commercial prints all use this technique. The scanner's sampling grid and the print's dot grid interact to create repeating wave-like patterns across the image. Enable the "descreening" option in your scanner's software before scanning halftone originals. If the scan is already done, the JPEG artifact remover followed by the denoiser can suppress the pattern because neural networks distinguish repetitive interference from real photographic texture.

Scanner Glass Dirt as Smudges

Unlike sharp dust particles, smudges from fingerprints or oily residue scatter light before it reaches the optical sensor. The result is a soft, cloudy area rather than a sharp line. Prevention is the right approach: wipe the glass with a microfiber cloth and a drop of lens cleaner before each session. If you already have smudge artifacts in a scan, the photo deblurrer can recover some sharpness in mildly affected areas, though the soft-light effect of a large smudge can exceed what any AI enhancement can recover cleanly.

How Do Scanner Resolution Settings Affect Final Quality?

DPI (dots per inch) controls how much of the original photo's physical detail the scanner captures.

| DPI | Use Case | File Size (4x6 photo) | |---|---|---| | 300 | Web sharing only | ~2–4 MB JPEG | | 600 | Standard preservation | ~40–60 MB TIFF | | 1200 | Small originals, fine detail | ~150–200 MB TIFF | | 2400+ | Film negatives, slides | 500 MB+ TIFF |

At 300 DPI, AI enhancement tools have limited source data to reconstruct from — the model cannot invent detail that was never captured. At 600 DPI, every tool in the ArtImageHub pipeline has enough pixel information to work effectively. Beyond 1200 DPI for standard prints, the law of diminishing returns applies: you are capturing grain in the paper rather than additional photographic detail.

What Is the Step-by-Step AI Enhancement Workflow for Scanned Photos?

The order of operations matters. Applying tools in the wrong sequence can amplify artifacts rather than suppress them.

Step 1 — Remove JPEG Artifacts First

If your scan has any JPEG compression in it (even at high quality), start with JPEG artifact removal. SwinIR's block-suppression model works best on unmodified compressed input — other enhancement tools applied first can reshape the block patterns in ways that make them harder to detect.

Step 2 — Denoise

After artifacts are removed, run photo denoising. NAFNet's denoising model targets random pixel noise — scanner grain, film grain transferred during scanning, and sensor noise from consumer scanner hardware. Denoising after artifact removal means the model is not mistaking JPEG blocks for genuine noise.

Step 3 — Old Photo Restoration

Run old photo restoration to address structural damage: scratches on the original print, fading, missing emulsion, and color cast. At this stage the image is clean enough that the restoration model can focus on genuine photo damage rather than scanning artifacts.

Step 4 — Upscale if Needed

If you need a larger print from a smaller original, the photo enhancer applies Real-ESRGAN upscaling with detail synthesis. Upscaling last — after all cleanup is done — prevents the upscaler from magnifying any residual artifacts.

Optional — Colorize

For black-and-white originals, add a colorization step using DDColor after restoration is complete. Colorizing on a clean, restored image produces significantly more accurate results than colorizing a noisy or artifact-heavy scan.

How Can You Prevent Scanner Problems Before They Happen?

Prevention eliminates most scanner artifact problems entirely:

• Clean the glass before every session. Use a microfiber cloth and lens cleaner, not paper towels (which scratch glass over time).
• Set output format to TIFF. This single change eliminates compression artifact problems entirely.
• Use the highest quality setting your scanner offers. Even when forced to use JPEG, setting quality to maximum significantly reduces block artifacts.
• Enable descreening for halftone originals. Any photo originally from a book, newspaper, or magazine needs this setting.
• Let the scanner warm up. Cold lamps have slightly different color output. Running a blank scan before your first photo takes 30 seconds and avoids a color shift on the first image in a batch.
• Handle originals by the edges. Fingerprints on the photo surface transfer to the glass and back to the photo during scanning.

A clean scan is always easier to work with than a problematic one — but if you are working from existing scans, the four-step AI workflow above addresses the most common problems systematically.