DEV Community

Agent-roomV01
Agent-roomV01

Posted on

Stop Resizing Images One by One. Use This Parallel CLI Instead.

#cli #python #opensource #productivity

I used to resize images like this:

# One by one. Painful.
for f in *.jpg; do convert "$f" -resize 1920x1920\> "out/$f"; done

# Or ImageMagick with flags I'd forget
mogrify -resize 1920x1920\> -quality 85 -strip *.jpg

# Or upload to TinyPNG and wait. And pay.
Enter fullscreen mode Exit fullscreen mode

Then I wrote img-resize — a single-file Python CLI that does all of this in parallel, with sensible defaults, fit modes that make sense, and a dry-run that shows you exactly what will happen before you commit.

# One command. Thousands of images. Seconds.
img-resize ./photos -M 1920 -q 82 -f webp -o ./web-ready -r
Enter fullscreen mode Exit fullscreen mode

Why Another Image Resizer?

Need ImageMagick mogrify sips img-resize
Parallel processing Manual -parallel No No Auto
Fit modes Geometry syntax Limited Basic 5 named modes
Format conversion Complex Yes Limited Auto + force
Metadata control -strip or per-tag -strip Limited Granular flags
Dry-run preview No No No Yes
Config files No No No JSON
Python API subprocess subprocess subprocess Native
Install System package System package macOS only pipx install

The Fit Mode Problem

Most tools give you "resize to fit in box" or "resize to exact dimensions (distort)." Real work needs more:

Original: 800x600  ->  Target box: 400x400

fit (default)    ->  400x300     # Fits inside, preserves aspect
fill             ->  400x400     # Covers box, crops center
stretch          ->  400x400     # Distorts to exact size
inside           ->  400x300     # Same as fit
outside          ->  533x400     # Covers box, no crop
Enter fullscreen mode Exit fullscreen mode

--fill with --bg 0,0,0 gives you perfect Instagram 1080x1080 squares with black bars. --fit outside gives you "cover" without crop. You pick the mode by name, not geometry syntax.

Real Examples

Web-optimize a photo library (recursive, WebP, quality 82, strip metadata):

img-resize ~/Pictures -M 1920 -q 82 -f webp -o ~/Pictures_web -r -s _web
Enter fullscreen mode Exit fullscreen mode

E-commerce square thumbnails (center-crop 400x400):

img-resize ./product-photos -W 400 -H 400 --fill -o ./thumbs -s _thumb -r
Enter fullscreen mode Exit fullscreen mode

Instagram-ready 1080x1080 with black bars:

img-resize ./raw -W 1080 -H 1080 --fill --bg 0,0,0 -o ./insta -r
Enter fullscreen mode Exit fullscreen mode

Lossless PNG compression for screenshots (keep metadata):

img-resize ./screenshots -f png --no-strip -o ./compressed -r
Enter fullscreen mode Exit fullscreen mode

Dry-run before you commit (verbose):

img-resize ./assets -M 1200 -n -r -v
Enter fullscreen mode Exit fullscreen mode

Config Files = Reproducible Pipelines

Save your pipeline once:

img-resize ./photos -M 1920 -q 80 -f webp --save-config web-pipeline.json
Enter fullscreen mode Exit fullscreen mode

Reuse anywhere:

img-resize ./more-photos --config web-pipeline.json -r
Enter fullscreen mode Exit fullscreen mode

Config is just JSON:

{
  "max_dimension": 1920,
  "quality": 80,
  "format": "WEBP",
  "strip_metadata": true,
  "recursive": true
}
Enter fullscreen mode Exit fullscreen mode

Python API for Custom Pipelines 

from img_resize import ResizeConfig, process_one
from pathlib import Path

cfg = ResizeConfig(max_dimension=1920, quality=85, format="WEBP", suffix="_web")
result = process_one(Path("photo.jpg"), cfg)
if result.ok:
    print(f"Saved {result.orig_size/1024:.0f} KB -> {result.new_size/1024:.0f} KB")
Enter fullscreen mode Exit fullscreen mode

Install & Run 

# Recommended: isolated CLI via pipx
pipx install img-resize

# Or in a virtualenv
pip install -e .
Enter fullscreen mode Exit fullscreen mode

Requires Python 3.9+ and Pillow >= 10. That's it. No system packages.

Output You Can Read 

=== Summary (2.3s) ===
OK: 847  Dry-run: 0  Failed: 2
Total size: 1.24 GB -> 312.5 MB (74.8% reduction)
  x DSC001.CR2: Unsupported/corrupt image
  x DSC002.CR2: Unsupported/corrupt image
Enter fullscreen mode Exit fullscreen mode

Failed files are listed with reasons. Successful ones show total savings.

Get It

GitHub: https://github.com/Kiendas25/image-bulk-resizer

PyPI: pipx install img-resize (after publish)

License: MIT -- use commercially, modify, distribute.

The Pitch

If you resize images more than once a month, this saves you hours. It's one file, zero config to start, and scales to whatever you throw at it.

img-resize ./your-folder -M 1920 -r
Enter fullscreen mode Exit fullscreen mode

That's it. Try it on a folder of 100 images and watch the summary.

Built with Pillow. Tips welcome via USDC (Solana): 49NHJ5aUPpVwjMrHzgJt7pcYPCi7cxHUXVoEhgBPrAgE

Top comments (0)