Image Pipeline
Strip the metadata from a batch of photos and shrink them for the web in one run.
What each step does
Strip metadata
JPEG and PNG files have their metadata blocks cut out without touching a pixel. GPS coordinates, capture time, camera model, and every other EXIF, IPTC, and XMP field are removed. Other formats, like WebP, are re-encoded once as high-quality WebP, which drops the metadata with them.
Compress
Each image is scaled so its long edge is at most 2200 pixels, then saved in its own format: JPEG and WebP at quality 0.78, PNG losslessly. If a file comes out bigger than it went in, the cleaned version from step 1 is kept instead.
Why the steps run in this order
- Strip first so the cleanup can’t be skipped. The compress step hands back its input whenever re-encoding would make a file bigger. Because that input is already the cleaned copy, every file you download is metadata-free.
- Compress last so there is only one lossy save. Step 1 leaves JPEG and PNG pixels untouched, so step 2’s resize and 0.78 save is the only time the image is re-encoded.
- Run both, even for photos you only want smaller. Phone photos carry the GPS location where they were taken, the time, and the phone model. A 4032-pixel-wide photo is also more than twice the width of a typical 1920-pixel desktop screen.
- Skip the compress step for print or for files someone will edit later. Download the cleaned files from step 1’s preview instead.
Worked example: Twelve product photos for a Shopify listing
You shot 12 photos of a jacket on an iPhone set to “Most Compatible”, so each one is a 4032 × 3024 JPG with your home’s GPS location inside.
- Drop all 12 onto the input and press Run all.
- Step 1 cuts the metadata out of each JPG without re-encoding it. The GPS location, capture time, and phone model are gone.
- Step 2 scales each photo to 2200 × 1650 and saves it at quality 0.78.
- Open the preview on each step to compare file sizes, then press Download zip to get image-pipeline-output.zip.
Result: Twelve 2200 × 1650 JPGs with their original file names and no location data, each under a third of the pixel count of the original. Your originals are untouched.
What it won’t do
- The long edge is capped, whichever way the photo faces. A portrait 3024 × 4032 photo comes out 1650 × 2200.
- There’s no quality slider. JPEG and WebP end at quality 0.78. PNGs stay lossless, so they mostly shrink by being resized.
- Formats are kept, so transparent PNG and WebP images stay transparent. Animated GIFs keep only their first frame. When a file’s format changes, its extension is changed to match.
- HEIC photos only load in a browser that can display HEIC, which today means Safari.
Tips for best results
- Keep your originals. Both steps throw away data you can’t get back: metadata in step 1 and pixels in step 2.
- Rename files before you drop them in. The output keeps the original names, so IMG_4471.jpg stays IMG_4471.jpg.
- Outside Safari, set your iPhone camera to “Most Compatible” or export photos as JPG before running a batch.
- Need exact sizes, like 2048 × 2048 squares? Crop and resize first with the Image Cropper and Image Resizer. This pipeline only caps the width.
Questions
What metadata is removed?
All of it: the EXIF, IPTC, and XMP blocks, which cover GPS coordinates, capture time, camera and lens model, serial numbers, and editing software. For JPEG and PNG the blocks are cut out and the pixels are left exactly as they were.
Why is one of my files the same size as before?
The compress step keeps its input when re-encoding didn’t make the file smaller. That happens with small PNG graphics and with JPEGs that were already heavily compressed. The file still went through step 1, so it’s clean.
How is this different from the Image Compressor and EXIF Data Remover tools?
It uses the same code. Step 1 is the EXIF Data Remover, and step 2 is the Image Compressor with the 2200-pixel resize switched on. The standalone tools let you skip the resize or convert to WebP; this pipeline runs both jobs on a whole batch in one go.
Will my photos look worse?
On a web page, rarely. Quality 0.78 with a 2200-pixel long edge is a common setting for photos on the web. Zoom in on fine texture like fabric or hair and you may spot softness, so keep the originals for print and editing.
Are my photos uploaded?
No. Both steps run in your browser, and the ZIP is built on your device too.