You want to convert a PNG to JPG. You search for an online tool. The first result asks you to create an account. The second wants you to upload the file to their servers. The third does both, and also wants your email address for a free trial that will auto-convert to a paid subscription.
I kept running into this. I work with images constantly — in website design, in narrative formation for public affairs campaigns, in social media audience activation. Every time I needed a simple operation — a format conversion, a social media crop, stripping location data before sharing a photo — the available tools wanted my files on their servers first. In 2026, that means handing your images to companies whose incentives around that data are, at best, unclear.
So I built ImageNurse — 120 image tools that run entirely in the browser — to prove that none of it is necessary.
What accounts are actually for#
When an image tool requires an account, it is not because the operation needs one. Converting a PNG to JPG is a stateless transformation — image data goes in, different image data comes out. There is no user state to persist, no history to maintain, no preferences that require server-side storage.
Accounts exist for the business, not the user. They enable usage tracking, subscription gating, marketing email capture, and the retention metrics that drive valuation. The account is the product — the image tool is the bait.
Canva, Adobe Express, and similar platforms have built large businesses on this model. They require accounts because their revenue depends on converting free users into subscribers, and accounts are the mechanism for that conversion. The image processing is the hook. The subscription is the business.
Why server-side image uploads exist#
The same logic applies to file uploads. When you upload an image to a server for processing, the server is doing work that your browser could handle directly. I built ImageNurse on the Canvas API for pixel manipulation and WebAssembly for high-performance encoding — the same browser capabilities that let me resize, crop, compress, convert, and batch process images locally at near-native speed.
Server-side processing exists because it creates a control point. If the server does the work, the provider can meter usage, enforce limits, and push users toward paid tiers. If the browser does the work, there is nothing to meter.
This is the same dynamic that drives server-side PDF tools — and the same reason I built both PDF Pony and ImageNurse to process everything locally.
How image uploads expose your private metadata#
Image files carry more information than most users realize. EXIF metadata can include GPS coordinates, camera serial numbers, timestamps, and device identifiers. A photo taken on your phone and uploaded to an image conversion tool sends all of that metadata to the server along with the image itself.
I built the EXIF viewer in ImageNurse so users can see exactly what metadata their images carry. The remove metadata tool I built strips it. The batch strip metadata tool does the same across an entire folder. None of these operations require uploading anything — I designed them to read and modify metadata locally in the browser.
For professional use cases — marketing teams processing brand assets, HR departments handling employee photos, medical offices preparing patient images — uploading files to third-party servers introduces compliance risk that does not need to exist.
Why uploading images is riskier in 2026 than ever before#
The privacy stakes around image uploads have changed. It is not just about metadata anymore. AI companies need vast quantities of image data to train their models, and the terms of service on free image tools often grant broad rights over uploaded content. Governments are deeply invested in the outcomes of these massive AI players and the surveillance capabilities they enable. Bad actors can use image training data to generate convincing deepfakes and power scams.
When you upload a photo to a free image conversion tool, you have no meaningful control over what happens to that image after it leaves your device. The tool’s privacy policy might say the right things today and change tomorrow. The company might be acquired. The servers might be breached. The only way to guarantee that your images stay private is to never upload them in the first place.
This is the architectural decision at the core of ImageNurse — and the reason I built PDF Pony and GIS Tools on the same principle. If the browser can do the work, the server should not touch the data.
What I built to run entirely in the browser#
The gap between server-side and client-side image processing has effectively closed for the operations most people need. I built ImageNurse to demonstrate this. Every one of these capabilities runs locally:
- Convert between eleven image formats including HEIC, AVIF, WebP, SVG, PSD, and TIFF
- Compress to exact file size targets with quality preview
- Generate social media assets for every major platform in one operation
- Remove backgrounds using AI models running client-side
- Detect faces and crop intelligently
- Batch process entire folders with consistent settings
- Generate favicons, app icons, sprite sheets, and responsive image sets
- Create animated GIFs from image sequences
- Work offline after the initial page load
All 120 tools run in a standard browser tab. No plugins, no extensions, no desktop software.
Why client-side architecture eliminates the need for subscriptions#
Client-side image tools cost almost nothing to operate. The browser does the computation. The server delivers static files. There are no GPU instances processing uploads, no storage costs for temporary files, no bandwidth bills for transferring images back and forth.
This is why I built ImageNurse without accounts, subscriptions, or upload-based processing. The architecture I chose eliminates the costs that force other tools into those models. The result is 120 tools that are genuinely free — not free-with-limits, not free-with-watermarks, not free-for-now.
ImageNurse processes everything locally, works offline, and is available in the thirteen languages I internationalized it into.
