What browser-based image tools can and cannot do
Understand local processing, device memory, large files, decoder support and the difference between downloaded code and uploaded images.
Local processing is useful, with real limits
Browser-based tools can decode a selected file, draw pixels to canvas and create a downloadable result on your device. That avoids sending ordinary source images to our application server and reduces server processing. It also means your phone or computer does the work. A 5000 × 4000 image has 20 million pixels; a single uncompressed RGBA buffer can use around 80 MB before previews, copies, history or encoder memory are counted. Several open files and effects can multiply that demand.
If a tab slows down, use fewer files at once, close other heavy tabs or resize a copy first. A browser may reject a very large canvas even when the file itself is small. File size in megabytes and decoded pixel count measure different things: a compressed 3 MB photograph can still require a large memory buffer.
Formats depend on the browser
JPG, PNG and WebP are broadly supported by current browsers. HEIC decoding may require an additional library and can be slower. Animated GIFs converted through canvas typically become a still frame; use the dedicated GIF maker when you need animation. A PDF tool needs a page renderer, and a PDF with unusual fonts, protection or damaged structure may not render correctly. Check what a particular tool actually says it accepts rather than relying on a file extension alone.
What 'no upload' means here
The selected image or PDF bytes for the image workspaces are not posted to Do The IMG for conversion. The page itself, libraries, fonts, ads (if enabled) and ordinary account or payment requests still involve network traffic. Some tools obtain JavaScript library code from third-party CDNs. Your browser extensions and operating system also influence privacy. Download the result and inspect it before posting it elsewhere; that next destination will have its own policies. For sensitive work, keep offline originals and verify metadata separately.