Writing
Mostly measurements. When we say a tool uploads your file, or that ours does not, there is a run behind it: the file we used, the date, and what the network showed. The method is published every time so you can re-run it against any tool, this one included.
Method ·
Chrome's built-in translator, measured
Chrome translates on your device with no server. What shipping it taught us: the click it demands, the tags it breaks, the words it splits, and offline.
Comparison ·
A free remove.bg alternative, and what changes on 1 December
remove.bg is folding into Canva, and its API moves to Leonardo.Ai on 1 December 2026. What changes, and a background remover that runs in your browser.
Method ·
The remove.bg API kept nothing. Its replacement keeps your image
remove.bg is moving to Canva and its API to Leonardo.Ai. One row in the migration table takes result storage from nothing kept to kept by default.
Method ·
MP3 vs WAV: we encoded the same audio six ways and weighed it
WAV 5.3 MB, FLAC 1.8, MP3 320 at 1.2, Opus 96 at 0.26. What lossy discards, measured sample by sample, and which format to keep.
Method ·
CSV vs Excel: which is smaller depends on how many rows
We wrote the same table both ways. XLSX was 15x larger at 4 rows and 38% smaller at 10,000. What each format keeps, and what Excel guesses.
Method ·
What is DPI, and why changing it does nothing to your image
We saved one image at 72, 300 and 1200 DPI. All three were the same bytes and not one pixel differed. What DPI is and what to send instead.
Method ·
What is a codec, and why .mov to .mp4 can change nothing at all
A codec compresses the video. A container is the box around it. We put one H.264 stream in three containers and the video bytes were identical.
Method ·
A pdf.js render stops when the tab is hidden
pdf.js paints on requestAnimationFrame, which a hidden tab never fires. No error, no rejection, no timeout. Measured, with the line of source that does it.
Method ·
Favicon size: the answer is six numbers, not one
16 decides whether your mark reads at all. The rest are Windows, bookmarks and shortcuts. What each size is for, and why one file holds them.
Method ·
JPG vs JPEG: the same file, and one of them is a 1980s filename limit
The two extensions are one format. We diffed both: byte identical, same type, same header. What the difference actually was, and when it still bites.
Method ·
What is a TIFF file, and why will nothing open yours?
TIFF is a container, not one format. The same image in it ran 15 KB to 2.3 MB in our test. That range is why yours opens in one program and not another.
Method ·
PNG vs JPG: the answer flips depending on the picture
We weighed the same image both ways. JPG was 5x smaller on a photo, PNG was 1.5x smaller on a graphic. What is in the picture decides.
Method ·
What is a CSV file, and why does yours open wrong?
A CSV is a text file of rows and columns. The comma is the least reliable part of it. What the format really allows, and why Excel mangles your data.
Method ·
Six characters that freeze a regex tester
(a+)+b runs inside one call the page cannot interrupt. Measured: 5ms at 19 characters, 4.7 seconds at 28. Only a worker can be stopped.
Method ·
One flipped module cannot fake a barcode. Two can.
A single damaged bar falls outside the pattern tables and reads as nothing. Two make a barcode that scans cleanly as the wrong product. Measured.
Method ·
Chrome can play an MP3 and cannot make one
WebCodecs encodes AAC and Opus and refuses MP3. Why the browser has half the codec, and what it takes to put the other half back without breaking the LGPL.
Method ·
Half your password’s characters are likelier than the rest
random % alphabet.length makes 76 of 90 characters 50% more likely. The entropy cost is tiny, the fix is four lines, and what it tells you is worth more.
Method ·
WhatsApp video size limit, and how to get under it
WhatsApp caps media at 16 MB and documents at 2 GB. Why a phone video passes that in seconds, and why compressing first arrives looking better.
Method ·
Chrome takes a full second to encode one canvas image
Extracting 9 frames from a video took longer than extracting 174. The cause is a 1000ms idle task timeout in Blink, and the fix is a worker.
Method ·
A diff tool should limit work, not lines
A line cap refuses a 20,000 line log that diffs in 55ms and still allows a 5,000 line case that takes a second. The budget belongs on the work.
Method ·
Sentence case is where text tools destroy your acronyms
Lowercase everything, then capitalise after each full stop. That turns The NASA API returns JSON into the nasa api returns json. Almost every tool does it.
Method ·
Privacy at filetity
No account, no cookies, and your files are never uploaded. What the three analytics scripts do record, what is kept in your browser, and how to check.
Method ·
HEIC to JPG Windows, without installing iCloud
Windows shows a HEIC as blank or asks you to buy a codec. A browser decodes it instead: one 347 KB download, kept after that, and the photo stays on the PC.
Method ·
Compress image to 100KB for an upload form
A phone photo is several megabytes because of its pixel count, and a form wants 100 KB. What compression throws away, and what to do when it will not fit.
Method ·
Merge PDF offline, with no connection at all
Merge PDF works with the network off, once you have merged on that device before. What the browser keeps, what it does not, and how to prepare on purpose.
Method ·
Compress PDF to 1MB, 500KB or 200KB
Three PDFs, four size targets, and the settings each one needed. Why resolution shrinks a file further than a quality slider does.
Method ·
How to check if PDF is password protected
Password protected covers everything from 40-bit RC4 to AES-256. The PDF says which it is in four fields. Here is how to read them, on any file.
Method ·
How to tell if a PDF is scanned or has a text layer
Four checks you can run in any PDF viewer to tell a scan from real text, what each result means, and which kind of tool each one needs next.
Method ·
Check EPUB for DRM before converting it
Four files inside the zip say whether an EPUB is locked and by which store. One of them looks like DRM and is not. How to check in about a minute.
Comparison ·
Best free HEIC to JPG converter, tested
The same iPhone photo through five free HEIC converters, with the network watched. Two sent it to a server, one before we pressed Convert.
Comparison ·
Do online PDF tools upload your file? We measured four
We sent the same 1,054-byte file through four PDF tools and watched the network. Three uploaded it. The method is included so you can re-run it.
Method ·
How filetity processes files locally
Which network requests happen on a filetity tool page, when, and how to verify for yourself that your file is not one of them.