EVIDENCE, NOT PROMISES
Conversion behavior & testing.
Reviewed October 13, 2026
Our examples use synthetic data and files. Checks exercise the actual conversion engine and browser downloads; they do not prove compatibility with every document or device.
Repeat a conversion yourself
- Nested JSON to CSV: compare dotted columns, array text and missing values.
- CSV to JSON: verify leading zeros, long identifiers and quoted multiline text.
- PNG to JPG and WebP to JPG: download a transparent 480 × 300 sample and flatten it onto white at 90% quality.
- Merge PDF: combine labeled sample A and B, then check page labels, dimensions and order.
What our checks cover
| Area | Checks |
|---|---|
| Data | Quoted fields, Unicode, multiline CSV, leading zeros, unsafe integers, nested fields, XML entities, invalid inputs and published text examples. |
| Images | Output encoding, resized and rotated dimensions, flips, transparency, mixed inputs, failure retries and ZIP filenames. |
| Mixed image embedding, merging, individual and grouped splitting, page order, cancellation and ZIP contents. PDF previews, page image export, page organization, and watermark/number overlays on rotated, cropped and scaled pages. | |
| Browser | All 30 tools, download flows, JavaScript errors, and representative desktop and 390-pixel mobile layouts. |
| Network | During tested conversions, requests outside the tested site and non-GET requests are flagged. |
How the samples are produced
Text examples are checked against the same data engine used by the tools. The image fixture is a Canvas drawing encoded in Chromium; JPEG references use a white background at 90% quality. PDF fixtures are unsigned documents with explicit page labels. These are reproducible examples, not a cross-browser performance benchmark.
Known limits
- Chromium checks are not a Safari or Firefox compatibility certification. Byte sizes can differ between browser encoders.
- PDF forms, bookmarks, signatures and complex document features are not guaranteed to survive editing. No OCR is included.
- JSON parsing can normalize number spellings and duplicate keys. Preserve original bytes when signatures or exact representations matter.
- Browser processing still uses local CPU and memory. Large-file behavior depends on the device; PDF and ZIP workers can be terminated; an image already encoding may finish before its result is discarded.
- The network checks cover the exercised paths in this version. They are not an independent security audit.
We do not claim measured speed advantages over competitors or a guaranteed compression ratio. If an output matters, compare it with your source before relying on it.
Explore the guides ↗