Examples, trade-offs and checks for real tasks—not promises of perfect results.
Extract, rearrange or place several PDF pages on one sheet?
These operations solve different problems. Extracting changes which pages are in a document; reordering changes their sequence; pages-per-sheet changes how they are laid out for printing. Choose the operation before changing the file.
Extract PDF Pages: create a new document containing selected pages. For a six-page report where you only need pages 2 and 5, extract those pages; the output should have two pages.
Reorder PDF Pages: keep pages but change their order. After moving a cover to the front, verify the cover is first and that no required page is missing.
PDF Pages Per Sheet: put several source pages on a larger output page. Four source pages in a 2×2 layout should occupy one output sheet. Text will appear smaller when printed at the same paper size.
Check the result before printing
Open the downloaded PDF and compare page count, sequence, orientation and legibility with your original. Pages-per-sheet is not booklet imposition: a folded booklet needs a special page order. Use PDF Booklet Maker for that job and test a short print first. Password-protected PDFs may be rejected; obtain an authorised unprotected copy rather than expecting the tool to bypass protection.
When text extraction is the wrong conversion
PDF to Text reads an existing text layer. A scanned image of a page may have no text layer; this tool does not perform OCR. Text-only Word/PDF conversions also lose fonts, images and layout. For a faithful document export, use the original document application's PDF export instead.
Clean copied text without destroying its structure
Start with one small sample and keep an untouched copy. A cleanup that is useful for prose can damage code indentation, poetry or table columns.
Use Invisible Character Detector if a word looks correct but search or matching fails. Inspect the reported code points before removing them.
Use Normalize Unicode Spaces for non-breaking spaces copied from documents. “Price: 10 EUR” becomes “Price: 10 EUR”. Intentional no-wrap spaces will also be lost.
Use Smart Quotes to Straight Quotes before placing text into code that requires plain quote characters. This changes quote characters, not the rest of the punctuation.
Compare filters instead of deleting blindly
To review a log, Keep Lines Containing preserves matching lines; Remove Lines Containing does the opposite. With phrase “error” and case sensitivity off, “ERROR timeout” matches but “INFO ready” does not. A substring filter also matches “errors”; use a regex with word boundaries when you need a whole word.
Check line breaks and punctuation after every step. Removing invisible joiners can split combined emoji, and removing line breaks can merge unrelated paragraphs. An empty filter result can be legitimate: it means no lines matched, not that an invisible file is waiting to download.
What browser AI can and cannot do for your images
The two model-based image tools download their models from Hugging Face and process the selected image in your browser. The first run requires a network download and can be slower than later runs. If the model host is unavailable or your browser runs out of memory, processing may fail; these are not offline-ready on a first visit.
Portrait cutout versus general object removal
AI Background Remover uses MODNet portrait matting. Choose a portrait or selfie, not a product shot expecting reliable object isolation. Inspect hair, glasses and clothing edges on both light and dark backgrounds. The PNG supports transparency, but a PNG extension alone does not prove that the cutout is clean.
2× enlargement versus recovered detail
AI Image Upscaler uses Swin2SR at 2×. A 120×150 image should produce a 240×300 PNG. Input is limited to 800 pixels on its longest side. Compare fine text and edges at the same display size: generated detail can look sharper without being faithful to the original. Upscaling cannot reveal a true detail that the source never captured.
Not every text helper is AI
The Text Summarizer selects existing sentences using word-frequency scoring. Template title generators and rule-based rewriters do not understand the subject, browse sources or fact-check. Review their output as a starting point, not an authoritative answer.