Yes. You can merge multiple image formats together with existing PDF documents in one operation. There is no need to convert your images into PDFs before uploading them.

Yes — and precisely, not just 'as best as possible'. JPEGs that need no changes are embedded byte-for-byte with an identical hash to the original; anything that must be re-encoded uses quality 90 with full colour resolution preserved. See How the Merge Preserves Your Files below for the exact handling of every file type.

Yes. Simply drag and drop your files into the preferred order before creating the PDF. The generated document follows exactly the sequence you choose.

If a PDF is password protected, you can provide the correct password during processing. Once unlocked, its pages can be merged with your other documents and images.

Using one tool saves time and simplifies the process. Instead of converting every image into a separate PDF and then merging those files, you can upload everything at once and create a single PDF in one step.

If a JPEG has no orientation tag to correct, no transparency to flatten and no rotation you asked for, the original file is embedded exactly as it is — not decoded, not re-encoded, not re-compressed. A second generation of JPEG loss is the most common quality complaint with merge tools, and this path avoids it entirely.

Verified on a 509,185-byte photograph: the stream embedded in the finished PDF was 509,185 bytes with an identical SHA-1 hash to the file on disk.

Flat artwork is detected by counting distinct colours rather than guessing from the file extension, so a photograph saved as PNG still goes to JPEG and a chart saved as JPEG still gets lossless treatment. When re-encoding is unavoidable the default quality of 90 keeps chroma subsampling switched off, which preserves full colour resolution on thin coloured lines and small text — the detail that usually smears first.

The order you arrange files in is the order they are processed in, and it is carried through as an explicit sequence rather than inferred from filenames, upload timing, or whatever order the network happened to return them in. Where an explicit position is supplied it takes priority, and files with the same position keep their relative arrangement — a stable sort, so nothing shuffles unpredictably between two runs of the same list.

Every page is tracked as it lands. The finished document reports which source file produced which pages and the exact page number each file starts on, so a 50-file merge can be audited page by page rather than trusted.

Each source file becomes a top-level bookmark. If a PDF already carried its own outline, those entries are re-nested one level underneath it with every page reference recalculated to its new position in the combined document. Bookmarks pointing outside the pages that file actually contributed are dropped rather than left pointing at the wrong page.

Orientation is handled in a fixed order, and the order is what decides whether the result looks right:

• The EXIF orientation tag is read and applied first. A phone photo is normally stored landscape with a tag telling the viewer to turn it. Ignoring that tag is the single most common reason merge tools produce sideways pages.
• Your own rotation is then added on top, relative to what the file already showed — so a page already at 90° rotated another 90° ends at 180°, which is what you saw on screen.

For images the pixels are transposed — an exact index move, with no interpolation and no resampling. Verified: rotating an image through four 90° turns returned pixel data exactly identical to the original, with dimensions preserved. For PDFs, rotation is written as a page attribute instead, so the page box always matches what is drawn and printing and thumbnails behave. Verified: a page already at 90°, rotated another 90°, came out at exactly 180°.

Your original files are never modified. Every rotation applies to the copy inside the merged document, which is what makes the operation repeatable.

In automatic mode the page is the image — same proportions, no white bars. The physical size comes from the image's own DPI, which is the only honest source: a 300 DPI scan of a letter should come out letter-sized, not screen-sized.

• DPI is only trusted inside a sane band (roughly 36–2400). Outside it, the value is metadata noise and 96 DPI is assumed instead — what browsers and word processors use.
• The result is then clamped so the longest side lands between 2 inches and A4's long edge. Without an upper bound, a 6000-pixel photo carrying no DPI metadata becomes a 60-inch page no printer will accept; without a lower bound, a 32-pixel icon becomes a stamp.

Choose a fixed size instead — A3, A4, A5, Letter or Legal — and each image is centred on that sheet, with the sheet turned to match the picture's own orientation so a landscape photo is not letterboxed into a portrait page. You can fit the whole image inside the sheet, fill the sheet and let the edges crop, or stretch to fill, and add a margin measured in points (72 points = 1 inch).

After the merge is written, the finished file is re-opened and verified: the page count must match what was assembled, and the bookmark tree must be readable. A file that fails that check is reported as an error rather than handed over. Damaged PDFs that could only be read by repairing them on the way in are flagged in the result, so you are told when content might be missing instead of finding out later.

Large scans are supported up to 300 megapixels per image, and truncated or partially uploaded images are salvaged where enough data decoded, rather than failing the whole job over one incomplete file.