Image Files to PDF 1.5.4

[1.5.4] - 2026-10-05

Changed

• Starting with 1.5.4, first-party project files are licensed under Mozilla
  Public License 2.0 (MPL-2.0). Earlier published versions remain under MIT;
  dependencies retain their own licenses. Distributors must preserve notices
  and make covered source and modifications available as required by MPL 2.0.
• The standalone JXA header includes the MPL notice and a link to the source
  for its version. Releases include the full LICENSE as an attested asset.
• Breaking: pdfcpu 0.16.1 or later is required. Older releases and prerelease
  builds are unsupported; run brew update and brew upgrade pdfcpu to upgrade.
• Breaking: the maximum resolution is now 1017 DPI, reduced from 1041,
  to fit pdfcpu's stateless 100,000,000-pixel image limit. Settings above the
  new ceiling must be lowered before conversion.
• PDF creation, page counting and strict validation use pdfcpu's stateless
  --conf disable mode. Personal pdfcpu settings, fonts and certificates are
  no longer loaded.

Fixed

• An incompatible personal pdfcpu configuration no longer produces a false
  "installed but too old" setup error. Setup checks the supported release floor
  without loading configuration and includes an upgrade command when needed.
• Setup recognizes Homebrew in its standard installation locations when
  Shortcuts provides a minimal PATH.

Image Files to PDF 1.5.3

[1.5.3] - 2026-09-27

Changed

• The output choice now reads One PDF — one image per page or Separate
  PDFs — one page per image. It is the first setting. Every image still gets
  its own page, and the choice never changes that.
• The settings window says how many images you selected, or how many were
  found in the folders you selected. It also says what each choice would
  create, in files and pages, where the PDFs will be saved, and the page order.
  Its button is now Create.
• Items in your selection that cannot be converted are listed before the
  settings, and you can cancel there instead of finding out afterwards.
• When the stepwise dialogs stand in for the settings window, they end by asking
  you to confirm the settings before anything is created, and they keep the
  answers you had already corrected in the window. Settings are remembered only
  once you confirm them.
• The completion message counts PDFs and pages actually saved, says "No PDF was
  created." when none was, and lists images that failed to convert separately
  from selected items that were not included.
• Every setting in the settings window has an accessible name and description.

Fixed

• A PDF missing pages could be published when all its pages were added in one
  step, which includes every single-image PDF. The page count of every PDF is
  now checked before it is saved, and a count that cannot be read fails the
  conversion.
• Malformed output from vipsheader, such as 2junk for a size or an
  orientation outside 1–8, now fails the image. Previously it could be taken
  as a number and place the image wrongly on the page.

Image Files to PDF 1.5.2

[1.5.2] - 2026-09-19

Added

• Instructions for removing the action. INSTALL.txt now ends with how to take
  it off your Mac and what it leaves there: one preferences entry, holding the
  page setup of your last run.

Internal

• Property-based tests for file ordering and byte-length handling, and a
  revised quality record.

Image Files to PDF 1.5.1

[1.5.1] - 2026-09-14

Fixed

• No more stray text files beside your PDFs. The action handed the list of
  PDFs it made back to Shortcuts, which wrote each path out as a file of its
  own — :Users:you:Pictures:output_20260907_010101.txt and one for every
  other PDF in the run. It hands back nothing now; the PDFs and the completion
  message are the whole of what a run produces. If you have collected some,
  they are safe to delete.

Image Files to PDF 1.5.0

[1.5.0] - 2026-09-13

Added

• A panel that shows what the action is doing while it does it: the file being
  prepared, the count of images, a bar, and then creating, validating and
  saving the PDF. It floats above the window you started from, never takes
  focus, and is gone before the message that says what was produced.
• The tool check and the search for images are now reported too. A selection
  of folders can take a while to read, and it used to be the longest a run
  could go without saying anything.

Changed

• A run that finishes in a moment puts no window up at all, and neither does a
  quick conversion after a slow answer to the settings — there is no point
  flashing one up on the way past. The panel also steps out of the way while
  the settings are being asked.
• Progress is still written to JavaScript for Automation's own progress
  object, which Script Editor, a script applet and the system script menu
  display in their own way. It is no longer the only place it is written: a
  Shortcut displays none of it, so a conversion said nothing for its entire
  length.
• The panel changes nothing for a headless run: nothing is displayed and
  nothing is built. Other changes below do affect one, and say so.

Fixed

• A damaged image is now refused instead of being half-converted. A photograph
  cut short — an interrupted download, a failing card — used to be salvaged
  into whatever could be read from it, and that partial picture was laid out,
  written into a PDF and reported as a finished conversion, with nothing said.
  It is now reported as a failure, by name. In Separate PDFs mode the rest of
  the selection still converts.
• A request to stop the run is now acted on. Pressing Stop — in a host that
  offers one; a Shortcut does not — used to be discarded, and the conversion
  ran to the end, or reported a file as missing that had never been written.
  It now takes effect at each point where the action says what it is about to
  do next, so nothing further is written into your folder after you stop and
  no PDF is ever left half saved.
• A stopped run says what it produced. The PDFs already saved are listed,
  including one that was saved at the very moment you stopped; a stop that
  arrived before anything had been made says nothing at all, as a cancellation
  always has; a headless run that was stopped part way reports that it did not
  do everything it was asked; and a run that had in fact finished every image
  by the time the stop arrived is reported as the complete run it was.
• A failure now names the file once. Every per-image failure used to read
  "photo.jpg: photo.jpg: ..." in the message and in the headless receipt.
• A timestamp in a headless configuration file must now be YYYYMMDD_HHMMSS.
  Anything else is refused before a single image is converted, and says what
  is expected. A value containing a / used to be accepted and put the PDF in
  a folder nobody asked for.
• A file URL the action cannot read is now reported as something that could
  not be converted, rather than guessed at. One naming a host other than this
  Mac used to be read as a path relative to wherever the action happened to be
  running, and one whose escapes are malformed used to be read as a literal
  filename — so a malformed address and a correctly written one for a file
  whose name really contains %20 meant the same thing, and the wrong
  photograph could be picked silently.
• Separate PDFs mode no longer keeps every page it has made until the run is
  over, including pages left behind by the check that rejected them. A long
  run used temporary space in proportion to the number of images converted; it
  now uses it for one image at a time.
• In Separate PDFs mode, an image that could not be converted now counts
  towards the progress. Only a saved PDF used to move the count, so three
  images with the second failing stopped at two of three, three failures
  stopped at none of three, and the file counter beside it went on to "3 of 3"
  regardless.
• The later stages of a combined run no longer report themselves beside a
  filename they have nothing to do with. "Creating PDF" now says how many
  images were prepared, instead of naming whichever one happened to be last.
• The headless failure message no longer says "0 not converted" for a run that
  left images unconverted. That count is of files refused before conversion
  began, and it is now named as such.

Image Files to PDF 1.4.0

[1.4.0] - 2026-09-09

Added

• The settings you chose last time are what the next run opens on — all six of
  them, not just some. Every value is shown before anything is converted, so
  nothing is applied that you have not seen, and changing one is the same as
  it ever was.
• If the settings are asked one question at a time, they are remembered there
  too, including a colour of your own: the list opens on "Custom colour..."
  with your colour already in the prompt behind it.

Changed

• Headless runs are unaffected in both directions. A configuration file says
  everything about the run it describes, so it neither reads what you chose in
  a window nor changes it.

Fixed

• A headless configuration file that is valid JSON but is not a set of
  settings — null, a number, a string, a list — now says so, instead of
  failing later with a message about the failure rather than about the file.

Image Files to PDF 1.3.0

[1.3.0] - 2026-09-09

Added

• The page background can be any colour, not only the four that were offered.
  The control is a list you can also type into: pick one of the presets, or
  write a colour of your own. #C7DAE8, c7dae8 and C7DAE8 are all
  accepted, and six hexadecimal digits is the whole rule; colours with
  transparency are not supported. The control says or type #RRGGBB beside
  itself, and a value that is not a colour is reported with everything else
  that needs correcting, with your other answers left in place.
• Headless configurations accept the same colours, written the same way. The
  four existing values keep working exactly as before.

Changed

• The background control shows colours rather than names. Its list holds
  #FFFFFF, #000000, #8E79E0 and #204486, and the field shows the one
  in force, with or type #RRGGBB beside it; the names are above the form
  instead. Showing "White (#FFFFFF)"
  made a control you can type into look like one you can only choose from, and
  editing what it put there was refused even when what you had written was a
  perfectly good colour.
• The colour swatches beside each preset are gone with the names. A control
  you can type into cannot carry them, and a swatch that stayed put while you
  typed a different colour would be showing you the wrong one.
• Where the settings are asked one question at a time — the fallback used when
  the form cannot be shown — the background list ends with "Custom colour...",
  which asks for the colour you want.

Fixed

• A value typed into the form and submitted without first leaving the field is
  no longer read as the value it replaced. This affected the resolution and
  quality fields as well.
• Where the settings are asked one question at a time, a rejected answer comes
  back for you to correct instead of being replaced by the default, which
  looked as though it had been accepted. The reason is shown in the dialog
  that asks again rather than in a separate one before it. This covers the
  resolution and quality prompts as well as a colour of your own.

Image Files to PDF 1.2.7

[1.2.7] - 2026-09-08

Fixed

• A save that fails no longer guesses why. It says what could not be done and
  then quotes the system's own explanation, so a folder that denies permission
  and a drive that cannot do the job are told apart. Both used to produce the
  same sentence, blaming the drive — including on a disk that was perfectly
  capable and a folder that simply would not accept a new file.
• Saving into a folder on another drive now works in cases where it used to
  fail outright. The PDF is put onto its name there by the ordinary method
  first, so a drive that supports it no longer depends on a system feature
  that some hosts do not provide.

Image Files to PDF 1.2.6

[1.2.6] - 2026-09-08

Changed

• Saving into a folder on an exFAT drive now stops rather than working around
  what the drive cannot do. Nothing is written to the drive, the message says
  the drive cannot take the output name in one step, and the finished PDF is
  kept — the message says where. Putting the name on such a drive meant
  creating it and then filling it, and a name that exists before your document
  is in it can be left empty by an interrupted run, or taken by something else
  in between. Some large camera cards are formatted this way; every other
  drive, including FAT-formatted cards, is unaffected.
• A failed save reads in the order you need it: what went wrong in plain words
  first, and the system's own message after it.

Image Files to PDF 1.2.5

[1.2.5] - 2026-09-08

Fixed

• Saving to a card or drive formatted for cameras is now a single step on most
  of them, rather than making the name and then filling it. There is no longer
  a moment in which your document's name exists with nothing in it, and a run
  that is interrupted cannot leave an empty PDF behind. FAT-formatted cards
  take this path; the older way remains only for exFAT, which cannot do it.
• On the drives that still need the older way, a file that appears at the name
  at the wrong moment is no longer deleted during cleanup: only the file this
  action itself created is removed.