FormatCore

FormatCore is a Windows script for a very specific case: bringing down the weight of images and remaking PDFs without depending on external tools like PDF24 or QPDF.

With images it works as you would expect: it resizes, recompresses and, if needed, changes the format. With PDFs there is one important catch that should be clear from the start: it does no structural compression of the PDF. What it does is render every page as an image and assemble a new PDF from those images.

That has consequences. You can lose selectable text, search, vectors, layers and other elements of the original PDF. For scans it usually makes sense. For manuals, invoices, technical documentation or text-heavy PDFs, not always. Some may end up worse, or even larger.

The important part before using it on PDFs

• If you want to keep selectable or searchable text, this is not the best option.
• If the PDF is already well optimized, rasterizing it can make it worse.
• -PdfMode Auto tries to avoid those cases with a prior estimate.
• -PdfMode Skip ignores PDFs entirely.
• -PdfMode Raster forces the remake even when the estimate says it is not worth it.

How to use it

There are three normal ways:

• Drag files onto the .exe if you have it packaged.
• Open the .exe or the .ps1 with no arguments to get the file picker.
• Run it from a console passing paths.

Examples:

powershell -ExecutionPolicy Bypass -File .\FormatCore.ps1 "C:\Docs\report.pdf" "C:\Docs\photo.png"

powershell -ExecutionPolicy Bypass -File .\FormatCore.ps1 "C:\Inbox" -Recurse

powershell -ExecutionPolicy Bypass -File .\FormatCore.ps1 "C:\Docs\scan.pdf" -PdfMode Auto -LogPath "C:\Logs\formatcore.log"

Options that are actually useful

• -PdfMode Auto|Skip|Raster: decides what to do with PDFs.
• -Recurse: also goes into subfolders when you pass directories.
• -WhatIf: shows what it would do without writing anything.
• -LogPath: saves a transcript of the run.
• -OutputDirectory: sends all results to another folder.
• -Overwrite: reuses the same output name. Without it, FormatCore creates _001, _002, etc. suffixes.
• -KeepLargerOutput: keeps results that do not improve on the original size.
• -OutputFormat Auto|Original|Jpg|Png: output format for images.
• -KeepTransparency: if the image has transparency, avoids ending up in JPEG.
• -PDFSplitMarginPercent: makes PDF splitting more conservative or more aggressive.

How output names are decided

By default it never overwrites previous results. If a file like photo_REDUCED.jpg already exists, the next one will be photo_REDUCED_001.jpg, then photo_REDUCED_002.jpg, and so on.

If you prefer that it always replaces, use -Overwrite.

About image formats

FormatCore can work with jpg, jpeg, png, bmp, gif, tif, tiff, jfif and heic.

Two practical nuances:

• HEIC depends on the codec installed in Windows. If the system does not have it, the file will not open properly.
• If an image has transparency and the output would end up as JPEG, -KeepTransparency forces a compatible output, normally PNG.

About PDF

The PDF path relies on Windows.Data.Pdf, so it is meant for Windows PowerShell 5.1 on a compatible Windows.

The actual process:

1. The PDF is opened with Windows.Data.Pdf.
2. Every page is rasterized to a JPEG image.
3. A new PDF is rebuilt from those images.
4. If it still exceeds the threshold, it is split into several parts.

Rasterization has no fixed megabyte limit for the input PDF: it depends on available memory, the number of pages and the size of each rasterized page. In this implementation the rebuild keeps the PDF and the JPEG bytes in memory before writing it out, so the practical limit for a generated PDF is around 2 GB on Windows PowerShell 5.1. A PDF or a single page getting close to that size can fail on memory even with plenty of disk space; it is not considered a supported case. Threshold uses Int64 so sizes and split decisions do not overflow near 2 GB, but that does not raise the rasterization limits of System.Drawing or Windows.Data.Pdf.

PDFSplitMarginPercent does influence the cut. A lower value makes the script put fewer pages per part, leaving more headroom on the final size.

Keep in mind that final PDF quality depends heavily on the DPI and JPEG quality chosen for the rasterization.

The repository also carries a small regression check at tests/Verify-FileSizeTypes.ps1. It inspects the source and fails if the file-size handling ever goes back to Int32 — exactly the overflow this script hit once near the 2 GB mark. Run it after touching any size-related code:

powershell -ExecutionPolicy Bypass -File .\tests\Verify-FileSizeTypes.ps1

CI runs the same check on every push.

Default behavior

• No files passed: opens a file picker.
• Folders passed: processes direct files from the first level only.
• -Recurse: also goes into subfolders.
• Result is not smaller and -KeepLargerOutput unused: discarded.
• Output name exists and -Overwrite unused: creates a new name.

Requirements

• Windows
• Windows PowerShell 5.1
• System.Drawing
• Windows.Data.Pdf available on the system

If you package it as an .exe, the expected behavior stays the same: drag files onto it or open it with no arguments to pick them.