TIFF to Searchable PDF: Converting Fax and Archive Images

To get from TIFF to searchable PDF, group the TIFF images into documents, wrap each group in a PDF without recompressing the pictures, then run text recognition on each PDF. A TIFF holds page pictures and no text layer, so the searchable part is always added after the images are in a PDF.

Why do legal files still arrive as TIFF images?

TIFF reaches a legal file from three kinds of sender: a fax service or fax server that saves each transmission as an image, an older imaging or records system that stored scans in that format, and a document production delivered as page images with separate text.

Each sender leaves a different job on your desk.

When the PDFs come back, search one of them for a Bates number on the page, the producing party's name, and a deposition line if the set is testimony. Open the hit and compare it with the page image. Do the same for an affidavit in the proceeding. A protective order quote you can see and cannot find means that page never received text.

For the first and last rows, the cheapest fix is at the source. If the fax service or the scanner offers PDF output, switch it and the conversion disappears for every later file. The post on converting a fax to a searchable PDF covers received faxes one transmission at a time.

For productions, read the specification, agreement or cover letter that came with the set to see what format was called for. The post comparing TIFF and PDF for document production explains why specifications ask for page images. Before you make searchable PDF working copies of a production, read the protective order or agreement that governs it, and keep the production itself exactly as it was received either way. The post on OCR for documents under a protective order covers what such an order may say about outside tools.

Single-page and multi-page TIFF: what you hold before converting

Before converting, find out whether each TIFF file is one page or one document. A single-page set needs grouping into documents first. A multi-page file is already grouped and can go straight to a PDF. The two cases share a file extension.

The post on what a TIFF file is explains the two layouts as a format. For conversion, three facts about your set matter.

Take an inventory

1. Count the files. Note the number in each folder.
2. Count the pages. Open several files in a viewer that shows a page count. One page in every file you open suggests a single-page set; test a few more from other folders before you believe it.
3. Look for companions. Text files with matching names, a .dat or .opt file, a spreadsheet index, a cover letter. These tell you where documents begin.
4. Check that names sort in page order. PROD0000098.tif sorts before PROD0000099.tif. page98.tif sorts after page100.tif in a plain alphabetical listing, because there are no leading zeros.
5. Note colour. Black-and-white pages and colour pages convert differently, and a mixed set needs both handled.

Write the totals down. Files, pages and documents are three different counts, and the checks after conversion compare against all of them.

Find the document breaks

A single-page set does not show where one document ends and the next begins. Sources for the boundaries, in the order to try them:

• A load file or index that lists the first and last page of each document.
• The folder structure, if each document has its own folder.
• Gaps or prefixes in the file names.
• The pages themselves: a letterhead, a caption, a fax cover sheet, a signature page.

The post on document unitization covers how boundaries and attachment families are recorded in a scanned set.

Turning a TIFF set into document-level searchable PDFs

Work in three stages: group the pages of each document, convert each group to one PDF, then run recognition on each PDF. Keep the TIFF set untouched and build the PDFs in a separate folder.

Stage one: group

For a multi-page TIFF, the grouping is done. For single-page files, make a list with one row per document: its name, its first file and its last file. A spreadsheet is enough. Everything after this stage follows the list, so check it against the boundaries you found before going on.

Stage two: wrap the images in a PDF

Use any tool whose documentation says it takes TIFF as input and that you are permitted to use on the matter. Three documented examples:

• A PDF editor. Adobe's list of file formats Acrobat converts to PDF includes .TIFF among its image files. Its guide to combining files into one PDF describes adding files, arranging them and merging them into a single document.
• The libtiff tools. The tiff2pdf manual says the program "converts one TIFF file to one PDF file, including multiple page TIFF files". For a set of single-page files it says: "If you have multiple TIFF files to convert into one PDF file then use tiffcp or other program to concatenate the files into a multiple page TIFF file." The tiffcp manual describes that tool as one that "combines one or more files" into a single TIFF file.
• img2pdf. The img2pdf project page describes it as "Lossless conversion of raster images to PDF" and lists TIFF with CCITT Group 4 data among the inputs it embeds directly.

The command-line tools suit litigation support staff working through hundreds of documents from a list. For a dozen faxes, a PDF editor's menu is quicker.

When the reverse is needed, the tiffsplit manual describes a tool that takes a multi-page TIFF file and creates single-page files from it. That is useful when one enormous TIFF holds many documents and has to be regrouped.

Stage three: recognize

Now each document is a PDF made of page pictures, which is the ordinary starting point for OCR. Run recognition and save a searchable PDF. The guide to making a scanned PDF searchable covers that step and how the text layer is built.

When the recognition tool does not accept TIFF

Check the accepted file types before you plan the job. Schema Law accepts PDF, PNG and JPEG uploads of up to 20 MB, and TIFF is not on that list, so a TIFF has to be converted before it is uploaded. Convert to PDF as described in stage two and upload the PDF; one PDF per document keeps the grouping, where a folder of PNG pages would lose it. Once the pages are read, searchable PDF is one of the export formats, and the pricing page gives the page limit per document on each plan.

A long multi-page TIFF can produce a PDF over a tool's size limit. The post on converting a large scanned PDF online covers splitting one oversized file, and the post on image file types as OCR input covers TIFF beside HEIC, screenshots and the other formats a tool may refuse.

A worked example

Odalys Ferreira, a paralegal, receives a production of 3,180 single-page TIFF files named KST0000001.tif to KST0003180.tif, a folder of matching text files and a load file.

1. She copies the production to a read-only folder and works from a second copy.
2. From the load file she builds a list of 412 documents, each with a first and last Bates number.
3. She totals the page ranges on the list. They add up to 3,180, so no page is unassigned and none is claimed twice.
4. She converts each range to one PDF named by its first Bates number, KST0000001.pdf, using a tool that copies black-and-white image data without re-encoding it.
5. She runs recognition on the PDFs in batches, saving the searchable copies to a third folder.
6. She reconciles the result against her list: 412 PDFs, 3,180 pages in total, and the right first and last page in a sample of documents.

The production folder is never modified. If a question later arises about what was produced, the answer is in that folder and not in her working copies.

Can you reuse the OCR text files that came with a TIFF production?

Yes, for searching and reading; no, as a ready-made text layer. A text file that arrives with a TIFF page holds the words and nothing about where each word sits on the page. A searchable PDF needs each word placed over its picture, which recognition on the image supplies.

Where that text lives in a production, and how it is tied to the images, is set out in the section on where the OCR text lives in each production format. What matters for a conversion is the choice between two uses.

Before relying on supplied text, test it. Open ten text files beside their images, chosen from different parts of the set, and compare: is there text for every page, does it match the picture, are names and amounts right? Look also at pages the producing party redacted or withheld, and check that the text you were given for them matches what the image shows and no more. If something there looks wrong, that is a question for the supervising lawyer before anyone works with the text.

Running recognition yourself gives you a second, independent reading of the same pages. Where the two disagree on a name or a figure, the image decides. Keep the supplied text with the production either way; it is part of what was produced.

If the review platform is where the searching will happen, find out whether it runs recognition on what you load, since that changes what you need to prepare. The post on OCR before uploading to a review platform covers that question.

Does converting TIFF to PDF lose image quality?

It does not have to. A PDF can hold the same compressed picture data a TIFF holds, so a converter can copy the page across unchanged. Quality is lost only when a tool decodes the picture and compresses it again with a lossy method or at a lower resolution.

The documentation of the tool tells you which kind it is.

• The tiff2pdf manual says that when the input contains single strip CCITT G4 Fax compressed information, "that is written to the PDF file without transcoding" unless options are set to prevent it. For colour and greyscale data it says the output "can be compressed using either JPEG compression" or Zip/Deflate, with a quality option from 1 to 100 for JPEG.
• The img2pdf project page states as its first priority that "the image embedded in the PDF will always have the exact same color information for every pixel as the input".
• A PDF editor's conversion settings may include compression and downsampling choices for images. Open the settings and read them before a large run.

Three places quality goes missing

A JPEG step in the middle. Converting black-and-white TIFF pages to JPEG, to satisfy a tool that will not take TIFF, re-encodes sharp letter edges with a method built for photographs. If an image format is needed, PNG keeps every pixel. Converting to PDF avoids the question.

A size-reduction option left on. "Reduce file size", "optimize" and "downsample" settings exist to make files smaller by discarding detail. Switch them off for the conversion and decide about size separately.

Printing to PDF. Printing a TIFF from a viewer to a PDF printer redraws the page at the printer's settings and can add margins or scale the image. Use a convert or combine command instead.

The compression methods themselves, and which are lossy, are explained in the post on lossy versus lossless compression in scans.

A quick comparison

Pick one page with small print, such as a footnote or a fax header line. Open the TIFF and the converted PDF side by side at 400% zoom. The letter edges, the specks and the background should look the same. If the PDF page shows blur, blocky halos around letters, or a page of a different size, change the tool or its settings before converting the rest.

Checking page counts and document breaks after a TIFF conversion

Check three numbers and two boundaries. The number of PDFs equals the number of documents on your list, the total pages across the PDFs equals the number of TIFF pages, and each PDF's page count equals its range. Then confirm that each sampled PDF starts and ends on the right page.

The reconciliation

A multi-page TIFF that is opened in a viewer showing only its first page is one way to end up with counts that disagree: the converter sees every page and the person checking saw one. Count pages with a tool that reports them, not by eye.

Checking the breaks

For a sample of documents, including the longest and several of the shortest:

1. Open the PDF at page one. It should look like the start of a document: a letterhead, a caption, a cover sheet.
2. Go to the last page. It should look like an ending, and the next PDF in the list should begin with a new first page.
3. Where pages carry Bates numbers, read the number on the first and last page against the file name and the list.
4. Where a document has attachments, confirm they follow it and were not merged into it or cut off.

Checking the text

Searchability gets its own pass. Search each sampled PDF for a word you can see on a page in the middle, not the first page. The guide to checking whether a PDF is searchable on every page gives a method for a whole set. Fax-resolution pages deserve a closer look than clean scans, since coarse images are harder to read.

Record the result in the grouping spreadsheet: a column for the PDF's page count, one for "matches range", one for "search tested". When the set is used months later, that sheet is the evidence that the working copies match the images they were made from.