How to Reduce PDF Size for Court Filing and Keep the Text

To reduce PDF size for court filing and keep the text, find the court's limit, then work in this order: rescan oversized pages in black and white at the court's resolution, compress the page images in a way that leaves the text layer in place, and split what is still too large into labeled volumes. Test search on every volume.

The order matters because each step costs something different. Rescanning costs time at the scanner. Compression costs image quality. Splitting costs nothing in quality but adds files, labels and docket entries. This post takes each in turn for a scanned exhibit, using limits and instructions read on named courts' own sites. It explains document handling and is not legal advice.

Why is a scanned exhibit so much larger than the brief it supports?

A scanned exhibit stores a picture of every page, and a brief converted from a word processor stores characters. A picture of a letter-size page holds millions of dots. The text of the same page is a few thousand characters. Scanner settings decide how many dots there are and how many bits each one takes.

The arithmetic is short. A letter-size page is 8.5 by 11 inches. At 300 dots per inch that is 2,550 by 3,300 dots, or about 8.4 million.

Scanning software compresses the pages it saves to PDF, so real files are much smaller than those figures. The ratios between the rows are the point. Before compression, a grayscale page carries eight times the data of a black and white page, and a color page carries twenty-four times. Doubling the resolution to 600 dpi multiplies every row by four, because both the width and the height double.

Compression then narrows or widens the gap depending on what is on the page. Black and white pages of clean type compress very well, because long runs of white dots are easy to describe briefly. Photographs, shaded backgrounds, highlighter, coffee stains and the gray speckle from a photocopy of a photocopy compress poorly, because nearly every dot differs from its neighbor.

The brief has none of this. A page of typed argument is roughly 3,000 characters plus font and layout instructions. One federal court puts a number on the difference: the Eastern District of Oklahoma's CM/ECF technical information document says "a document you convert to PDF using your word processor could be approximately 20% of the size of a scanned version of that same document." That is one court's estimate, and your own files will vary.

Two practical conclusions follow.

• Anything you hold as an electronic file should never pass through a scanner. An email, a spreadsheet or a contract that exists as a file converts directly to a small PDF.
• The hidden text layer is not the problem. The words that make a scan searchable are characters, so they add about as much as the text of a brief would. If an exhibit is over the limit, look at its images.

To see where the weight sits in a particular file, divide its size by its page count. Then page through and look for the outliers: the color pages, the photographs, the pages scanned at a higher setting than the rest.

How do you find your court's per-document and per-filing size limits?

Look in the court's e-filing rule, its administrative procedures or user manual, and the clerk's notices and how-to pages, and read for two numbers: the cap on one PDF and the cap on everything uploaded in one transaction. Some courts publish both, some only one. Note the date on whatever you find.

Limits are published in different kinds of documents from one court to the next. Five examples show where to look and how the numbers are framed. Each was read on the court's own site, and each can change.

Two numbers, not one

The North Carolina and Oklahoma documents show why the per-filing number matters. In the Eastern District of North Carolina, a 50 MB main document leaves 10 MB for every attachment combined. The guide warns that the system "does not provide reminders of the total file size used" and that an oversized upload looks normal "until the filer tries to submit/commit the filing." You find out at the last screen.

So when you read your court's material, write down:

1. The limit for a single PDF.
2. The limit for all PDFs in one docket entry or transaction, if one is stated.
3. Any separate limit for exhibits, as in California's rule 8.74(c)(6).
4. What the court tells you to do when you are over: split, file in several entries, or deliver on physical media.
5. The date of the document, and whether a newer notice replaces it.

Some limits are not in the court's own rule at all. The Texas Rules of Appellate Procedure, rule 9.4(j)(4), refer to "the size limit prescribed by the electronic filing manager", which sends you to the e-filing system's documentation for the number.

The post on searchable PDFs for court filing has limits from three more courts and what each says to do when a file is over. For how federal district limits are set and where each district's administrative procedures sit, see the post on ECF PDF requirements. If the documents you find conflict or look stale, call the clerk's office and ask for the current figure.

Shrink the exhibit only after you write down the size limit for that electronic filing. A motion with attachments can cross the combined limit even when each PDF looks small. Read the clerk of court page for the system you are using, and put the limit in your notes. After you compress, search for a word on the first page and a word on the last page.

Rescanning an oversized exhibit at filing resolution and color mode

When you still have the paper, rescanning is the cleanest fix. Scan in black and white at the resolution your court names, keep color only for pages where color carries meaning, and add the text layer after the scan. A page scanned at the right settings needs little or no compression afterwards.

General scanner setup is covered in the guide to scanner settings that make OCR work.

Read the court's settings before you touch the scanner

Courts that publish scanning settings do not all say the same thing. The Eastern District of Oklahoma's document says "Use 300dpi for scanning documents" and "Set image type to black and white bitmap, text (image only), or line art". It also says "Do not use OCR or Textbridge. For filing purposes only an image of the document is required." California's rule 8.74(a)(1) points the other way and requires "text-searchable portable document format (PDF)". Follow your own court's instruction for the filed copy, and note that a court's position on OCR can be the opposite of the one you learned somewhere else.

Color has its own rules in places. California's rule 8.74(a)(8) says a document with a color component "may be electronically filed or manually filed on electronic media, depending on its file size."

A rescan procedure

1. Find the heavy pages. Sort out which pages of the exhibit are color, grayscale or photographic. A 300-page exhibit can be over the limit because of twenty pages.
2. Decide page by page whether color matters. A photograph of an injury, a color-coded chart, a map with colored boundaries and a document where the dispute is about ink color all need it. A letterhead logo does not.
3. Set the scanner for the plain pages. Adobe's list of scan settings in Acrobat names the controls you will find in some form on any scanner: a color mode of "Autodetect, Color, Black and White, or Grayscale", a resolution in DPI, and an Optimize Image control that Adobe describes as adjusting "the balance between file size and quality." Choose black and white and the court's resolution.
4. Avoid automatic color detection for a filing. Autodetect can decide that a yellowed page or a page with a blue signature is a color page. Set the mode yourself.
5. Scan the color pages as a separate batch in color at the same resolution, then put them back in their places.
6. Look at the smallest print. Black and white scanning turns every dot fully black or fully white. Fine print, faint carbon copies and pencil can break up. Open the rescanned page at 100% and read the footnotes, the fax header and any handwriting. If a page is not legible in black and white, rescan that page in grayscale and accept the extra size.
7. Add the text layer if your court requires or allows one. The guide on how to make a scanned PDF searchable covers that step.
8. Count the pages against the original and check the new size.

When you no longer have the paper

If the exhibit came to you as a PDF and the original is with a client or a third party, you cannot rescan. You can ask the sender for a copy scanned at lower settings, which is worth one email when the deadline allows. Otherwise the choice is between compressing the images you have and splitting the file.

One compression setting deserves a caution for exhibits. Adobe's settings page lists "JBIG2 (Lossy)" for monochrome scans and says lossy mode "reduces file size by up to 60%". Lossy means the saved image is no longer an exact copy of the scan. For a document offered as evidence, think about whether you want to explain that, and consider the lossless option on the same list.

Does compressing a filed PDF damage its text layer?

Whether compressing a filed PDF damages its text layer depends on how the file is made smaller. Recompressing the page images at a lower quality leaves the hidden text in place. Rebuilding the pages, by printing to PDF or saving each page as a picture, can discard it.

Test the output either way, because the documentation read for this post does not promise that search survives.

Why print to PDF and image export rebuild the page is explained in the post on the steps that stop a PDF being searchable. The table here is limited to what each route means for a file you are shrinking to fit a court's limit.

What the optimizer's other panels remove

Size reduction tools do more than shrink pictures. Adobe's description of the optimizer lists a Discard Objects panel that can "Remove elements like embedded thumbnails, search indexes, and flatten form fields", a Discard User Data panel that can "Remove comments, forms, multimedia, and private data", and a Clean Up panel that can "Remove invalid elements such as bookmarks or links". Some of that you may want gone. Some of it, such as bookmarks a rule requires, you do not. Compress before you add bookmarks where you can, and see the post on how to bookmark a PDF for court filing for the order of steps.

Compression can hurt the text without touching the text layer

Three situations cause trouble even when the hidden words survive.

Compressing before OCR. Recognition reads the image. If you downsample a scan heavily and then run OCR, the software is reading a worse picture and will make more mistakes. Run OCR on the best image you have, then compress.

Compressing until a person cannot read the page. The text layer may still find "indemnify" on page 40, but if the visible clause has turned to gray blocks, the judge cannot read what the search found. The image is the exhibit. The text layer is an index to it.

Running OCR again on a compressed file. If a later step drops the text and you re-run recognition on the compressed images, the new text layer is built from the degraded page. Keep the uncompressed, searchable original so you can go back to it.

A workable order is: scan, OCR, redact and verify, compress, split, bookmark, test. If you use an online tool for the OCR step, its upload limit applies before the court's does. Schema Law accepts PDF, PNG and JPEG files of up to 20 MB, and the homepage describes its searchable PDF export as your original file with a hidden text layer. Nothing on the site says it compresses files, so do not count on it to get an exhibit under a limit. The formats section of the homepage describes each export.

Splitting an oversized exhibit into volumes the clerk will accept

Split when settings and compression cannot bring a readable exhibit under the limit. Cut at document boundaries, keep every part under the cap with room to spare, number the pages straight through, and label each part so the docket shows what it is and where it falls in the whole.

Courts say different things about how parts are filed, so read your court's text before you cut.

• Central District of California: the clerk's notice says PDF files "shall contain no more than one document or portion of one document per file" and that files over the limit "must be divided into sub-volumes."
• Eastern District of Oklahoma: an oversized document "must be split into smaller PDF files, each less than 10 MB. These additional parts may be filed as attachments to the main document."
• Eastern District of North Carolina: when the parts together exceed the per-filing cap, the guide shows the filer creating a placeholder document, in its example titled "Notice of Continuation of Filing Regarding the Complaint", and filing the remaining attachments in a second entry linked to the first.
• Tenth Circuit: appendices larger than 30 MB "must be docketed in multiple CM/ECF transactions", and "If you split a single volume of the appendix into multiple parts to accommodate e-filing, each part is considered a separate volume" for the hard copy requirement.
• California reviewing courts: rule 8.74(a)(5) says a document over the limit must be submitted "in more than one file", that "The first file must include a master chronological and alphabetical index stating the contents for all files", and that each file must have a cover page stating the file number and total number of files, the volumes it contains and the page numbers it contains. The files must be "paginated consecutively across all files in the document, including the cover pages for each file."

How to cut

1. Set a target below the cap. If the limit is 35 MB, aim for parts of about 30 MB. Cover pages, bookmarks and a final save all add a little.
2. Cut between documents. End a volume at the last page of a sub-exhibit, a tab or a letter. Do not leave the signature page of a contract alone at the start of the next volume.
3. Use your editor's split function, then adjust. Adobe's instructions for splitting a PDF say you can split "by the maximum number of pages, maximum file size, or top-level bookmarks." The page does not say that a split by file size respects document boundaries, so look at where each cut fell and move it if it landed mid-document. A split by top-level bookmarks follows your exhibit boundaries if the file is already bookmarked that way.
4. Add the cover page or label your court expects to each part.
5. Paginate after splitting only if your court wants each part to restart. Otherwise number the whole exhibit first so the page numbers run through.
6. Name the files so they sort in order.
7. Bookmark each volume on its own.

A worked example

Fictional facts: in Halvorsen v. Reyes Mechanical, the plaintiff's Exhibit 12 is 412 scanned pages of maintenance records, 96 MB after rescanning the plain pages in black and white. The court's cap is 35 MB per PDF. The color photographs in the middle of the set are the heavy part.

The volumes are uneven in page count on purpose. Each cut falls where one group of records ends, and the photograph volumes are short because their pages are large. The docket text for each attachment would follow the same pattern, for example "Exhibit 12, Volume 2 of 4 (pages 151 to 198)". If the court also had a 60 MB cap per filing, these four parts total 96 MB and would need two entries.

A size and search check on each exhibit volume before upload

Check every volume as its own file and then the set as a whole. Each part must be under the per-document cap, the parts filed together must be under any per-filing cap, the page counts must add up to the original, and a search must find words at the start, middle and end of each part.

Run the per-volume checks on the exact files you will upload, after the last save.

Then the checks across the set:

• Add up the sizes of everything going into one docket entry and compare the total with the per-filing limit.
• Add up the page counts and compare with the original exhibit. A cut is where a page goes missing or gets doubled, so look at the last page of each part beside the first page of the next.
• Read the labels in sequence. Volume 1 of 4 through Volume 4 of 4, with page ranges that meet and do not overlap.
• Confirm the index names every part, if your court requires a master index.
• Keep the uncompressed, unsplit original in the matter folder with a note of the settings you used.

If a volume fails

A record of the check is worth keeping. A three-column note of file name, size and "searched and read on" date takes a minute and answers the question if a clerk or opposing counsel later says a volume was defective.