OCR Microfilm Prints: Film, Fiche and Photostat Copies

OCR can read a page printed or scanned from microfilm or microfiche, with more errors than a scan of the paper would give. The film adds grain, scratches, dark borders and sometimes white letters on a black page. Crop the frame, turn a negative into a positive, sample a few pages, and ask whether a direct film scan exists.

Can OCR read a page printed or scanned from microfilm?

Yes, and how well depends on the single frame in front of you. A sharp, evenly lit frame of a typed or printed page converts into usable text. A dim, scratched frame of a handwritten ledger may give you almost nothing. Results change from page to page within one reel, so judge by sampling, not by the first page.

A microfilm copy reaches you in one of four forms, and it helps to know which you hold.

Sample before you commit time to a long file:

1. Pick three frames: the cleanest, a typical one and the worst.
2. Convert them and read each result against its image, line by line.
3. Count what went wrong in names, dates, amounts and reference numbers, since those are what you will search for.
4. Decide per document: convert and spot-check, convert and proofread, or read from the image and type what you need.

Schema Law's free plan takes one document a day of up to three pages, which fits that three-frame sample. It reads pages into text and tables, its recognition is strongest on clean print, and nothing marks a passage as uncertain. The steps from upload to export are on the home page. What the film did to your pages is something only a test on those pages will show.

What does filming and reprinting do to the letters on a microfilm page?

Filming shrinks a page to a small frame, and every later copy and print enlarges that frame again with some detail lost. Letters come back thicker or thinner than they were, fine strokes break, enclosed letters fill in, and marks on the film itself are printed as though they were on the page.

How many copies stand between you and the paper

The reel a member of the public threads onto a reader is not the film that passed through the camera. Cornell University Library's preservation tutorial on microforms and the three generations of film describes the practice for preservation filming: "three generations are produced: the camera or master negative, the print master negative, and the positive-use copy." It adds that the positive copy "is what readers use."

Not every record series was filmed to that practice, and you cannot tell from a print which generation it came from. Count the steps you can see. A paper print from a public reader is a print, of a projected image, of a use copy, and that use copy is a copy of an earlier film. Each step is a chance to lose the thin parts of letters.

What the losses look like

Soft focus deserves a separate look, because a page can be large enough in pixels and still unreadable. If the letters have no hard edges at any zoom, a higher scan resolution of the same print will not help. The post on blurry scans explains how to tell whether a page is blurry or only low in resolution.

What is not the film's fault

The original mattered too. A reel of typed court papers carries every habit of the typewriter that made them: uneven ink, strikeovers and a lowercase L standing in for the digit one. Those survive filming and are covered in the post on OCR for typewritten documents. A reel of handwritten registers is a handwriting problem first and a film problem second.

White-on-black negative prints and photostats

A negative print shows white letters on a black page. The words are all there, reversed in tone. Recognition software built around dark text on a light page may read a negative poorly or not at all, so the safe course is to turn the image into a positive yourself and keep the negative as received.

Why a page arrives reversed

Film can hold an image in either polarity. If the reel is a negative and the reader-printer was set to print what it saw, you get white on black. If the machine was set to reverse the image, you get black on white. A photostat in an old file can have the same look: a sheet of heavy photographic paper with white writing on a black ground. The cause does not change what you do next.

Turning a negative into a positive

1. Save the file as received under a name that says so. This is the copy you would show anyone who asks what you were given.
2. Open a duplicate in an image editor, or in a PDF tool that edits page images, and use the command named Invert or Negative. Every tone flips: black becomes white and white becomes black.
3. Look at the result at full size. The page background should now be light and even. If it is a muddy gray, adjust brightness and contrast a little, and stop as soon as the thin strokes begin to break.
4. Check for pages that were positive all along. Inverting a whole file turns those into negatives. Flip page by page when a file is mixed.
5. Convert the positive copy. Compare the text with the positive image, and for anything you will quote, with the negative as received.
6. Note in the file what you did: "Pages 3 to 11 received as negative images; inverted for conversion; originals retained."

Inverting is a change to the image, and a visible one. If the copy is certified, or will be an exhibit, the unaltered copy is the one that goes forward. The inverted copy is a working tool for finding and reading text.

What reversal does to the marks around the text

• Scratches change tone. A scratch that was a white line on the negative becomes a black line through the text after inverting, where recognition is more likely to read it as a character.
• Stamps and seals may vanish. A red or purple stamp filmed in black and white may sit at the same tone as the paper. Look at the image for stamps; do not expect them in the text.
• Borders flip too. A black surround becomes white, which is convenient. A white surround on a positive that you inverted by mistake becomes a black frame.
• Handwritten notes on a print. Someone may have written in dark pen on the negative print itself. After inverting, that note is white on white paper and easy to miss. Read the as-received copy for annotations.

If a tool does read your negative page without help, test it on a dense page before trusting it on a file, and compare against a result from the inverted copy.

Black borders, film edges and two frames on one microfilm image

A microfilm image carries more than the page: a black surround, the edge of the film, and sometimes a neighboring frame. Recognition tries to make sense of all of it. Crop each image to the page, one page per image, before you convert, and keep the uncropped version.

What the clutter is

• The surround. The page was filmed on a dark copyboard, or the frame is larger than the page. On a positive print the surround is black.
• Film edges. The top and bottom of the frame may show the edge of the film as a hard line, with or without markings.
• Blips and counters. Small marks beside each frame, or a number in a corner, used by a reader to count frames.
• Targets. Frames at the start of a reel that show a title sheet, a resolution chart or a certificate from the camera operator.
• A neighbor. Part of the previous or next frame at the side, cut off mid-line.
• Two pages in one frame. An open book filmed as a spread, or two small documents laid side by side.

What each does to converted text

A black surround can be read as a table cell or a thick rule, which pushes the real text into odd blocks. A neighbor's half-lines appear as fragments at the start or end of your lines. A spread is the worst case: software may read straight across both pages, joining line one of the left page to line one of the right. The post on whether OCR can handle two-column documents explains how reading order goes wrong and how to spot it.

Cropping without losing evidence

1. Keep the full frame. The surround, the frame counter and the target frames show where the page sits on the reel, and you may need them to cite or re-find it.
2. Crop a working copy to just outside the paper's edge on all four sides. Leave the whole page in, including margins, since stamps and filing notes live there.
3. Split a two-page frame into a left image and a right image, and name them in reading order. The guide to converting a scanned book covers splitting a two-page spread without scrambling the page order.
4. Straighten each page after splitting, not before. The two halves of a book lean in opposite directions.
5. Record the reel and frame for each working image, for example reel-0412_frame-0237_left.png.
6. Convert the cropped pages, then check that the first and last words of several lines match the image. Missing line ends mean the crop was too tight.

Do not throw away target frames. A title target names the record series and the dates filmed, and a certificate frame says who filmed the records and when. Those frames are not part of any document, but they explain what the reel is.

Should you ask the records office for a new scan from the microfilm reel?

Ask when the text you need falls in a part of the print that the print itself damaged: soft focus, heavy speckle, a cut-off edge or a muddy negative. A digital scan taken directly from the film removes the print and the paper scan from the chain. Whether an office can supply one is for that office to say.

Signs that a better copy would help

• You can read the passage on the reader screen and cannot read it on the print.
• The print is dark at one side and pale at the other.
• A scratch or blob sits on the print and moves or disappears when the film is repositioned.
• The page was printed at a size where small type is only a few dots tall.
• The edge of the page is cut off by the print area.

Signs that it would not

If the film image is itself faint, out of focus or cut off, a new scan from the same reel gives you a cleaner picture of the same defect. Then the questions change: does the office hold the paper original, an earlier generation of film, or a second filming of the same series?

What to ask for

Offices differ in equipment, fees and what their rules allow, so put the request as questions.

1. Can you supply a digital image scanned directly from the film, in place of a paper print?
2. Can it be grayscale, saved as TIFF or PDF without heavy compression?
3. Can each frame be a separate page, with the full frame showing?
4. Can negative film be delivered as a positive image, or both ways?
5. Is the reel I viewed a use copy, and is there an earlier generation or the paper original that a copy could be made from?
6. If I need a certified copy, can the certificate be issued for a copy made this way?
7. What are the reel, frame and series identifiers, so I can cite the page and ask for it again?

A worked example: a paralegal at the fictional firm Aldous and Reyes needs a 1958 judgment from a closed civil file. The clerk's office mails a reader-printer sheet. The caption and the body read well, but the dollar amount in the final paragraph sits in a dark corner of the print, and conversion returns "$3,8 .00". She does not guess the missing digits. She writes back with the case number, the reel and frame printed on the sheet, and questions 1 to 3 and 6. The image that comes back is lighter, the amount is legible on screen, and she types it into the working copy from the image, with a note of which copy she read it from.

Court files, land records and vital records held on microfilm

Closed court files, recorded land instruments and vital records are among the records an office may keep on film. Each puts the hard-to-read content in a different place: file stamps and docket entries, legal descriptions and book-and-page references, or handwritten names and dates on a printed certificate.

Closed court files

A filmed case file mixes typed pleadings, printed forms, handwritten minute entries and exhibits of every kind, filmed in whatever order the folder held them. Check that you have the whole file: frame numbers or page counts should run without gaps. File stamps and clerk's notes are in margins, where a tight crop removes them. Docket sheets are tables with handwriting in the cells; read them from the image.

Land records

A deed copied from film is a recorded instrument, sometimes itself a photographic copy bound into a book before filming. The legal description is the part where one wrong character matters: bearings, distances, lot and block numbers. Read every figure against the image, and check the book and page stamp, since that reference is how you ask for the page again. Index books on film are handwritten ledgers in columns; the post on OCR for handwritten notes sets out what recognition can and cannot do with them.

On a filmed warranty deed you read the habendum clause aloud, then the stamp that shows where the office recorded it. A gap in the chain of title often sits in a scratched book or page figure, so you match that stamp to the reel label before you accept a distance or a bearing. Whether the recording act treats this print as the recorded instrument is for counsel.

Vital records

A birth, death or marriage certificate from film is a printed form with entries typed or written in. The printed labels convert. The entries are short, are proper names and dates, and have no context to correct a misreading. Treat converted text as a way to find the certificate, and take the facts from the image. The certified paper copy, not your conversion of it, is the document.

Newspapers and other material

A newspaper archive may exist only on film, and a clipping or legal notice printed from a reel has narrow columns and small type on top of every film fault. The post on OCR for newspaper clippings covers columns, hyphenation and articles that jump pages.

A note for the file

Whatever the record type, write down four things with each converted page: the office that supplied it, the reel and frame or other locator, the form in which it arrived (paper print, film scan, negative or positive), and anything you did to the image before conversion. A later reader who finds a gap in the text can then tell whether the gap is in the record or in the copy. Before uploading records that name clients or third parties to any service, read what it promises about storage and deletion; Schema Law's commitments are under security on the home page.