# Best Scanner App Settings for Pages You Plan to OCR

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2022-04-08T09:51:28Z
- Updated: 2026-10-11T19:22:20Z
- Category: Phone scans and photos
- Canonical: https://www.schemalaw.com/blog/best-scanner-app-settings

> The best scanner app settings for OCR: which filter to pick, when to switch off auto capture, how to check the crop, and why you keep the untouched photo.

The best scanner app settings for pages you plan to OCR are the ones that change the picture least: a color or grayscale filter instead of black and white, a crop you have checked by eye, manual capture for anything irregular, the highest quality the app offers, and a saved copy of the unprocessed photo.

A scanning app does four things to a photo before it hands you a file. It finds the page, cuts away the desk, pulls the page into a rectangle and repaints the tones. Each step has a setting or a correction screen, and each can remove something a matter turns on. The sections here take the controls one at a time and say what to look at after you touch them.

## What are the best scanner app settings for OCR?

The best scanner app settings for OCR keep detail and leave decisions to you. Choose color or grayscale, review every crop, use manual capture when pages vary, save at the top quality setting, and keep the original picture. Then read the smallest print on the result before the paper leaves your hands.

As a list you can work down the first time you open any scanning app:

| Control | Set it to | Then check |
| --- | --- | --- |
| Filter | Color, or grayscale for plain print | Pencil, pale stamps and highlighter still show |
| Crop outline | Dragged out to the paper's true corners | Page number, footer, margin notes, hole-punch strip |
| Perspective | Left on, with the phone held flat over the page | Top line and bottom line are the same size |
| Capture | Manual for mixed or awkward pages, auto for a uniform stack | One image per page, none skipped, none doubled |
| Quality or size | The highest offered | Footnotes are readable at full zoom |
| Output | PDF for a document, image for a single page | Pages are in order and upright |
| Original photo | Kept, if the app has the option | A second, unprocessed copy exists somewhere you control |

Not every app shows every control. Apple's guide to [scanning documents on an iPhone or iPad](https://support.apple.com/en-us/108963) describes an Auto mode, a shutter button, corners you drag "to adjust the scan to fit the page" and, in iOS 26, "options such as flash or an image filter". Google's guide to [scanning documents with Google Drive on Android](https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid) lists a Manual and Auto capture toggle, Crop & Rotate, Filter, Clean and a choice of .pdf or .jpg. Neither page mentions a resolution setting, so on those apps the quality row comes down to how you hold the phone. The per-app steps for Android are in the guide to [scanning a document on Android](https://www.schemalaw.com/blog/how-to-scan-a-document-on-android).

One rule sits over all of the rows: a setting that discards information cannot be undone later. A filter that turns a gray pencil note white has deleted the note from that file. That is why the untouched photo matters more than any single switch.

On a client's medical bill, that untouched photograph is the file to hold if the scan may be documentary evidence. A black and white filter can wipe a pale stamp or a penciled total. Leave color on, confirm the crop includes the account line, and date the original so authentication and the chain of custody start from it.

## Color, grayscale or black and white filter in a scanning app?

Pick the color filter when the page has stamps, highlighter, colored ink or a signature you may need to describe. Pick grayscale for plain black print on white paper. Use black and white only when file size forces it, and only after comparing the result with the paper.

A filter in a scanning app is a repaint of the photograph. The app has already taken a full color picture. The filter then decides what each pixel should become. The names differ from app to app, and the meaning of a name is not standard, so judge a filter by what it does to a test page and not by its label.

### What each kind of filter does to a legal page

| Filter family | What the app does | What can go missing |
| --- | --- | --- |
| Original or photo | Leaves the picture as shot | Nothing, but shadows and a gray background stay in |
| Color document | Whitens the paper, raises contrast, keeps hues | Very pale marks near the paper tone |
| Grayscale | Drops hue, keeps shades | The difference between a red stamp and black type |
| Black and white | Forces each pixel to pure black or pure white | Pencil, faint stamps, light signatures, shaded table rows |

The reasoning behind the three modes, and what recognition software makes of each, belongs to the post on [grayscale versus black and white scanning](https://www.schemalaw.com/blog/grayscale-vs-black-and-white-scanning). The point specific to a phone is that the choice is made after capture and can be changed, as long as you change it before you save and close the scan.

### A filter test on one hard page

Take one page from the file that has the hardest marks on it: a received stamp, a pencil annotation, a blue ink signature, a highlighted line. Scan it once and step through every filter the app offers. For each one, ask three questions.

1. Can you still see every mark that is on the paper?
2. Can you still tell which marks are handwritten and which are printed?
3. Is the small print thicker and joined up, or thinner and broken, compared with the paper?

A filter that fails the first question is out for that matter. A filter that fails the third will produce misread letters, because recognition depends on the shapes of the strokes.

Google's Drive scanner also offers a tool it calls Clean, described in the [Drive scanning help page](https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid) as a way to "Erase stains, fingers, and more". Treat a tool like that with care on legal paper. Something the app takes for a stain may be an initial, a correction or a notary's mark. If you use it, compare the cleaned page with the original photo afterwards.

## What does automatic edge detection cut off a page?

Automatic edge detection can cut off anything near the rim of the sheet: page numbers, footers, Bates numbers, fax header lines, margin notes, hole-punch strips and the edge of an oversized stamp. The app looks for a contrast boundary, and a boundary inside the page fools it as easily as the real edge.

Edge detection is the outline you see on screen before capture. Google's [Drive scanning guide](https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid) puts it this way: "A blue line outlines the document. This indicates where the picture will be cropped." Everything outside that line is thrown away when the scan is saved.

### Where the outline goes wrong

- **White paper on a pale desk.** With no contrast at the true edge, the app may settle on a printed border, a table rule or a letterhead box.
- **A page with a ruled frame.** Pleading paper, certificates and forms with a printed border invite the app to crop to the frame and lose whatever sits outside it.
- **A page lying on another page.** The outline can run out to the sheet underneath, or stop at a sticky note.
- **A folded or dog-eared corner.** The app draws a straight side through the fold and trims the corner off.
- **A bound volume.** The gutter shadow reads as an edge, and the inner margin goes.

### What to check after every crop

Look at the four sides in turn, not at the middle of the page. On each side, find the outermost printed or written thing on the paper and confirm it is inside the image with a strip of blank paper beyond it. A crop that touches ink has probably removed some.

If the outline is wrong, fix it on the correction screen instead of retaking and hoping. Apple's [iPhone scanning guide](https://support.apple.com/en-us/108963) says to "drag the corners to adjust the scan to fit the page". Google's guides for Drive and for [Files by Google](https://support.google.com/files/answer/15598575?hl=en) both name a crop and rotate control for adjusting the scan area. Drag each corner slightly past the paper's corner. A sliver of desk in the picture costs nothing. A missing page number costs you the ability to cite the page.

Capture on a dark, plain surface whenever you can choose. It gives the app a clear boundary and makes a bad outline obvious before you press the shutter.

## How perspective correction in a scanning app reshapes the page

Perspective correction stretches the four-sided shape the camera saw into a rectangle. It makes a tilted shot look as though it was taken square on. It does not add detail, so the part of the page that was farther from the lens stays softer than the nearer part.

When the phone is tilted, the paper appears as a trapezoid. The app takes the four corners of the outline and maps them to the corners of a rectangle, then redraws every pixel in between. The far edge of the sheet, which occupied fewer pixels in the photo, is spread over the same width as the near edge. The lines of text come out parallel and the page looks flat.

Samsung's page on [scanning documents with a Galaxy phone or tablet](https://www.samsung.com/us/support/answer/ANS10003696/) describes the same idea in its own words, saying that scanning "will also align rectangular shapes so your document will appear perfectly straight and legible". That is the maker's description of the feature. Whether a given page is legible afterwards is something you confirm on the page.

### Limits to know about

- **Stretched text is soft text.** Compare the top line and the bottom line at full zoom. If one is visibly blurrier, the phone was tilted and the correction covered it up. Retake from straight over the center of the sheet.
- **Wrong corners give a warped page.** The correction is only as good as the outline. If one corner of the outline sat on a table rule instead of the paper's corner, the whole page is skewed toward it and straight lines of type lean.
- **Curved paper stays curved.** The correction assumes a flat sheet. A page that bows near a binding or a curled thermal slip keeps bent lines of text after the outline has been squared.
- **Proportions can change.** A badly tilted shot can come out slightly too tall or too wide. That matters if anyone will measure from the image, as with a drawing or a plat.

A page that still sits at a slant after the app has finished has ordinary skew, which is a separate fault. The post on [how to OCR a crooked scan](https://www.schemalaw.com/blog/how-to-ocr-a-crooked-scan) covers what a remaining tilt does to recognition and how to check for it.

The setting to change here is your grip, not a switch. Hold the phone so the outline on screen is already close to a rectangle. Then the correction has little to do and little to hide.

## Automatic shutter or manual capture for a stack of pages

Use the automatic shutter for a stack of same-size, single-sided, clean sheets on a contrasting surface. Switch to manual capture for mixed sizes, double-sided originals, bound pages, glossy paper and anything where you need to see the outline settle before the picture is taken.

Auto capture fires as soon as the app believes it has found a page and the phone is steady. Apple's [guide to scanning on an iPhone or iPad](https://support.apple.com/en-us/108963) says that in Auto mode "your document will automatically scan", and that you can press the shutter or a volume button "to manually capture a scan". Google's guides to the [Drive scanner](https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid) and the [Files by Google scanner](https://support.google.com/files/answer/15598575?hl=en) both describe a switch between Auto capture and Manual.

### What auto capture gets wrong in a long run

| Fault | How it happens | How you catch it |
| --- | --- | --- |
| Doubled page | The app fires twice while you are still reaching for the next sheet | Page count is higher than the paper count |
| Skipped page | You swap two sheets during one pause | Page count is lower; a page number is missing |
| Half-turned page | The shutter fires while the sheet is moving | Motion smear, or your hand across the text |
| Wrong outline accepted | No chance to look before capture | Cropped margins found only afterwards |

### A routine for a stack in manual mode

1. Count the sheets and write the number down.
2. Lay the first sheet on the dark surface and square it to the phone.
3. Wait for the outline to sit on all four paper corners.
4. Press the shutter, or a volume button if the app allows it, so the phone does not dip.
5. Slide the next sheet into the same spot without moving the phone.
6. At the end, compare the app's page count with the number you wrote down.
7. Open the first, middle and last pages at full zoom and read the smallest print on each.

For twenty uniform pages, auto mode with the count check at the end is a reasonable trade. For a file with exhibits, tabs, half sheets and stapled receipts, manual capture keeps you in charge of what each image contains.

## Image quality and file size options worth changing

Set image quality to the highest option the app has, leave any reduce-size or compress switch off, and save a multi-page document as a PDF. If the app can keep the original photo beside the processed scan, turn that on. Storage is cheap and a page you cannot retake is not.

What a scanning app exposes here depends on the app. Apple's [scanning instructions for iPhone and iPad](https://support.apple.com/en-us/108963) describe no quality setting at all. Open your app's settings screen and look for controls of these kinds.

| Kind of option | What it trades | Suggested position |
| --- | --- | --- |
| Quality or resolution (low, medium, high, or a pixel size) | Detail against file size | Highest |
| Compression or "optimize for sharing" | Smaller files against smudged letter edges | Off for anything going to OCR |
| Save original image | Storage against a fallback copy | On |
| Default filter | Convenience against lost marks | Color, or whatever leaves the page unprocessed |
| Output format | One file per document against one file per page | PDF for documents |
| Auto upload or sync | Convenience against control of where the file goes | Decide deliberately; see the firm's policy |

### PDF or JPG

Google's [instructions for scanning with Drive](https://support.google.com/drive/answer/3145835?hl=en&co=GENIE.Platform%3DAndroid) say that to save the scan in a specific format you "select either .pdf or .jpg". Its [instructions for scanning with Files by Google](https://support.google.com/files/answer/15598575?hl=en) say scanned documents are "automatically saved as PDFs". A PDF holds every page of the session in order, which suits a contract or a set of records. A JPG is one picture, which suits a single sheet. Both Google guides also say the PDFs they save are searchable. Neither says how the text is produced, so read it against the page before you rely on it. The post on [whether a free OCR scanner puts text in the file](https://www.schemalaw.com/blog/free-ocr-scanner) shows how to test a phone scan for a text layer.

### Size against an upload limit

High quality pages add up. If the finished file is going to a converter, check its limit first. Schema Law accepts PDF, PNG and JPEG files up to 20 MB, and the [pricing page](https://www.schemalaw.com/pricing) gives the page allowance per document on each plan. When a scan is too big, split it into parts in page order. Lowering the quality setting to squeeze under a limit removes the detail that recognition needs.

### The untouched photo

Whatever the app produces is a processed copy. If the app has no option to keep originals, take one plain camera photo of any page that carries a signature, a seal or handwriting before or after you scan it, and file it with the scan. Then if a filter, a crop or a cleanup tool has altered something, you have a picture that shows the paper as it was. Convert whichever copy reads cleanly at full zoom, and keep the plain photo whichever you choose. The [export formats](https://www.schemalaw.com/#formats) you choose afterwards do not change either image.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)