How to Turn a Phone Photo of a Document into a PDF: A Step-by-Step Document Scanning Guide

You have a contract, an invoice, a petition or a signed form in front of you, and the other party wants it "as a scanned PDF." But there's no scanner at home, there's a queue at the office, or the document is already with you, in your hands. What most people do is obvious: take a photo with their phone and send it. And the result is usually the same — a crooked, yellowish photo with a corner of the desk showing and your own shadow lying across it.

The document scanner tool exists precisely to close that gap. It takes a document photo shot with a phone, finds the edges of the paper, corrects the perspective, cleans up the shadow, and gives you a PDF that genuinely looks like it came out of a scanner.

In this guide I'll walk through how to use the tool step by step, which filter to pick when, and which small mistakes you make while taking the photo end up directly affecting the result.

What does the tool do behind the scenes?

Before we get to the steps, knowing briefly what happens makes choosing settings easier. Every photo you upload goes through the following stages in order:

1. Edge detection: The contour structures in the image are analyzed in an attempt to find the four corners of the paper.
2. Perspective correction: The four corners that were found are pulled into a rectangle using a four-point perspective transform. In other words, a page shot at an angle is flattened as if it had been photographed straight on.
3. Shadow removal: A blurred copy of the image is extracted to produce a "background map," and then the original image is divided by that map. This brings the light falling on one side of the page and the shadow on the other side to the same level.
4. Filter application: Depending on the mode you select, contrast enhancement, grayscale conversion or thresholding is applied.
5. A4 normalization: Every page is fitted to A4 size; portrait shots become portrait A4 pages (595x842 points) and landscape shots become landscape A4 pages (842x595 points). That way photos taken from different distances end up as consistent pages.
6. Text recognition (OCR, optional): If you picked a language, an invisible text layer is added to the finished PDF.

Step by step: turning a photo into a PDF

Step 1: Take the photo properly

This step is the most important one, because it sets the limit of what the tool can do. The tool can make a good photo perfect; a bad photo it can only rescue up to a point.

• Use a contrasting surface. If you put white paper on a white desk, edge detection struggles. Placing it on a dark table, a carpet or a book cover makes detection noticeably easier.
• Get the whole sheet into the frame. If one corner falls outside the frame, the four corners can't be found.
• Leave a little space around it. If you frame the paper right at its edges, there's no margin left for the contour algorithm to work with.
• Don't cast your own shadow onto it. Shadow removal balances gradual differences in light well, but the sharp, black shadow of your hand or your phone can blend into the text on the paper.
• Flatten the paper as much as you can. The bulge in the middle of a folded page won't be fully corrected by the perspective transform.

Step 2: Upload the photos to the tool

On the Document Scanner page, drag and drop your files onto the upload area or pick them from the file selector. For a multi-page document you can upload photos of all the pages at once; they all become pages of a single PDF.

When picking from your phone gallery, don't assume the file names follow the order in which you shot them; you'll be checking the ordering in the next step anyway.

Step 3: Arrange the page order

You can change the order of the photos you uploaded before starting the process. This matters especially with multi-page contracts or reports: a signature page landing in the middle of the document means reproducing the whole file.

A practical habit: shoot the pages in order from start to finish, then just do a quick check before uploading.

Step 4: Choose the filter mode

There are four modes, and the choice depends on the type of document:

• Document (default): Shadow removal, contrast enhancement and sharpening are applied together. This is the right choice for the vast majority of pages containing written text. The result is dark, legible text on a white background.
• Grayscale: Color information is removed, the shadow is still cleaned up, but the contrast is kept gentler than in document mode. That's what preserves the gray tones: on documents containing photos, charts or shaded figures it retains far more detail than black & white does. If you want no shadow removal at all, original is the only mode that skips it.
• Black & white: Thresholding is applied; every pixel becomes either black or white. This mode gives the smallest file size. It's ideal for plain-text documents destined for an archive. However, on pages containing photos or shaded drawings you'll lose detail.
• Original: The color image is left untouched; only edge detection and perspective correction are applied. Choose this mode when color information matters — a colored stamp, the color of a signature, color-coded tables, passport or ID images.

If you're not sure, start in document mode; you can look at the result and try again with a different mode if needed.

Step 5: Start the process and check the result

Processing happens on the server, and the files are deleted afterwards. Once you've downloaded the output, look at these three things:

• Is the crop right? Are the page edges clean, or is there still some desk in a corner?
• Is the text readable? Especially in black & white mode, make sure fine print hasn't disappeared.
• Is there shadow left? If there's still a visible gray band along one edge of the page, you may need to retake the photo under more even lighting.

What to do if you want the text to be searchable

There's no separate step for this: before you start the process, pick a language from the Text recognition (OCR) dropdown on the tool page. The options are "Don't add" (the default), "Turkish + English", "Turkish" and "English". Pick one and an invisible text layer is added to the output PDF; the page image stays exactly as it is, but the document becomes searchable and its text can be copied.

The reason the default is "Don't add" is time: text recognition noticeably lengthens the job. If all you want is a clean-looking PDF, leave the setting alone; for documents you'll archive or search through, picking a language is worth it.

The order in which correction and recognition happen isn't a coincidence either: recognition runs last, on a page whose perspective has already been corrected and whose shadow has already been cleaned up. On a crooked, shadowed, low-contrast image recognition errors go up; a corrected page is a far better input. If you already have a properly scanned PDF and only want to add a text layer to it, the OCR tool does exactly that on its own.

Practical tips

• Making the text as legible as possible is the real goal. What matters is that the other party can read the document without trouble, rather than an aesthetically pleasing result. When in doubt, pick document mode.
• Consider black & white if you're archiving. If you're storing a file of dozens of pages, black & white mode shrinks the file size noticeably. If you want to shrink it further, you can run the output through the PDF compression tool.
• Use original mode if there's a colored signature or stamp. Some institutions want to see that the signature is blue; black & white mode destroys that information.
• Upload the pages all at once. Instead of making each page a separate PDF and merging them afterwards, uploading them together both keeps the page sizes consistent and saves time. If you shot in mixed orientations, portrait photos produce portrait pages and landscape photos produce landscape ones; if you want them all facing the same way, shoot them the same way.
• You don't need this tool if you already have properly scanned images. If the images you have are already flat and clean, going straight to the image to PDF tool is a faster route.

Common mistakes

Mistake 1: Putting white paper on a white desk. Edge detection relies on contour differences. If there's no visible difference in tone between the surface and the paper, the corners can't be found and the photo gets used as it is. A dark surface solves this problem almost every time.

Mistake 2: Leaving a corner of the paper outside the frame. If all four corners aren't visible, the perspective transform can't be performed. Before taking the photo, check that all four corners appear on screen.

Mistake 3: Choosing black & white mode for a document with photos. Thresholding removes gray tones entirely. On a page containing an ID photo or a chart, this mode can render the content unrecognizable. For pages like that, go with grayscale or original mode.

Mistake 4: Leaving the OCR dropdown on its default and then being surprised you "can't search." Text recognition is off by default. If you want a searchable PDF you have to pick a language before starting the job; noticing afterwards means processing the file all over again.

Mistake 5: Sending the result without checking it. Especially with official applications, a single page that ended up badly cropped or illegible can cause the whole file to be rejected. Opening the PDF you downloaded and looking through it page by page takes a few seconds.

Conclusion

The difference between a document photo taken with a phone and a scanned document really comes down to a handful of image processing steps: find the edge, correct the perspective, even out the shadow, adjust the contrast. The document scanner performs these steps for you, but the quality of the result still depends largely on how you took the photo. A contrasting surface, a sheet of paper fully inside the frame and shadow-free lighting let the tool do the best it can. And if the text needs to be searchable, don't forget to pick a language from the text recognition dropdown before you start.