Two PDFs can show the same sentence and behave differently when you try to select it. One has text objects. The other has a picture of the words. That difference matters when you extract pages, turn them into images, or combine the images into a new PDF.
The .pdf extension doesn't tell you which one you have. A document can also mix both kinds of page, or include an OCR text layer over a scanned image.
Compare two small documents
Start with the text sample. It has the sentence “Search for lantern. This sentence is stored as PDF text.” Open it in a PDF viewer and search for lantern. Try selecting the words too.
The image-only sample is a rasterized copy of that page. It was generated for this comparison, not scanned from a physical sheet. Its words are pixels, with no PDF text layer behind them. It can look readable while giving the viewer nothing to select or search as text.
This is a good first check on your own document, but it won't tell you everything. A viewer may have its own OCR feature, and a searchable scan can still have recognition errors. Try a distinctive word on more than one page before deciding what the whole document contains.
Keep pages as PDF when you need their text
Use Split PDF to keep selected pages in PDF form, or Merge PDFs to combine documents. These tools copy PDF pages instead of rendering every page as an image. Text on the source page stays available as page content.
Copying pages doesn't create text that was missing. Combining several scans won't make them searchable by itself. Don't assume every interactive form or document-level feature survives in the new file either. Check the saved PDF in the software that needs to use it, especially if the original had forms, bookmarks, or signatures.
Rendering turns the page into pixels
Open the text sample in PDF To Images. Select PNG, set Resolution to 72 DPI, and export the ZIP. The file inside is page-001.png. For this US Letter sample, that export is 612 by 792 pixels. At 144 DPI, each side doubles to 1,224 by 1,584, so the image has four times as many pixels.
Those dimensions follow the page size and rendering resolution. A differently sized page produces different output. Higher DPI can make small lettering clearer in the raster copy, but it also gives the browser more pixels to hold and encode. It doesn't recover detail from a blurry scan.
PNG keeps the rendered pixels without another lossy image encoding step. JPG or WebP may be useful for smaller delivery copies, but inspect lettering and thin lines before accepting the result. The selected PDF pages are exported as images in a ZIP. The ZIP is packaging, not a PDF text layer.
Putting the images back doesn't restore the text
Load the exported PNG into Images To PDF. That tool embeds each selected image on a page, with controls for page size, orientation, margins, and order. The result is a valid PDF with the image you added.
It doesn't run OCR. The sentence is still visible, but the original text objects aren't rebuilt from the pixels. The image-only sample shows the same thing. Search and selection depend on real text content, not on whether the output file ends in .pdf.
If your next step requires a searchable document, keep the original text PDF and use the page tools where possible. If all you have is a scan, OCR is a separate job. Either way, open the saved copy, check its page order and size, and test the features you need before replacing your source.