A compressed PDF looks wrong when the saved file does not match the size you asked for, the embedded images appear blurry or blocky, or the tool produced no file at all — and the reliable fix is to discard the bad output and restart with a tool that only releases a download after the verified file size is at or below your target. Most "wrong-looking" results trace back to one of three things: the compressor estimated size from the source rather than measuring the saved output, it re-encoded images with no quality floor, or it over-promised and silently shipped a file that missed the limit. Compress PDF to Size is built around the opposite rule — it tries a bounded sequence of JPEG quality and maximum-dimension profiles, saves each candidate, compares the actual byte length to the target, and only exposes a download when the real saved file is at or below the requested MB value. If no candidate passes, no file is offered and the reason is shown instead. That contract is what makes it the right tool for fixing a result that looks wrong, because the next attempt either delivers a file that genuinely fits the limit or tells you honestly that the target is unreachable from the current source.
What "Looks Wrong" Usually Means After PDF Compression
Most complaints about a compressed PDF fall into four failure modes, and each one needs a different fix. Pinpoint which one you are seeing before you restart, because the recovery workflow differs by symptom.
| Symptom | What you see | Typical cause | What fixes it |
|---|---|---|---|
| Wrong size | The downloaded file is larger than the target you typed in | Tool measured the source, not the saved output | Use a target-aware tool that compares the saved bytes to the target |
| Blurry images | Photos look soft, jagged, or full of JPEG noise | Image quality dropped too low during re-encoding | Restart with bounded quality profiles that stop at the first fit |
| No download at all | The tool returns nothing or a "target unreachable" message | Source is text-only, vector-heavy, or dominated by unsupported images | Lower the target, or switch to splitting the PDF or removing images |
| Broken page structure | Text layers disappear, links die, forms vanish | Tool rasterized whole pages or rewrote objects it shouldn't touch | Use a tool that only swaps eligible JPG streams and leaves the rest of the page alone |
The first two rows are the most common, and both come from tools that promise a target but cannot prove they hit it. The last two are usually a sign that the compressor tried to do too much: it rewrote objects it did not understand, or it gave up without telling you why.
Why a Verified, Target-Aware Tool Is the Fix
The reason "wrong" results keep happening is that most PDF compressors compare the source size to your target and report success based on a ratio, not on the bytes of the file they actually saved. If their pipeline drops three images and re-encodes the rest, they hand you whatever lands on disk and call it a day. You only find out the file is still over the limit — or looks worse than you expected — when you open it.
Compress PDF to Size takes a stricter path. You set a target between 0.1 and 25 MB that is smaller than the source. The tool runs through four fixed profiles — 0.70 quality at a 1400 px longest side, 0.55 at 1000 px, 0.40 at 800 px, and 0.30 at 600 px — saving a fresh PDF at each step and measuring the real byte length against your target. The first profile whose saved file meets the target wins. If none of them do, the tool does not guess or fabricate a result: it tells you no candidate passed and offers no download.
Two safety rules sit on top of that loop. First, the tool only swaps JPG image streams it can safely replace — unmasked DCTDecode DeviceRGB or DeviceGray streams with no decode parameters — so anything it does not understand, including fonts, vectors, forms, annotations, and masked or CMYK images, is left alone. Second, if a candidate JPG would land larger than the original stream, that image is left untouched, so the tool never makes an image bigger in the name of saving bytes. Page count is also re-checked against the source before a download is exposed, which protects against silent truncation.
How to Fix the Result With Compress PDF to Size
The recovery workflow is short because most of the decisions are made by the tool. Treat the bad file as disposable, keep your original, and start over with the verified-output path.
- Open Compress PDF to Size in your browser. The job runs locally — PDF reading, image decoding, encoding, measurement, and verification all happen on your device, so the original never leaves your machine.
- Choose the original local PDF from disk. The source limit is 25 MB and the tool caps eligible images at 80 and 100 megapixels across the images it decodes, which is enough for typical reports and slide decks but will surface a clear message if your document is outside that envelope.
- Enter a target size in MB between 0.1 and 25 that is smaller than the source. Pick the smallest target that still leaves room for the document type you have — a JPG-heavy scan can usually go much lower than a text-only PDF, vector-heavy report, or file dominated by unsupported images.
- Select Compress to target size. The tool walks through the four quality and longest-side profiles in order, saves a candidate PDF at each step, and compares the actual bytes to your target.
- Wait for the download button to appear. If it appears, the saved file is at or below your target by construction; the button only exists when the bytes check out. Click to download.
- Open the saved file and compare it page by page against the original. Confirm the page count matches, scan the images, and check that any links, form fields, or selectable text you depend on still work. If something looks off, keep the original and try a higher target.
If the download button never appears, that is itself a result. It means none of the four bounded profiles produced a saved PDF at or below your target, and the tool would rather tell you that than hand you a file that silently misses the limit. The next section covers what to do in that case.
What the Tool Changes and What It Preserves
Because the tool only swaps a narrow set of image objects, "looks wrong" should not include structural changes. The table below summarizes what gets rewritten and what stays byte-for-byte identical in the output.
| Property | Handled by the tool | Notes |
|---|---|---|
| Eligible JPG streams | Re-encoded at lower quality or smaller longest side | Only unmasked DCTDecode DeviceRGB or DeviceGray streams with no decode parameters |
| Ineligible images | Left untouched | Masked images, CMYK, JPX/JPEG2000, and any image with decode parameters are not replaced |
| Page structure | Preserved | Page count, page boxes, content streams are not rewritten wholesale |
| Selectable text and fonts | Preserved | Text layers and embedded fonts are not rasterized |
| Vectors, forms, annotations | Preserved | Form fields, links, comments, and vector drawings pass through untouched |
| File size | Reduced only when a candidate passes | Verified against the saved bytes, not estimated from the source |
If you see one of the "preserved" rows broken in the output, the original file was probably already broken before recompression, or the tool could not complete the verification step and surfaced a clear message instead of a download.
When No Download Appears — and What to Try Next
No download is a feature, not a bug. It means the tool ran every bounded profile it is willing to run and none of them produced a saved PDF at or below your target, which is the honest answer to "why does this still look wrong after I compress PDF to size." Three shapes of source file usually hit this path:
- A text-only or near-text-only PDF with no large JPG streams to rewrite. With no eligible images, the tool has nothing to swap and the file size barely moves.
- A vector-heavy report or presentation. Vectors compress poorly and the tool does not rasterize them, so the file cannot drop to a small target.
- A document dominated by images the tool is not allowed to replace, such as masked JPGs, CMYK scans, or JPEG 2000 streams.
For a deeper read on why a saved file may not always land at your target, the guide Will a Compressed PDF Always Meet Your Target Size walks through the same logic. In practice, you have three options when no download appears:
- Raise the target by a few tenths of a megabyte and try again. A 12.0 MB target often passes where 10.0 MB does not, with no visible quality change.
- Split the document with Split PDF by Size so each part fits a hard limit, instead of forcing one file to carry the whole target.
- Strip image content from pages where it is not needed using Remove Images from PDF, then retry the recompression on the leaner source.
Alternative Recovery Paths if Compression Won't Reach Your Target
If the document genuinely will not compress to the size you need, the fix is to reduce what the file has to carry rather than to compress harder. Three local workflows cover most cases without leaving the browser:
- Split by size: For email attachments or upload forms with a hard cap per file, Split PDF by Size chops a source PDF into consecutive parts measured against a maximum serialized file size, so each piece lands under the limit without any recompression.
- Drop pages: If the size problem is really a "too many pages" problem, Delete PDF Pages trims the document to just the pages you need before you retry compression, so the target has less work to do.
- Remove images: For drafts or internal review copies where images are decorative, Remove Images from PDF creates a no-upload copy that replaces eligible embedded JPG image objects with white pixels, and you can then rerun the target-aware recompression on the lighter source.
Whichever path you take, keep the original PDF until you have confirmed the saved file opens cleanly, has the right page count, and still does what you need it to. A wrong-looking result is almost always a sign to start over from the source rather than to chain another compression on top of a broken file.
For a deeper look, see Look Up Paper Sizes for Multiple Files at Once.