Compress PDF made simpler, in your browser

A smaller attachment is useful only if the recipient can still do what they need with it. A searchable report, a bookmarked manual, and a photograph of a receipt have different requirements. This tool offers two methods. The default optimizes internal compression while retaining text and document objects. The image-only method rebuilds visible pages as JPEG images with resolution and quality settings you choose.

Start with the default when search, copying, bookmarks, links, or fillable fields matter. It does not lower image resolution or deliberately change page appearance, so its size saving can be modest. Choose image-only output when you specifically need a visual scan copy and accept losing interactive and text features. The interface reports actual source and output sizes, including cases where processing does not make the file smaller.

Choose your PDF and leave the compression method set to preserve text, bookmarks, links, and forms. Process the document, compare the reported sizes, and inspect the first-page comparison. The tool checks page count, visible boundaries, rotation, extracted text positions, annotation details, form values, and bookmark targets before offering the optimized download. Those checks help catch an unexpected change in the features people commonly need.

Reopen the downloaded copy in your usual reader. Search for a phrase, copy a representative paragraph, and follow a chapter bookmark or link if the source contains one. Check a filled field when working with a form. Important documents still deserve a review in the software that will use them. Keep the original separately so that you can compare another result or return to the source if an unusual viewer behaves differently.

The default method does not render the document into lower-resolution page images. It optimizes PDF stream compression and uses a compact representation for internal objects. Existing text remains text, vector drawings remain vector drawings, and images are not deliberately resized or converted to lower-quality JPEGs. A clear scan should therefore not become softer simply because you ran this mode.

That also explains why it cannot promise a dramatic reduction for a large photo-heavy file. If the source images already use efficient compression, optimizing surrounding objects may save little space. This method does not invent a smaller high-quality version of every image. Judge it by the actual size summary and the features preserved, rather than by expecting the same percentage saving for every kind of document.

Preserving these features is the purpose of the default mode. An existing OCR layer should remain searchable, bookmarks should point to the same physical pages, and links and normal interactive form values should remain available. The processing path avoids a whole-page image conversion and checks common structural features after optimization. It does not automatically add a missing OCR layer or create new navigation.

The image-only mode behaves differently. It creates page pictures and omits original text layers, active links, form fields, bookmarks, and structured reading information. A visible filled form value may still appear in its page image, but the field is no longer editable. Read the selected method before processing, particularly when passing a newly searchable scan or a document with a clickable contents page into this tool.

Use it when a visual copy meets the task and you need control over page image size. For example, you might prepare a receipt scan for a upload form that does not need selectable text. Switch the method explicitly, choose resolution and JPEG quality, and acknowledge the loss of document features. Optional grayscale conversion is available for pages where color does not communicate useful information.

The preview compares the first original page with the first output page. Do not stop there: download and check representative later pages, including dense tables, fine print, handwriting, and stamps. A large cover heading may remain clear while a small footnote becomes unreadable. If that happens, process the original again at higher resolution or quality. Repeatedly compressing the same output can compound damage without recovering detail.

DPI controls how many pixels represent the physical page in image-only mode. Available settings are seventy-two, one hundred, one hundred fifty, and two hundred DPI. More pixels can preserve smaller detail but increase memory use and file size. Rendering an already soft scan at a higher DPI adds pixels without creating missing information. Use a setting appropriate to the text and drawings you actually need to read.

JPEG quality controls how aggressively those pixels are encoded. The allowed range is fifty to ninety-five. Lower values can create blockiness, halos around letters, and blurred punctuation. Raising quality cannot repair a resolution too low to capture a thin stroke. Change one setting at a time when comparing alternatives. These controls apply to image-only output and are not used by the default structure-preserving optimizer.

No fixed output size is guaranteed. Documents differ in page count, image content, fonts, and existing compression. Check the actual requirement of the destination, then compare it with the generated file. If the default mode does not reach your target, decide whether an image-only copy is acceptable before trying lower resolution or JPEG quality. Do not sacrifice necessary readability solely to reach a number.

When the size target conflicts with clarity or required features, consider removing genuinely unnecessary pages or splitting the document into appropriate sections. Confirm that the receiving system accepts the resulting files. This tool does not know a particular email provider's attachment limit or an upload form's validation rules. The result summary gives measured bytes and percentage change, which are more useful than a promised universal compression ratio.

An already compact file may have little remaining redundancy. Rewriting it can add a small amount of structural overhead, even in the default mode. The image-only method can grow much more: compact text and vector drawings may need a large number of pixels when converted into full-page photographs. A monochrome scan may also use a compression scheme more efficient for its content than JPEG.

The tool reports no size saving when that is what happened. Keep the original if it better serves your task. The optimized copy is still offered for inspection, but a completed operation is not presented as evidence that the result is smaller. Compare actual size, readability, and document behavior together. A difference of only a few bytes rarely justifies replacing an otherwise reliable original.

Grayscale is an optional setting in image-only mode. It can suit black text on plain paper and may reduce the visual information the encoder needs to store. It does not guarantee a saving, automatically whiten a gray background, or create a specialized one-bit archival scan. The output remains a JPEG representation of each page.

Keep color when it identifies chart series, highlights revisions, distinguishes stamps, or carries information in photographs. Different colors can become similar gray shades and make the result harder to interpret. Inspect a relevant color page before accepting the output. For scan brightness, contrast, or small-angle correction, use the scan adjustment controls in the OCR tool and review those changes separately.

Choose an unencrypted, unsigned working copy. Signature fields, certified documents, and XFA forms are unsupported. Rewriting a signed document is not a way to keep a valid digital signature, so such files are rejected. For a password-protected source you are authorized to open, use the known-password unlock tool first, then review the unlocked copy before optimizing it.

Normal supported forms can use default optimization, but check their values and appearance after download. Image-only mode requires explicit acceptance of feature loss. Specialized scripts, calculation behavior, and unusual rendering features should be reviewed in the receiving application; ordinary page previews do not certify those behaviors. Keep an authoritative original if the document has requirements beyond making a practical working or sharing copy.

One input file can be up to 30 MB and one hundred pages. Output is limited to 120 MB, and image rendering is bounded to twelve million pixels per page. Large physical pages at high DPI may exceed that pixel limit even when the source file is small. Select a lower resolution for a suitable scan or split a large document into sections when necessary.

Processing runs in your browser without uploading the selected document to a conversion service. You can cancel and retry. Changing settings removes the previous download until you process again. Before keeping a result, reopen it, verify its page count and visible content, and test the specific text, navigation, or fields needed for your task. Retain the source as the starting point for future revisions.

Yes. Select two through eight documents together to use the sequential batch queue. Your compression method and any image settings apply to each file. Each source is checked separately, so a damaged, encrypted, or unsupported file produces its own failure message while later files continue. The original files remain on your device. Inputs are capped at thirty megabytes per file and one hundred twenty megabytes per batch; successful outputs together are capped at one hundred twenty megabytes.

After the queue finishes, download each successful result or a ZIP containing the results and a manifest.json report. The report records individual sizes and outcomes, including a document that did not become smaller. A small optimized file can already be efficiently compressed, so no fixed reduction is promised. The same image-only consent applies to all documents in that batch; use separate batches when some files must preserve text and others may become images. Duplicate file names are disambiguated in the archive. Canceling discards queue results, and changing settings removes the previous download so that you can run the queue again with the intended options.

Choose a tool for the next step and review each output.

• Split a large document
• Remove unnecessary blank pages