Choose compression by file structure
Chanjify’s Compress category contains twelve focused tools: a general image compressor, dedicated JPG, PNG, WebP, AVIF, TIFF, and SVG routes, safe structural PDF compression, and package-aware DOCX, XLSX, PPTX, and ODS compression. Each route uses a format-appropriate engine. A name change or generic ZIP wrapper is never presented as genuine compression.
Choose the dedicated format page when you know the source type. The general image page is useful for mixed JPG, PNG, WebP, AVIF, TIFF, and SVG inputs because it detects the actual format and keeps it. Office documents and PDF have different structural risks, so they use purpose-built validation rather than passing through an image pipeline.
How the three presets should be used
Light prioritizes fidelity and conservative packing. Balanced is the default for common web, email, and collaboration workflows. Strong applies a more aggressive quality or palette choice where the format supports it. These labels describe encoder settings, not guaranteed reduction percentages; two files with the same extension can have completely different compressibility.
Start with Balanced, open the result, and check an area that is likely to reveal change. For photos, inspect hair, texture, text, and high-contrast edges. For transparent graphics, inspect gradients and alpha boundaries. For documents, confirm pages, sheets, slides, formulas, relationships, selectable text, and embedded media in the intended office or PDF application.
Real measurements and honest no-op results
Every compressor measures exact input and output bytes. A successful screen shows original size, compressed size, saved bytes, and percentage saved. Those values come from the generated output buffer, not from an estimate based on extension or quality. The resulting file receives a structural or decode check before it becomes downloadable.
If the candidate is not smaller, Chanjify reports Already optimized and does not offer a larger replacement. This happens often with tiny icons, well-tuned modern images, efficient ZIP packages, and PDFs dominated by compressed scans. Trying Strong can help in some image cases, but an honest no-op is safer than repeated generation loss or fabricated savings.
Image and SVG compression
Sharp handles raster image compression on Chanjify’s own server route. It validates magic signatures, bounds decoded pixels, preserves dimensions, applies JPG, PNG, WebP, AVIF, or TIFF settings, and decodes the output again. Metadata is removed by default and can be retained with the visible option when color, camera, copyright, or workflow records are necessary.
SVG is different: it is vector markup, not a pixel grid. SVGO runs locally in the browser and removes redundant XML, editor metadata, groups, and numeric verbosity without rasterizing the artwork. More aggressive precision can affect geometry that depends on exceptionally small differences, so complex diagrams, maps, filters, or interactive graphics deserve a destination check.
Safe PDF and office compression
PDF compression uses a conservative structural rewrite with compressed object streams. It preserves page objects and selectable text rather than flattening every page into a picture. This approach cannot dramatically shrink a PDF dominated by already compressed images, but it avoids claiming lossy scan downsampling that the engine does not perform. Page count is validated after writing.
DOCX, XLSX, PPTX, and ODS are ZIP packages with strict internal relationships. Chanjify blocks traversal paths, excessive entries, oversized expansion, and suspicious compression ratios; requires the defining package files; optimizes supported embedded images; rebuilds the archive; and opens the result again. ODS receives its special uncompressed mimetype entry, while OOXML documents retain their content types and relationship structures.
Privacy matches the actual engine
SVG and PDF compression happen in browser memory and do not upload the selected file through these tools. Raster images and office packages use Chanjify’s own Next.js routes for dependable native encoding and container validation. Those routes work with in-memory buffers, return the result immediately, and do not intentionally save a permanent upload or conversion history.
No third-party file conversion API is involved. Normal hosting and network infrastructure can still process operational request data, so users handling regulated, secret, or unusually sensitive files should follow the responsible organization’s approved workflow. The Privacy page explains the paths separately rather than making one inaccurate promise for every tool.
Build a reliable compression habit
Preserve the source, choose the least aggressive preset that meets the real delivery limit, and open the result where it will be used. Do not repeatedly compress the same lossy derivative. When a target system permits a different format, a dedicated conversion may achieve a better result, but that is a separate decision and is exposed through related conversion tools rather than hidden inside compression.
For repeated work, test a representative source first and keep filenames explicit. A suffix such as “-compressed” makes the derivative obvious. Record measured savings and preset choice when teammates need reproducible output. Treat Already optimized as useful evidence, not a failed interface, and avoid deleting the master until visual and structural checks are complete.