Learn how to convert files without losing quality. Covers lossless conversion techniques, optimal settings for video, audio, images, and documents, plus common mistakes that destroy quality.
You have a high-quality video, a crisp photo, or a crystal-clear audio recording, and you need to convert it to a different format. The moment you click "convert," a question hangs in the air: will the output look (or sound) as good as the input?
The honest answer is: it depends entirely on how you do it. Some conversions are inherently lossless -- zero quality loss, guaranteed. Others involve lossy re-encoding where quality loss is unavoidable, but the amount of loss depends on your settings. And some common mistakes can destroy quality unnecessarily.
This guide explains the science behind quality preservation, gives you the optimal settings for every major file type, and teaches you to spot and avoid the mistakes that silently degrade your files.
Some conversions involve no quality loss whatsoever. These operations change the container or metadata without touching the actual media data:
| Operation | Example | Why It's Lossless |
|---|---|---|
| Container change (remux) | MKV to MP4 (same codec) | Only the wrapper changes, not the data |
| Metadata editing | Changing title, tags, chapters | Media streams are untouched |
| Lossless-to-lossless | FLAC to ALAC, PNG to WebP lossless | Both formats preserve all data |
| Stream extraction | Extracting audio from video without re-encoding | Original stream copied bit-for-bit |
| Format rename | .jpeg to .jpg (literally the same format) | No conversion happens at all |
For video, the remux operation is incredibly useful. If you have an MKV file with H.264 video and AAC audio, you can convert it to MP4 with zero quality loss:
ffmpeg -i input.mkv -c copy output.mp4
The -c copy flag tells FFmpeg to copy the streams without re-encoding. This is near-instant and produces a bit-identical output. The MKV to MP4 conversion guide explains when this works and when you need to re-encode.
When you must re-encode (change codec, resize, apply filters), quality loss is inherent in the process. But the amount of loss is under your control:
The key insight is that one generation of lossy compression at high quality settings is virtually indistinguishable from the original. Problems arise when you re-compress multiple times (generation loss) or use unnecessarily aggressive settings.
For H.264 encoding, the CRF (Constant Rate Factor) value is the single most important quality control:
| CRF Value | Quality Level | SSIM (Structural Similarity) | Recommended For |
|---|---|---|---|
| 0 | Mathematically lossless | 1.000 | Archival only (huge files) |
| 15-17 | Visually lossless | 0.990+ | Master copies, editing sources |
| 18-20 | Excellent (transparent) | 0.975-0.990 | High-quality distribution |
| 21-23 | Good (default) | 0.950-0.975 | General use, streaming |
| 24-28 | Acceptable | 0.900-0.950 | Web delivery, social media |
| 29+ | Noticeable degradation | <0.900 | Previews, thumbnails only |
The quality preservation formula:
# Maximum quality preservation (visually lossless)
ffmpeg -i input.mov -c:v libx264 -crf 17 -preset slow -c:a aac -b:a 256k output.mp4
# High quality (imperceptible loss for most content)
ffmpeg -i input.mov -c:v libx264 -crf 20 -preset medium -c:a aac -b:a 192k output.mp4
For a complete explanation of how CRF, bitrate, and preset interact, the video bitrate explained guide covers the theory in depth.
Pro Tip: Use -preset slow or -preset slower when quality preservation is the priority. Slower presets produce the same visual quality at a smaller file size, or equivalently, better quality at the same file size. The encoding takes longer, but the result is worth it for important content.
One of the most common quality-destroying mistakes is upscaling -- converting a 720p video to 1080p or a 1080p video to 4K. Upscaling does not add detail; it just makes the same pixels larger, increasing file size without any quality improvement. In fact, the re-encoding introduces additional artifacts.
Rules for resolution:
# Correct: downscale 4K to 1080p
ffmpeg -i 4k_video.mp4 -vf "scale=1920:1080" -c:v libx264 -crf 20 output.mp4
# Wrong: upscaling adds no quality
ffmpeg -i 720p_video.mp4 -vf "scale=1920:1080" output.mp4 # DON'T DO THIS
Every time you decode and re-encode a lossy file, quality degrades slightly. This is called generation loss. A single re-encoding at CRF 20 is virtually transparent. But re-encoding the same file five times accumulates visible degradation.
Best practices:
The lossless vs lossy compression guide explains generation loss in detail.
Converting between lossless formats preserves quality perfectly:
# FLAC to ALAC (lossless to lossless -- zero quality loss)
ffmpeg -i input.flac -c:a alac output.m4a
# WAV to FLAC (uncompressed to lossless -- zero quality loss, smaller file)
ffmpeg -i input.wav -c:a flac output.flac
When converting to a lossy format (MP3, AAC, OGG), use the highest quality settings appropriate for your use case:
# Highest quality MP3 (320 kbps CBR -- overkill for most listeners)
ffmpeg -i input.wav -c:a libmp3lame -b:a 320k output.mp3
# High quality MP3 (V0 VBR -- recommended for music)
ffmpeg -i input.wav -c:a libmp3lame -q:a 0 output.mp3
# High quality AAC (256 kbps -- better than MP3 at same bitrate)
ffmpeg -i input.wav -c:a aac -b:a 256k output.m4a
For the WAV to MP3 conversion, 192-256 kbps is the sweet spot where quality is indistinguishable from the original for most listeners.
Converting one lossy format to another lossy format always degrades quality, even at high settings. Converting MP3 to AAC does not improve anything -- it just adds another generation of lossy compression on top of the first.
If you must change lossy formats (e.g., OGG to MP3 for device compatibility), use the highest bitrate possible and accept that some quality will be lost.
# PNG to WebP lossless (smaller file, identical quality)
ffmpeg -i input.png -lossless 1 output.webp
# BMP to PNG (uncompressed to lossless -- zero loss)
ffmpeg -i input.bmp output.png
JPEG uses a quality scale from 1 to 100. Here is how the settings affect output:
For web images, the image compressor automatically finds the optimal quality level. The optimize images for website guide covers the complete optimization workflow.
RAW files from cameras are not images -- they are sensor data. Converting RAW to JPEG or PNG is a developing process (like developing film), not a simple format conversion. Use dedicated software:
Export at the highest quality settings to TIFF (for further editing) or high-quality JPEG/WebP (for delivery).
Pro Tip: When saving JPEG files that will be edited further, use quality 95 or higher. Every time you open, edit, and re-save a JPEG, it is re-compressed, and the artifacts compound. For files that will go through multiple edit cycles, use PNG or TIFF during editing and only convert to JPEG at the final export stage.
When converting Word documents to PDF, the main quality concern is font rendering and image handling:
The Word to PDF converter handles font embedding and image optimization automatically.
PDF compression reduces file size but can affect quality:
When using the PDF compressor, choose the quality preset carefully:
JPEG to PNG to JPEG. MP3 to AAC. AVI (with Xvid) to MP4 (with H.264). Each step adds a generation of quality loss.
Fix: Always convert from the highest quality source. If you only have a lossy file, convert it once to the final format at high quality settings.
Converting 720p video to 1080p, enlarging a 800px image to 4000px, or increasing audio sample rate from 44.1 kHz to 96 kHz. None of these add quality -- they just increase file size.
Fix: Match or reduce resolution. Never increase it.
Default settings are designed for general use, not quality preservation. Most online converters default to medium quality to keep file sizes manageable.
Fix: Explicitly set quality parameters. For video, use CRF 17-20. For images, use quality 85-95. For audio, use 256-320 kbps.
If you are converting MKV to MP4 and both use the same codecs, re-encoding is unnecessary and wasteful:
# Wrong: re-encodes everything (quality loss, slow)
ffmpeg -i input.mkv -c:v libx264 -c:a aac output.mp4
# Right: copies streams (lossless, instant)
ffmpeg -i input.mkv -c copy output.mp4
Fix: Use -c copy whenever the target container supports the source codecs.
Taking a screenshot of a document, video frame, or image and using that as your "conversion." Screenshots capture at screen resolution, which is typically far lower than the original file's quality.
Fix: Use proper conversion tools, not screenshots.
After conversion, verify that quality has been preserved:
# Compare SSIM (Structural Similarity) between original and converted
ffmpeg -i original.mp4 -i converted.mp4 -lavfi "ssim" -f null -
# Compare PSNR (Peak Signal-to-Noise Ratio)
ffmpeg -i original.mp4 -i converted.mp4 -lavfi "psnr" -f null -
Open the original and converted images side by side at 100% zoom. Pay attention to:
Listen to the original and converted audio with good headphones. Focus on:
| Conversion | Lossless Possible? | Best Quality Setting | Tool |
|---|---|---|---|
| MKV to MP4 | Yes (remux) | -c copy | MKV to MP4 |
| MOV to MP4 | Often yes |
| MOV to MP4 |
| AVI to MP4 | Rarely (different codecs) | CRF 18, preset slow | AVI to MP4 |
| PNG to WebP | Yes (lossless WebP) | -lossless 1 | WebP Converter |
| JPEG to PNG | Preserves current state | Direct conversion (no re-compression) | PNG Converter |
| WAV to FLAC | Yes | Any FLAC setting (all lossless) | FLAC Converter |
| Word to PDF | Yes (for text/vector) | Embed fonts, 300 DPI images | Word to PDF |
Quality preservation during file conversion comes down to three principles: use lossless operations whenever possible (remuxing, lossless-to-lossless conversion), use high quality settings when lossy re-encoding is necessary (CRF 17-20 for video, quality 85-95 for images, 256+ kbps for audio), and avoid the five common mistakes (format-to-format lossy conversion, upscaling, default settings, unnecessary re-encoding, and screenshot-based conversion). Always convert from the highest quality source, never delete originals, and verify quality after conversion.
Marcus Rivera
Systems engineer writing about video transcoding, hardware acceleration, and large-scale media processing.