Strictly speaking, you cannot make a video file much smaller without losing some quality - but you can compress it so the loss is invisible. The most effective approach is to re-encode with a more efficient codec (H.265 or AV1 instead of H.264), use a constant quality setting such as CRF rather than a fixed bitrate, and use a slower encoder preset. Reducing resolution or frame rate shrinks files further, but only when the extra detail is not needed.
This guide explains what "without losing quality" really means, which settings give the biggest savings, and how to check that your compressed file still looks as good as the original.
Lossless vs visually lossless
True lossless compression (for example FFV1 or lossless x264) keeps every pixel identical, but the files are usually far larger than the camera or download you started with, because those sources were already lossy. So for everyday compression, the realistic goal is visually lossless: a file where you cannot see a difference at normal viewing distance, even though the data is not identical.
Every lossy re-encode throws away a little information, and the losses add up across generations. Always compress from the best source you have, and keep the original until you have checked the result.
Where the savings come from
Many videos are much larger than they need to be. Phones, action cameras and screen recorders often record at high bitrates so they can encode in real time with little processing. A careful software encode can usually reach similar visual quality at a fraction of the size. The main levers are:
| Lever | Typical effect on size | Quality cost |
|---|---|---|
| More efficient codec (H.264 to H.265 or AV1) | Large | None visible if quality settings are sensible |
| Constant quality (CRF) instead of high fixed bitrate | Often large for high-bitrate sources | None visible at moderate CRF values |
| Slower encoder preset | Moderate | None - costs encoding time instead |
| Lower resolution (e.g. 4K to 1080p) | Large | Loss of fine detail on big screens |
| Lower frame rate (e.g. 60 to 30 fps) | Moderate | Less smooth motion |
| Re-encode audio (e.g. PCM to AAC) | Small to moderate | Usually inaudible at decent bitrates |
Step 1: Choose an efficient codec
H.265 (HEVC) typically produces files noticeably smaller than H.264 at similar quality, and AV1 is usually more efficient again. The catch is playback: older TVs, phones and browsers may not decode them, or may struggle with high resolutions. If the file needs to play everywhere, H.264 is still the safe choice. Our comparisons of H.264 vs H.265 and AV1 vs HEVC go into the details.
Step 2: Use constant quality, not bitrate
CRF (Constant Rate Factor) tells the encoder to aim for a consistent level of quality and use whatever bitrate each scene needs. Simple scenes get few bits; complex, fast-moving scenes get more. That is far more efficient than forcing the same bitrate on everything. Lower CRF values mean higher quality and bigger files. As a rough guide:
- x264 (H.264): CRF 18 is often considered close to visually lossless; 20 to 23 is a good balance. The default is 23.
- x265 (H.265): CRF values around 20 to 24 are common; the default is 28. x265 CRF numbers do not map directly to x264 numbers.
- SVT-AV1: the scale runs to 63; values in the high 20s to mid 30s are a common starting range.
A good starting point for compressing a typical video with H.265 in FFmpeg:
ffmpeg -i input.mp4 -c:v libx265 -crf 22 -preset slow -c:a copy output.mp4
If you need to hit an exact file size instead, use two-pass bitrate encoding. Our guide to what CRF is explains the scale in more depth.
Step 3: Use a slower preset
Encoder presets (ultrafast through to veryslow for x264 and x265) control how hard the encoder searches for efficient ways to compress each frame. Slower presets give smaller files at the same CRF, at the cost of encoding time. slow is a sensible choice for files you will keep; going beyond it brings diminishing returns.
Step 4: Consider GPU encoding carefully
NVIDIA NVENC, Intel Quick Sync and AMD's hardware encoders are much faster than software encoders. For maximum compression at a given quality, though, a slow software encode generally still produces smaller files. GPU encoding is a good choice when speed matters more than squeezing out every last megabyte.
Step 5: Only reduce resolution or frame rate if it makes sense
If the video will only ever be watched on a phone or a laptop, downscaling 4K to 1080p can cut the size dramatically with little visible difference on those screens. Similarly, a 60 fps talking-head video loses little at 30 fps. But for content watched on large screens, or fast motion such as sport, keep the original values.
Step 6: Leave the audio alone (mostly)
Audio is usually a small part of the file. If it is already AAC or Opus, copy it unchanged. If it is uncompressed PCM, as in many camera and editing exports, converting it to AAC at around 192 kbps for stereo saves space with no audible difference for most listeners.
Step 7: Check the result objectively
Do not judge quality from a small preview window. Compare the original and compressed file at full size, on the screen you intend to use, paying attention to:
- Dark scenes and gradients such as skies, where banding appears first.
- Fast motion and fine textures like water, grass and confetti.
- Faces and text.
For a more objective comparison, metrics such as VMAF, SSIM and PSNR score the compressed file against the original. VMAF in particular was designed to predict how viewers perceive quality. FFmpeg can calculate it when built with libvmaf:
ffmpeg -i compressed.mp4 -i original.mp4 -lavfi libvmaf -f null -
The two files must have the same resolution and frame rate for the comparison to be meaningful. Encoding a short sample at a few CRF values and comparing their scores and sizes is the quickest way to find the sweet spot for your footage.
-pix_fmt yuv420p10le with x265) often reduces banding in gradients, even from 8-bit sources. Make sure your playback devices support 10-bit HEVC first.Doing it with Movie Encoder
Movie Encoder's Quality Lab runs real test encodes of a sample at several settings, such as CRF 18 to 26, and compares PSNR, SSIM, VMAF (when the FFmpeg build includes libvmaf), file size and encoding time with charts and a side-by-side A/B view, so you can pick the smallest setting that still looks right and apply it. The Smart Encoding Advisor can also recommend settings for a "Smallest file" or "Balanced" goal, with an estimated file-size range.
Once you have settled on a setting, the batch queue applies it to whole folders, and your originals are never overwritten or deleted.
Frequently asked questions
Is it possible to compress video with no quality loss at all?
Only with lossless codecs, which usually make files larger than a typical camera or download. For everyday use the realistic goal is visually lossless, where you cannot see a difference.
What CRF value should I use?
For x264, around 18 to 23 is a common range, and for x265 around 20 to 24. Test a short clip at a few values, because the best setting depends on the footage.
Is GPU encoding good for compression?
GPU encoders are much faster, but slow software encodes usually produce smaller files at the same quality. Use the GPU when speed matters more than the last few megabytes.
