Screen recording sounds simple: capture what's on screen, save to a file. But the format and codec choices during recording are different from normal video encoding because you're encoding in real-time while simultaneously using the computer for whatever you're recording. The wrong settings mean dropped frames, laggy performance, or corrupted files if the recording stops unexpectedly.
The right approach prioritizes reliability and low CPU impact during recording, then handles optimization (compression, format conversion) afterward. Here's what each recording tool outputs, why, and how to handle the files.
OBS Studio: Record to MKV, Remux to MP4
OBS Studio is the most popular recording tool for tutorials, gameplay, presentations, and general screen capture. It supports multiple output formats, but one matters most:
Record to MKV. This is OBS's own recommendation. Here's why:
- Crash safety: MKV files are recoverable if the recording stops unexpectedly (crash, power loss, accidental stop). MP4 files are not — if OBS crashes while recording to MP4, the entire file is corrupted and unplayable because the moov atom (file index) is written at the end.
- No limitation on streams: MKV supports unlimited audio tracks. OBS can record desktop audio, microphone, and application audio as separate tracks in one MKV file. MP4 technically supports multiple tracks but many players only see the first one.
- Same quality: MKV and MP4 support the same codecs. Recording to MKV instead of MP4 doesn't affect video or audio quality at all.
After recording: Use OBS's built-in Remux Recording feature (File → Remux Recordings) to convert MKV to MP4. This is a remux — it takes seconds regardless of file size and preserves quality perfectly. Or use ChangeThisFile's MKV to MP4 converter.
OBS's Fragmented MP4 option: OBS also offers "Fragmented MP4" which is crash-safe (because each fragment is self-contained). The downside: some editing software and players don't handle fragmented MP4 well. MKV remains the safer choice.
Recommended OBS Codec Settings
| Setting | For Tutorials/Presentations | For Gameplay |
|---|---|---|
| Encoder | Hardware (NVENC, QSV, or AMF) | Hardware (NVENC preferred) |
| Rate Control | CQP (Constant QP) | CQP |
| CQ Level | 20-22 | 18-20 |
| Preset | Quality | Performance or Balanced |
| Resolution | 1920x1080 (native monitor) | 1920x1080 |
| FPS | 30 | 60 |
| Audio | AAC 160kbps | AAC 192kbps |
Why hardware encoding for recording: Software encoding (x264) uses CPU, which competes with whatever you're recording. A software encode at medium preset can consume 30-50% CPU. Hardware encoding (NVENC, Quick Sync, AMF) uses a dedicated chip and typically consumes less than 5% CPU. The files are 20-30% larger than software encoding at the same quality, but you can re-encode later if size matters.
macOS QuickTime: Simple but Limited
macOS's built-in screen recording (Command+Shift+5 or QuickTime Player → New Screen Recording) outputs MOV files with H.264 video and AAC audio.
Advantages:
- Zero setup — built into every Mac
- Low CPU impact — uses Apple's VideoToolbox hardware encoder
- Reliable — Apple's own tool for their own OS
Limitations:
- Single audio track — can record microphone OR system audio (with a virtual audio driver like BlackHole), but managing multiple tracks requires third-party tools
- Fixed quality settings — no CRF or bitrate control. QuickTime decides the quality based on content complexity.
- Large files — QuickTime tends to encode at high bitrate (15-25 Mbps for 1080p), producing 100-200MB per minute at 1080p. A 30-minute tutorial is 3-6GB.
- MOV only — need to convert to MP4 for sharing. Convert MOV to MP4.
QuickTime screen recording is ideal for quick, short captures (under 5 minutes). For longer recordings, use OBS for better control over file size and audio tracks.
Windows Game Bar: Decent Default
Windows 10/11 includes Xbox Game Bar (Win+G) with built-in screen recording. Despite the name, it records any application, not just games.
Output: MP4 with H.264 (hardware encoded via the GPU's encoder) and AAC audio.
Settings (Xbox Game Bar → Settings → Captures):
- Video quality: Standard (30fps) or High (60fps)
- Resolution: Up to 1080p (regardless of monitor resolution — 4K monitors still record at 1080p by default)
- Audio: Records system audio and optionally microphone
Limitations:
- No desktop recording — only captures application windows. Can't record the desktop, File Explorer, or the Start menu.
- No multi-monitor support — records one window at a time.
- MP4 only — not crash-safe. If the recording stops unexpectedly, the file is corrupted.
- Limited quality control — two quality presets with no fine-tuning.
Game Bar is fine for quick application recordings (showing a bug, recording a Zoom call). For anything serious, use OBS.
Lossless vs Compressed Recording
Some recording tools offer lossless or near-lossless options:
- Uncompressed (RGB, YUV): Raw pixel data. Perfectly sharp but enormous: 1080p at 30fps generates ~5.6GB per minute. Only useful for professional capture cards and short clips where quality is paramount.
- Lossless codecs (FFV1, HuffYUV, Lagarith, UT Video): Lossless compression reducing size by 2-3x versus uncompressed. Still massive: ~2GB per minute at 1080p 30fps. Used in professional capture workflows where you need lossless but can't afford uncompressed disk throughput.
- Near-lossless (CRF 0 H.264, ProRes 422): Technically lossy but visually indistinguishable from source. CRF 0 H.264 produces about 500MB-1GB per minute at 1080p. ProRes produces about 150MB per minute. Good balance for editing workflows.
- Standard compressed (CRF 18-22 H.264): Visually lossless for screen content. Produces 30-80MB per minute at 1080p. This is the right choice for 95% of screen recording use cases.
Screen content compresses exceptionally well. Most of the frame is static (toolbars, background, text), and only small regions change (cursor, active window). H.264 CRF 20 on screen content produces files 5-10x smaller than camera video at the same CRF because there's so much less motion and so much more redundancy.
Post-Recording: Optimize Before Sharing
The file that comes out of your recording tool is optimized for reliability during capture, not for sharing. Post-processing steps:
1. Remux MKV to MP4 (if you recorded to MKV in OBS). This is instant and lossless. Convert MKV to MP4.
2. Trim dead time. Most recordings have dead time at the start and end (opening the recording tool, arranging windows). Trim it before sharing.
3. Re-encode if file size matters. Hardware-encoded recordings are 20-30% larger than necessary. If you're uploading a 2-hour tutorial to YouTube, re-encoding with software H.264 at CRF 22 will save significant bandwidth without visible quality loss.
4. Add faststart for web. If sharing the MP4 via a download link or embedding on a web page, ensure the moov atom is at the front of the file (-movflags +faststart) so playback can start before the entire file downloads.
5. Extract audio if needed. Tutorial recordings often need just the audio for a podcast version. MP4 to MP3 or MKV to MP3.
Screen recording format choice is about reliability first, optimization second. Record to MKV with hardware encoding so your computer stays responsive and the file is recoverable if something goes wrong. After recording, spend 30 seconds remuxing to MP4, trim the dead time, and you have a file ready to share or upload anywhere.
The temptation is to record directly to MP4 with aggressive compression to save space. Don't. The risk of losing a 2-hour recording to a crash isn't worth the disk space savings. Record generously, optimize afterward.