Poor Video Quality After Compression? CRF Parameter Tuning in Practice

Bottom line first: Video compression causes poor quality because of three issues: CRF value too high, wrong encoder selection, and excessive resolution reduction. The solution is CRF parameter tuning: set CRF to 18–23 (18 is visually lossless, 23 is the balance point), prioritize the H.265 encoder (saves 40%–50% file size at the same quality as H.264), and don't reduce resolution by more than half. A 2.3GB 4K video can be compressed to 280MB using x265 CRF 23 with virtually no quality loss. Below, we start by examining the causes of quality degradation, then provide detailed CRF parameters, encoder comparisons, and a complete case study.

If you're not yet familiar with the overall concept of video compression, we recommend reading the Video Compression Guide.

1. Why Does Video Quality Degrade After Compression?

Video compression reduces file size by discarding temporal redundancy (similar parts between adjacent frames) and spatial redundancy (similar pixels within the same frame). Quality degradation almost always comes from improperly set compression parameters that cause the encoder to discard too much visually sensitive information. Understanding the causes of quality degradation is key to targeted tuning.

Cause of Quality Loss	Typical Symptom	Root Cause	Frequency
CRF value too high	Block artifacts, blurriness, detail loss	CRF>28, discarding too much high-frequency info	★★★★★
Wrong encoder selection	Significantly worse quality at same size	Using an old encoder or preset too slow/fast	★★★★☆
Excessive resolution reduction	Overall softness, visible aliasing	4K directly reduced to 720p, too many pixels lost	★★★☆☆
Improper bitrate mode	Blurry in motion scenes, wasted in static scenes	Using fixed bitrate CBR instead of variable bitrate VBR	★★☆☆☆
Audio bitrate too high	File size consumed by audio	Using lossless FLAC or 320kbps MP3	★★☆☆☆

As the table shows, CRF value too high and wrong encoder selection are the two main causes, accounting for over 75% of quality issues. A 2.3GB 4K video compressed with CRF 35 becomes 180MB but has visible block artifacts and blurriness; with CRF 23 it becomes 280MB with virtually no quality loss — a CRF difference of 12 produces dramatically different results.

2. CRF Parameters and Encoders in Detail

For video compression quality issues, the core is CRF parameter tuning and encoder selection. The table below first compares three mainstream encoders, then details CRF parameters.

Encoder	Encoding Standard	Size at Same Quality	Encoding Speed	Compatibility	Recommended Scenario
x264	H.264/AVC	Baseline (100%)	Fastest	★★★★★	General/mobile
x265	H.265/HEVC	50%–60%	3–5x slower	★★★★☆	Storage/streaming
AV1	AV1	35%–45%	10x+ slower	★★★☆☆	Cutting-edge/8K

As the table shows, x265 strikes the best balance between file size and compatibility — saving 40%–50% compared to x264, with compatibility covering the vast majority of modern devices. AV1 has the smallest file size but extremely slow encoding and poorer compatibility, making it suitable for cutting-edge scenarios. SmartSlim defaults to x265, balancing size and compatibility.

1. CRF Parameters in Detail (Most Critical)

CRF (Constant Rate Factor) is the quality control parameter for x264/x265, ranging from 0–51. Lower CRF values mean better quality and larger file size; higher values mean worse quality and smaller size. The key is finding the balance between "file size" and "quality." Through extensive testing, CRF 18–28 is the practical range: CRF 18 is visually lossless, CRF 23 is the balance point, and CRF 28 is acceptable lossy.

An important rule: for every 6-point increase in CRF, the bitrate approximately halves and the file size approximately halves. CRF 18 to CRF 24 roughly halves the size, and CRF 24 to CRF 30 halves it again. This means going from CRF 18 to CRF 30 reduces the size to about 1/4 of the original. H.265 can use a CRF 2–3 points higher than H.264 at the same quality, meaning H.264 CRF 21 quality approximately equals H.265 CRF 23–24 quality.

CRF Value	2.3GB 4K Video After Compression	Compression Rate	Visible Distortion	Recommended Use
CRF 18	920MB	60%	None (visually lossless)	Archiving/master
CRF 20	560MB	75.7%	Almost none	High-quality display
CRF 23	280MB	87.8%	Not noticeable	General balance point
CRF 26	160MB	93%	Slight	Mobile/web
CRF 28	110MB	95.2%	Acceptable lossy	Low-bandwidth scenarios
CRF 32	68MB	97%	Visible block artifacts	Not recommended

2. Preset Selection

The encoder's preset determines the trade-off between encoding speed and compression efficiency. Slower presets mean the encoder spends more time finding the optimal compression method, resulting in smaller file size at the same quality. We recommend preset medium (default) or slow — medium is the balance point between speed and efficiency, while slow can save an additional 5%–10% but doubles encoding time. preset ultrafast is the fastest but produces files 30%–40% larger than medium at the same quality, and is not recommended.

SmartSlim defaults to preset medium, with the enterprise version configurable to slow for maximum size optimization. For detailed FFmpeg command-line usage, see FFmpeg Video Compression in Practice.

3. Resolution and Bitrate Strategy

Excessive resolution reduction is another major cause of poor quality. Reducing 4K directly to 720p drops pixels from 3840×2160 to 1280×720, losing 75% of pixels and inevitably causing softness. The correct approach is to not reduce resolution by more than half — 4K can be reduced to 1080p (retaining 50% of pixels), and 1080p can be reduced to 720p, but don't jump directly from 4K to 720p.

For bitrate mode, we recommend VBR (Variable Bitrate) over CBR (Constant Bitrate). VBR lets the encoder allocate more bitrate to complex motion scenes and less to static scenes, resulting in more balanced overall quality. CBR allocates the same bitrate regardless of scene complexity, causing blurriness in motion scenes and wasted bitrate in static scenes. For an in-depth comparison of encoding standards, see H.264 vs H.265 vs AV1 Encoder Comparison.

3. Case Study: 2.3GB 4K Video Compressed to 280MB

This is a corporate promotional video in .mp4 format, originally 2.3GB. 4K resolution (3840×2160), 30fps, H.264 encoded, 45Mbps bitrate, 5 minutes 42 seconds duration. It needs to be uploaded to the corporate website (500MB limit) and WeChat Video Channel (200MB limit), with a target of under 300MB and virtually no quality loss.

Document profile: 4K 30fps, H.264 CRF 18 original encoding, 45Mbps bitrate, AAC audio 256kbps, contains extensive motion shots and close-ups.

Parameters and size changes:

Step	Operation	Key Parameters	Size Change
1	Encoder switch	H.264→H.265 (x265)	2.3→1.2GB
2	CRF tuning	CRF 18→CRF 23, preset medium	1.2→320MB
3	Resolution adjustment	4K→1080p (retain 50% pixels)	320→290MB
4	Audio compression	AAC 256kbps→AAC 128kbps	290→280MB

Result: 280MB, compression ratio ~8.2:1, 87.8% size reduction. Below the website's 500MB limit, close to the Video Channel's 200MB limit (further reducing CRF to 26 can achieve 160MB). 1080p display quality on mainstream monitors is not noticeably different from 4K, motion shots have no block artifacts, close-up details are well preserved, and audio sounds identical. The entire process completed locally — the promotional video never passed through any third-party server, meeting the confidentiality requirements for corporate promotional materials.

4. Compression Recommendations for Different Video Types

Different types of videos have different optimal compression parameters. The table below provides recommendations for common video types.

Video Type	Typical Size	Recommended Encoder	Recommended CRF	Recommended Strategy
Corporate promo	1–5GB	x265	CRF 23	4K to 1080p, preserve detail
Screen recording tutorial	500MB–2GB	x264	CRF 20	Keep 1080p, high efficiency for static scenes
Meeting recording	1–3GB	x265	CRF 26	1080p to 720p, lower quality requirements
Product demo	200MB–1GB	x264	CRF 21	Keep 1080p, ensure clarity
Event recording	2–10GB	x265	CRF 24	4K to 1080p, many motion scenes
Surveillance footage	5–50GB	x265	CRF 28	Reduce to 720p, acceptable lossy

A general principle: for high-quality requirements, use CRF 18–21; for general scenarios, use CRF 23; for low-quality requirements, use CRF 26–28. Prioritize x265 (saves 40%–50% file size); only use x264 when compatibility is critical. Don't reduce resolution by more than half, and AAC 128kbps is sufficient for audio. With the right parameters, video compression can perfectly balance file size and quality.

5. FAQ

Q1: How can I fix poor video quality after compression?

Poor quality is usually caused by CRF value too high, wrong encoder selection, or excessive resolution reduction. Tuning method: set CRF to 18–23 (18 visually lossless, 23 balanced, 28 acceptable lossy), prioritize H.265 encoder (saves 40%–50% file size at same quality as H.264), and don't reduce resolution by more than half (4K can go to 1080p, not 720p). A 2.3GB 4K video can be compressed to 280MB with x265 CRF 23 and virtually no quality loss. SmartSlim automatically selects the optimal CRF and encoder based on video content.

Q2: What's the best CRF value to use?

CRF (Constant Rate Factor) recommendation is 18–28. CRF 18 is visually lossless (virtually identical to the original), suitable for archiving; CRF 23 is the balance point (good quality with significant size reduction), suitable for general use; CRF 28 is acceptable lossy (some detail loss but doesn't affect viewing), suitable for mobile. For every 6-point increase in CRF, the bitrate approximately halves and the file size approximately halves. H.264 recommends CRF 18–23, H.265 recommends CRF 20–26 (CRF can be 2–3 higher at the same quality).

Q3: Which encoder is better — x264, x265, or AV1?

At the same quality, file size ranking: AV1 < x265 < x264. x264 (H.264) has the best compatibility and fastest encoding, suitable for general scenarios; x265 (H.265) saves 40%–50% compared to x264 but encodes 3–5x slower, suitable for storage and streaming; AV1 saves an additional 20%–30% compared to x265 but encodes extremely slowly (10x or more) and has poor decoding compatibility on older devices. We recommend x264 for general use, x265 for size optimization, and AV1 for cutting-edge scenarios. SmartSlim defaults to x265, balancing size and compatibility.

Q4: How can I compress a 4K video to under 300MB?

There are two approaches to compress 4K video to under 300MB: first, maintain 4K resolution with x265 CRF 26–28, with quality loss but retaining 4K detail; second, reduce to 1080p with x265 CRF 20–22, with better quality but half the resolution. We recommend the second approach — 1080p display quality on most screens is not noticeably different from 4K, but quality is clearly better than 4K with high CRF. A 2.3GB 4K video reduced to 1080p x265 CRF 22 is about 280MB, with better quality than 4K CRF 28. Use AAC 128kbps for audio to further save space.

Summary

Video compression causes poor quality because of CRF value too high, wrong encoder selection, and excessive resolution reduction. The solution is CRF parameter tuning: choosing the right CRF value is most critical (18 visually lossless, 23 balance point, 28 acceptable lossy, every 6-point increase halves the size), prioritize x265 encoder (saves 40%–50% compared to x264), don't reduce resolution by more than half (4K to 1080p, don't jump to 720p), and use AAC 128kbps for audio. With the right parameters, a 2.3GB 4K video can be reliably compressed to 280MB with virtually no quality loss.

Remember three points: first, CRF 23 is the general golden balance point — below 18 the file is too large, above 28 the quality is noticeably poor. Second, x265 is the current optimal choice, saving nearly half the size compared to x264 with compatibility now covering mainstream devices. Third, reducing resolution by half is the bottom line — reducing too much inevitably causes softness. With the right encoder and parameters, video compression can perfectly balance file size and quality.