Best Export Settings for YouTube Without Blockiness

You made a good video. You uploaded it to YouTube and, an hour later, the dark scenes look like a mosaic and the grass looks like green felt. The file on your computer looks fine. What happened?

What happened is that YouTube did not play your file. It made a copy of it, compressed to its own settings, and that copy is what people watch. The best thing you can do is hand YouTube a file that is so clean and so generous that its encoder has plenty to work with. This article covers the settings I would use, why each one matters, and the usual mistakes.

The short version

If you only want the settings, here they are.

• Container and codec: MP4 with H.264 (AVC). H.265 if your editor does it well and you are comfortable with it.
• Resolution: the same as your timeline. For 1080p, 1920 by 1080. For vertical shorts, 1080 by 1920.
• Frame rate: the same as your footage. Do not convert 24 fps to 30 or 60.
• Bitrate for 1080p at 24 to 30 fps: 12 to 16 Mbps VBR, with a maximum around 20 Mbps. YouTube’s own recommendation is 8 Mbps, and giving more is not wasted.
• Bitrate for 1080p at 48 to 60 fps: 18 to 24 Mbps.
• Bitrate for 4K at 24 to 30 fps: 40 to 55 Mbps. At 60 fps, 60 to 80 Mbps.
• Audio: AAC, 48 kHz, 320 kbps stereo.
• Other: progressive scan, square pixels, color space Rec. 709 for SDR, and a “fast start” (moov atom at the front) if the option exists.

The rest of the article explains the reasoning so you can adapt it to your footage.

Why “YouTube recommends 8 Mbps” is not the whole story

YouTube’s help pages list recommended upload bitrates, and for 1080p at standard frame rates the number is about 8 Mbps. People take that as the target. But the recommendation is the minimum reasonable bitrate for an upload, not the level at which the footage will look best after YouTube recompresses it.

Think about what happens. Your 8 Mbps file is already compressed. YouTube decodes it and encodes it again, usually at a lower bitrate, and each generation of compression loses detail. If the upload is only just clean enough, the re-encode will make it slightly worse everywhere. If the upload is at 16 Mbps, YouTube’s encoder starts from a better picture, and the final result is closer to what you saw in the editor.

I tested bitrates on a clip from an open-source animated film that has fur, grass and fast camera moves. At 4 Mbps it was watchable but visibly soft in motion. At 8 Mbps it was clean. At 16 Mbps I could not tell it from 8 Mbps when I looked at it at full screen on its own. But a second compression pass amplifies small differences, so the extra headroom has value when the file goes through a platform. You can see the full numbers in what bitrate you need for 1080p.

Resolution: match the timeline, and consider going up

Export at your timeline’s resolution. Scaling in the export dialog is possible but adds an extra resampling step, and sometimes picks a poor filter.

The question that comes up is whether to upload in a higher resolution than you filmed. YouTube’s different resolution tiers get different encoding treatment. Creators have noticed for years that a 4K upload tends to hold up better than a 1080p one, even if the 4K is an upscale. The effect is real but it has a cost: the upload is larger, processing takes longer, and you are not adding detail. I discuss it in upscaling to 4K, the honest guide. If your footage is already 4K, upload at 4K.

Frame rate: keep it native

If you shot at 24 fps, export at 24. If you shot at 30, export at 30. Converting between them duplicates or drops frames and introduces judder. If you shot at 60 fps and want the smooth look, upload at 60.

Variable frame rate is another trap. Phone footage is often variable frame rate, and some editors misinterpret it. If audio drifts out of sync with the picture, the culprit is often variable frame rate. Convert the clip to constant frame rate before editing, or let the editor handle it on import.

Codec choices: H.264, H.265, ProRes

H.264 is the safe default. Every editor can export it and every platform accepts it. A well-configured H.264 export at a good bitrate is more than enough for YouTube.

H.265 (HEVC) is more efficient. For the same quality, it can use about 30 to 40 percent fewer bits. For uploading, the benefit is a smaller file. The risk is that some editors’ H.265 encoders, particularly hardware ones, are less efficient than the codec’s potential, so the savings do not always show up. I have a test of both in H.264 versus H.265.

ProRes or DNxHR are editing codecs. They are huge but nearly lossless. YouTube accepts them, and they give the cleanest possible start for the re-encode. A ten minute 1080p ProRes 422 file is on the order of 10 GB. If your connection and storage can handle it, it is the best-quality upload. For most people it is overkill.

Variable or constant bitrate

For a platform upload, use variable bitrate (VBR). A static scene needs few bits and a chaotic one needs many, and VBR spends them where they are needed. Set the target around the numbers above and the maximum at 1.3 to 1.5 times the target. If your editor offers two-pass VBR, use it for final exports: the first pass measures how hard each scene is and the second distributes bits accordingly.

Constant bitrate (CBR) wastes bits in easy scenes and starves hard ones. It makes sense for live streaming, not for uploads.

If your tool offers constant quality instead, as x264 does with CRF, use CRF 16 to 18 for YouTube uploads. The file size will vary with the content, but the quality will be consistent. CRF is explained here.

Audio

Do not skimp. YouTube re-encodes audio too, so give it a clean source. AAC at 320 kbps, 48 kHz is a standard choice. Keep loudness in a sensible range: YouTube normalizes playback around minus 14 LUFS, and a very loud mix just gets turned down.

Things that cause blockiness no matter what

Some footage is simply harder to encode. If your video contains any of these, expect more compression damage and raise the bitrate.

• Film grain and noise. The encoder treats it as detail.
• Confetti, snow, rain, dust and sparkles. Tiny, fast-moving, random detail.
• Foliage and water. Detailed and moving in unpredictable ways.
• Fast pans and shaky handheld footage. Everything moves at once.
• Fades and dissolves. Gradients over a moving image.
• Dark scenes with gradients. These produce banding.

There is a separate piece on why some footage breaks encoders: why busy scenes ruin video quality. A practical fix for noisy footage is a light denoise before export, which frees bits for real detail. If the video still looks blocky after YouTube processes it, see YouTube video pixelated after upload.

Settings for specific editors

Premiere Pro. Format H.264. Preset “Match Source - High bitrate” is a reasonable start. Set the bitrate encoding to VBR 2 pass, target 16 Mbps, maximum 24 Mbps. Turn on Use Maximum Render Quality. Leave “Render at Maximum Depth” off unless you know you need it.

DaVinci Resolve. On the Deliver page choose the YouTube preset as a starting point, then switch Quality to “Restrict to” and set 16000 Kb/s for 1080p. Choose the encoder, software if your GPU’s encoder looks poor. Set the Encoding Profile to High.

Final Cut Pro. Use the Share, YouTube and Facebook option or a custom export, and pick “Better Quality” rather than “Faster Encode”. For the most control, export a ProRes master and compress it with another tool.

CapCut. Choose the maximum resolution, the highest frame rate you filmed at, and “Higher” bitrate. Use H.264 unless you know your target supports HEVC.

OBS recordings. If you record gameplay with OBS, record at a high bitrate, such as 30 to 40 Mbps for 1080p60, and avoid re-encoding it more than once. See OBS recordings pixelated when moving.

Upload checks after publishing

When your video is processed, check it properly.

1. Wait until the HD or 4K version is available. YouTube shows lower resolutions first.
2. Set the player quality manually to the highest option. Do not rely on Auto.
3. Watch the part of the video most likely to show problems: the darkest scene, the busiest scene, and any title with fine text.
4. Compare against your local file at the same size.

If it looks fine in those four places, it is fine. If it does not, check your bitrate and the footage. It is usually one of the two.

Check your export before you upload it

It takes thirty seconds to confirm that the file you are about to upload is what you think it is. Open it in a player that shows media information, or run ffprobe -hide_banner yourfile.mp4. Look at four things.

The video stream line should show H.264 (or HEVC), the resolution you expected, and a frame rate that matches your footage. If it says 29.97 and you filmed at 30, that is normal. If it says 24 and you filmed at 60, something in your sequence settings was wrong.

The bitrate should be in the range you set. Editors sometimes ignore your number and use their own, especially when a preset overrides it. If you asked for 16 Mbps and the file says 6 Mbps, the preset won.

The pixel format should be yuv420p for H.264 uploads. Other formats such as yuv444p are valid but can cause trouble in some players and add nothing for YouTube.

The audio line should show AAC at 48000 Hz. Some editors default to 44.1 kHz, which is fine, but mixing sample rates in a project can cause drift.

Then watch the first thirty seconds and the last thirty seconds of the file. It sounds silly, but a surprising number of exports have a glitch at the end, such as a missing frame or a cut-off audio tail, and finding it before upload saves an hour.

Shorts, vertical video and screen recordings

Vertical video for Shorts should be 1080 by 1920 at 30 or 60 fps. Many people export these from phone apps at 720 by 1280, and the picture suffers when a 1080p phone screen shows it enlarged. Export at 1080 by 1920, with a bitrate of about 10 to 15 Mbps, and keep the important text away from the edges because the interface covers part of the screen.

Screen recordings behave differently from camera footage. They have large flat areas, razor sharp text, and sudden changes when a window opens. The sharp text suffers when chroma subsampling reduces color detail, so red text on a dark background can look fuzzy. Higher bitrates help, but a more reliable fix is to record at a size close to the final output so that no scaling happens, and to avoid scaling the recording in the editor by non-integer amounts. If you must scale, 200 percent of a 960 by 540 recording looks cleaner than 187 percent of a 1024 by 576 one.

HDR and color: a short warning

If your phone records HDR, a clip can look washed out or too bright when exported as SDR without tone mapping. This is not pixelation, but it shows up in the same complaints: “my video looks bad on YouTube”. If you shot in HDR, either export HDR properly with the right color tags, or convert the footage to SDR in the editor using a color management setting before you export. Do not mix HDR and SDR clips in one timeline without handling the conversion.

Mistakes I see repeated

Exporting the same clip several times as a test. Each export of an export loses quality. Always go back to the project, not the previous file.

Using the “web optimized” preset for the final. Presets that aim for small files are the number one reason for blocky uploads. Look at the bitrate before you click export.

Uploading with a restricted mobile connection. Interrupted uploads sometimes get re-encoded by the app. Use the desktop site for the final upload when you can.

Judging quality on the thumbnail player. Maximize the player to full screen, set the quality manually, and watch a few seconds of the busiest scene.

Trusting the first hour. YouTube may show you a blocky 360p version while the rest processes. If that is all you see, wait.

If you run through this list and the picture is still bad, the source is probably the limit, and that is a footage problem more than an export problem. Cleaning the source, even lightly, usually helps more than another export at a higher bitrate.

When the video is already exported and the quality is poor

Sometimes the project is gone, and all you have is a low quality export. In that case, a cleanup pass can help: deblock, denoise, scale to a higher resolution with a good filter, then re-encode at a generous bitrate before uploading. It will not restore the detail, but it makes the video easier on YouTube’s encoder, and the output looks calmer. You can try it on a short sample with the free preview on the home page.

A repeatable routine

When quality matters, I use the same routine for every upload, and it takes about as long as the export itself. First I export a ten second test of the hardest part of the video with my intended settings and look at it at full screen. If it is clean, I export the whole thing with the same settings. Then I run the quick ffprobe check above. Then I upload and, once YouTube has processed the HD version, I watch the hardest part again in the player and compare it with the local file.

If the hosted version is noticeably worse than the local one in that section, I raise the bitrate by about a third and try again. If it is still poor, I look at the footage itself: grain, noise and shake are the usual suspects, and a light denoise before export is the cheapest fix. A routine like this turns a mysterious problem into a short checklist, and it means that next time you will know which setting to touch before the video even leaves your computer.

Keep a note of the settings that worked for you, with the editor and version. Presets drift between software updates, and a line in a text file will save you from rediscovering your own best settings every few months.

Summary

Treat YouTube’s encode as a second compression and give it the cleanest, most generous file you can. For 1080p, 12 to 16 Mbps in H.264 is a sensible setting for most footage. Use the native resolution and frame rate, VBR with a sensible maximum, and a high-quality audio track. Add a light denoise for grainy footage. Then check the result after YouTube has finished processing, not before.