4K or 8K video for social media: which should you film and upload?

Should you film or upload 4K or 8K for each social platform?

Our suggestion, based on the documented limits in the next sections and the measurements further down, is below. It is a recommendation, not a platform rule. Where a platform has no official statement about what it does with a larger file, we say so.

Our suggestion; limits are from the official pages under Sources.

The row that decides most of it is the YouTube one. If your channel is the only place the full-resolution file will be watched, 4K is the practical ceiling, and we found no official YouTube statement that 8K improves reach or ranking. For short vertical clips, the recommended or documented sizes in the table are 1080 by 1920 or smaller.

What is the real difference between 4K and 8K?

Only the pixel count. The frame sizes below are the standard ones YouTube lists, and the pixel counts are plain arithmetic: width times height. 8K has four times the pixels of 4K and sixteen times the pixels of 1080p, which is why every step up multiplies storage, decoding and editing work.

16:9; sizes from the YouTube resolution page, pixel counts calculated by us.

YouTube adds a note that since 2022 it has been removing playback support for resolutions between 4K and 8K, so a 5K file may no longer play back at 5K. If you export for YouTube, use the standard sizes in the table.

What does each platform accept?

These are the maximums and recommendations on the official pages, read on October 21, 2026. The Instagram, Facebook, TikTok and X rows come from developer documentation for publishing by API; what the Instagram and TikTok mobile apps do with a larger file in practice is undocumented in what we read and not tested by us, and we did not upload a test file to any of them.

Official pages, read October 21, 2026.

Two reads of this table. First, keep two things apart. What the publishing API (and, for LinkedIn, the help page) accepts is documented in the table. What the Instagram and TikTok mobile apps do with a 4K or 8K file is undocumented in our sources and not tested: it may be accepted and downscaled, so treat 4K and 8K as unproven there. Second, a 25 Mbps cap on Instagram is a small fraction of what an 8K file runs at, as the measurement below shows.

For the general reason platforms re-encode what you send, see our guide to why social media compresses video.

What does 8K cost at the same quality setting?

We made two 2-second, 30 fps test scenes at 7680 by 4320 with ffmpeg: scene A is clean test graphics, scene B is the same graphics with heavy synthetic noise. We scaled each to 4K and 1080p and encoded all three with x264 at CRF 23, a constant quality setting, so the encoder chooses the bitrate it needs. The chart and table show what it chose.

X264 CRF 23, 2 second clips, ffmpeg 6.1.1 on Linux, October 21, 2026.

The spread between the two scenes is the point: how much an 8K file costs depends on the picture, not only on the resolution. Fine detail and noise, which dark or high-ISO footage has, are the expensive part. Our scene B noise is much heavier than a normal camera would produce, so read them as two data points, not a forecast for your camera. The panning scene further down gave a third, 3.3 times.

Multiply a bitrate by 7.5 to get megabytes per minute. At YouTube's recommended 4K bitrate of 35 to 45 Mbps that is roughly 260 to 340 MB a minute; at its 8K recommendation of 80 to 160 Mbps it is roughly 600 to 1,200 MB a minute (arithmetic from the official bitrates).

Does 8K look sharper once it is shown at 1080p?

Not in our test. We chose a 1080p viewing size as our assumption (not measured) and scaled every encode to 1920 by 1080 and compared it with the 1080p master using SSIM, a similarity score where 1.0 is identical. We ran it two ways: the quality setting above, and a fixed 15 Mbps cap so all sizes got the same bits.

Higher is closer to the master; Koldflux measurement, October 21, 2026.

At the same bitrate cap, 8K scored lowest in both scenes: the encoder spreads the same bits over four times the pixels. At CRF 23 the three sizes are within 0.003 SSIM of each other on the clean scene and 0.001 on the noisy one, at 9.4 to 54.6 times the bits on these two scenes. The achieved bitrates differ from 15 Mbps because the cap was applied to a 2 second single-pass encode; that is a limit of this test.

What this does not show: how an 8K file looks on an 8K or large 4K screen, which is where extra pixels could help. SSIM is also a computed score, not a viewing panel. We did not test that.

What happens with motion, HEVC and AV1?

To get closer to real footage than a static test pattern, we made a 6 second panning scene: a procedurally generated, multi-scale detailed image (not camera footage), moved across the frame at 30 fps. We encoded it at 8K, 4K and 1080p with three encoders, using the same one-setting-per-codec approach, and scored each at a 1080p viewing size. We planned 10 to 30 seconds, but raw 8K masters did not fit our machine beyond 6 seconds.

Koldflux measurement, October 22, 2026; CRF values are not equivalent across codecs.

Here 8K did score higher at a 1080p viewing size with x264 (0.9972 against 0.9948) and x265 (0.9961 against 0.9935), a small edge that cost 3.3 and 2.4 times the file size and about 4 times the encode time. This contradicts a flat "8K never helps" reading, so we state it. Two caveats cut the other way: our 1080p reference is scaled from the 8K master, which favors the 8K encode, and the AV1 setting we picked produced much lower scores at every size, so its numbers show the cost of resolution, not a fair codec comparison. We had no VMAF tool in this ffmpeg build, so SSIM is the only score, and we did not judge by eye.

Takeaway for this scene: HEVC gave a smaller file than x264 at every size here, but at a different quality setting, so we do not claim one codec is better. Full commands and output are in the test log.

When is 8K worth filming?

When you will crop, reframe or stabilize, and only if your storage and editing machine can take it. Extra pixels let you zoom into the frame and still export a sharp 4K or 1080p file. This is arithmetic on frame sizes, not a measured result.

Calculated by us.

Read the last column as the useful one for social: cutting a vertical clip out of a wide 4K frame already leaves enough pixels for a 1080 by 1920 export, with a small zoom to spare. 8K buys more zoom, not a better 1080 by 1920 file. Camera files also use heavier codecs and bitrates than our tests, and we did not test any camera, so check what yours does before you count on this.

How do you convert 8K to 4K or 1080p?

Scale it down, with a high-quality filter, and re-encode. The ffmpeg command below turned our 8K test encode into a 3840 by 2160 H.264 file in 7.1 seconds (3.19 MB). To go straight to 1080p, change the scale to 1920:1080.

ffmpeg -i in8k.mp4 -vf scale=3840:2160:flags=lanczos -c:v libx264 -crf 18 -preset slow -c:a copy -movflags +faststart out4k.mp4

No command line? Our free Compress video tool re-encodes in your browser with nothing uploaded, and its largest preset is 1080p. We ran it on the live site in headless Chromium: a 4K and an 8K H.264 test clip both came out as 1920 by 1080 H.264 (11 and 19 seconds). Its page states a 100 MB per file limit, and our 152 MB 8K file was rejected with the error "Processing failed - a file may be corrupt, too large for your browser, or unsupported". Real 8K camera files are usually larger than that, so for those use the ffmpeg command.

The raw commands and outputs for every run on this page are in the test log below.

Free tool:Compress video in your browser (set the 1080p preset)

Can a 4K screen or a phone play 8K?

A 4K screen shows an 8K file by scaling it down, and the picture cannot be sharper than the screen. Whether a given phone, TV or laptop can decode the file at all depends on its hardware and the codec, which we did not test. If an 8K clip stutters on an older device, exporting a 4K or 1080p copy is the practical fix.

Download the test log

Every command we ran, each exit code and time, and all results in one text file, so you can repeat the measurement.

Free tool:test-results.txt (commands, outputs, results)

Limits and how we checked

• Tested by us: the encodes, sizes, times and SSIM scores, the ffmpeg 8K to 4K command and a run of the Koldflux compress tool on the live site in headless Chromium, on October 21, 2026, plus the panning scene with x264, x265 and SVT-AV1 on October 22, 2026, all on one Linux machine (ffmpeg 6.1.1).
• Only read: every platform limit and bitrate figure comes from the official pages under Sources, read on October 21, 2026 (YouTube help in Chromium on a Mac; the Meta, TikTok, LinkedIn and X pages in headless Chromium). The Instagram, Facebook, TikTok and X pages are developer documentation, and the apps may accept or process more. We did not upload any test file to any platform, so what each one actually keeps after upload is not tested.
• Two test scenes are synthetic ffmpeg test graphics (one clean, one with added noise) and only 2 seconds long; the third is a 6 second procedural pan. Real camera footage and camera encoders will give different sizes. The panning scene (6 seconds, procedural, not camera footage) adds motion and the HEVC and AV1 encoders, but only through software x265 and SVT-AV1, not phone or camera hardware encoders. Results change with camera, codec, content and encoder settings.
• The measurements are single runs, not averages, and SSIM at a 1080p viewing size does not measure how 8K looks on an 8K display. We did not test playback on any device.
• We did not cover 5K, 6K or 16K, VMAF scoring, HDR workflows, or camera or phone model comparisons. The recommendations in the first table are our suggestion. Date of checks: October 21, 2026, and October 22, 2026 for the panning scene.