The safest way to compress 4K video is to decide first whether the destination actually needs 4K. If it does, keep 3840×2160 and give the encoder a realistic bitrate. If it does not, scale to 1080p before spending bits on a resolution the viewer will never see.
The mistake is keeping the 4K label while forcing the file into a tiny budget. A file can still report 3840×2160 after most of its fine detail has been replaced by blur, blocks and ringing. Dimensions are not the same thing as usable detail.
Quick takeaways
- Keep 4K for large displays, reframing, archiving or a platform that can use it. For email, chat, most web embeds and phone playback, 1080p is usually the better use of the available bytes.
- Bitrate is the real budget. One minute in 20 MB leaves only about 2.54 Mbps for video after 128 kbps audio; changing the codec cannot make that a healthy 4K budget.
- High-motion footage needs more. Water, grain, foliage, confetti and camera movement are harder than slides or a static product demo.
- Use target-size mode for a hard cap, but leave headroom. In our 4K tests, requested 10 MB outputs landed between 10.57 and 11.02 MB.
- H.265 can help camera footage, but H.264 is safer when the file must play on an unknown device. Codec choice comes after the bitrate decision.
- Check difficult frames at full size. A thumbnail can hide damage that is obvious on a 4K monitor.
Why 4K video becomes enormous
A 4K UHD frame contains 3840×2160 pixels—four times as many pixels as 1080p. The codec does not store every pixel independently, but it still has four times the frame area to describe.
Then duration, frame rate and scene complexity multiply the demand:
- twice the duration needs roughly twice the total bytes at the same bitrate;
- 60 fps gives the encoder twice as many frames as 30 fps;
- natural noise and fast motion reduce how much one frame can borrow from the previous one;
- HDR and 10-bit sources carry more tonal information and require a workflow that preserves or intentionally maps it.
This is why two one-minute 4K videos can have radically different sizes. A static screen capture and handheld footage of breaking waves are both “4K,” but the second contains far more new information from frame to frame.
Start with the destination, not the source
Ask one question before touching a setting: where will the result be watched or used?
| Destination | Sensible starting decision |
|---|---|
| TV, projector, 4K YouTube upload | Keep 4K if detail matters |
| Editing, cropping or reframing later | Keep the original 4K master |
| Website hero or product page | Usually export 1080p; loading speed matters more |
| Email or chat attachment | Use 1080p or 720p and target the platform cap |
| Phone-only viewing | 1080p is normally enough; 720p can be cleaner at tiny sizes |
| Archive you control | Consider H.265 and keep the original |
Downscaling is not automatically a quality failure. A clean 1080p output with enough bitrate often looks better than a starved 4K output at the same file size.
Use the dedicated 4K video compressor when you want to keep UHD or deliberately scale it for delivery. Use the video compression calculator first when the file must fit a specific number.
The bitrate budget for 4K
For a target file size, the approximate video bitrate is:
video kbps ≈ target MB × 8,000 ÷ duration in seconds − audio kbps
This uses decimal megabytes, which is how most upload limits are advertised. For a one-minute video with 128 kbps audio:
| Total target | Video bitrate left | What it means for 4K |
|---|---|---|
| 20 MB | 2.54 Mbps | Far too constrained for normal 4K detail |
| 50 MB | 6.54 Mbps | Possible only for easy content or visible compromise |
| 100 MB | 13.21 Mbps | Usable for some simple 4K footage; inspect carefully |
| 250 MB | 33.21 Mbps | Near a normal 4K delivery range |
| 300 MB | 39.87 Mbps | A comfortable starting point for many 30 fps uploads |
These are budgets, not universal quality thresholds. A screen recording can survive far below a noisy camera clip.
For context, YouTube's current upload guidance recommends 35–45 Mbps for 4K SDR at 24–30 fps and 53–68 Mbps at 48–60 fps. Those figures are upload recommendations for YouTube's transcoding workflow, not minimums for every 4K file. They are still a useful warning against trying to make a normal minute of 4K camera footage fit into 20 MB.
What happened in our measured 4K target-size jobs
We ran target-size compression on one real 11.7-second 3840×2160 iPhone waves scene through the same provider path used by the product, using two input variants:
- the original HEVC/MOV file at 32.1 Mbps;
- a normalized H.264/CFR MP4 derivative at 72.2 Mbps.
These rows test target-size behavior across input encodings, not two independent scenes.
The encoder kept the 3840×2160 dimensions in every row. We did not capture VMAF for these runs, so this table supports claims about size, bitrate, dimensions and target accuracy—not a claim that every result looked good.
| Source | Requested target | Actual output | Output bitrate | Size removed | Output dimensions |
|---|---|---|---|---|---|
| Original HEVC/MOV | 19 MB | 19.84 MB | 13.55 Mbps | 57.8% | 3840×2160 |
| Original HEVC/MOV | 10 MB | 10.57 MB | 7.22 Mbps | 77.5% | 3840×2160 |
| Original HEVC/MOV | 4 MB | 4.17 MB | 2.85 Mbps | 91.1% | 3840×2160 |
| Normalized H.264/CFR | 30 MB | 30.68 MB | 20.93 Mbps | 71.0% | 3840×2160 |
| Normalized H.264/CFR | 10 MB | 11.02 MB | 7.52 Mbps | 89.6% | 3840×2160 |
Three lessons matter.
Keeping 4K dimensions does not prove the detail survived
The 4 MB original-input result still reports 3840×2160, but only 2.85 Mbps is available for more than eight million pixels per frame. A media inspector will call it 4K. That does not mean a viewer receives 4K detail.
When the budget becomes that low, scaling to 1080p gives each remaining pixel more data and usually produces a cleaner picture.
Difficult footage makes exact targets less exact
Both 10 MB variants overshot the decimal target: the original HEVC/MOV input by 5.7% and the normalized H.264/CFR input by 10.2%. Even for the same scene, container overhead, input encoding, audio and rate-control decisions move the final byte count.
If an upload rejects anything above 20 MB, do not request exactly 20 MB. Use a platform preset with headroom or target roughly 5–10% below the hard ceiling. The harder the source, the more useful that margin becomes.
A larger target can still remove most of the file
The normalized H.264/CFR variant fell from 105.8 MB to 30.68 MB while remaining 4K. You do not have to chase the smallest possible number to achieve a meaningful reduction. Stopping at a realistic target preserves more room for motion and texture.
A practical 4K compression workflow
Step 1: keep the original
Compression is a delivery operation, not a replacement for the master. Keep the camera file until the compressed copy has been checked on its destination. This matters especially for HDR, vertical rotation metadata and footage you may edit again.
Step 2: trim before encoding
Removing unwanted time is the only reduction with no image-quality cost. Cut countdowns, accidental recording, duplicate takes and dead space before asking the codec to work harder.
Step 3: decide whether to keep 4K
Keep it when:
- the destination displays 4K;
- the footage will be cropped or stabilised later;
- small text or distant detail matters;
- you are creating an archive rather than a delivery copy.
Scale to 1080p when:
- the file is going into chat, email or a form;
- the viewer will watch in a small embedded player;
- the target bitrate is below what 4K can use;
- upload and download time matter more than future reframing.
Step 4: choose size mode or quality mode
Use target-size mode when a platform has a hard cap. The tool works backward from the number and assigns a bitrate budget.
Use quality mode when the result can land at any size and visual quality is the priority. Start moderately, inspect the hardest scene and adjust once. The CRF guide explains why one setting cannot fit screen, animation and camera footage equally.
Step 5: choose H.264 or H.265
- H.264: faster and safer for broad sharing.
- H.265: often smaller for camera footage when you control playback.
In our measured codec sweep, H.265 saved 25–35% on the camera clip but did not save bytes on the screen recording. The full H.264 vs H.265 comparison includes the 36-run table and compatibility trade-offs.
Step 6: preview the difficult scene
Do not judge a 4K result from the first talking-head frame. Check:
- water, smoke, rain, snow or confetti;
- leaves, grass, hair and fabric texture;
- fast pans and handheld movement;
- dark gradients and skies;
- small text and thin lines;
- faces in motion rather than paused faces.
Look at 100% scale on the largest screen that matters. Also listen through the whole file if the source came from a phone, screen recorder or multi-track container.
Settings by destination
4K YouTube upload
Keep the source frame rate, use progressive video and provide a clean file with sufficient bitrate. YouTube recommends MP4/H.264 and publishes 35–45 Mbps for 4K SDR at 24–30 fps, with higher ranges for 48–60 fps and HDR. It will transcode the upload again, so avoid repeatedly compressing the same intermediate file.
Website video
Most website players do not need a 4K master. Export 1080p, keep the duration short, remove audio if it is genuinely unused, and measure page performance. A smaller resolution with enough bitrate loads faster and often looks better than low-bitrate 4K.
Discord, email and chat
The cap, not the source resolution, is the binding constraint. Use a target-size preset for the hard ceiling. That mode keeps the source dimensions, so if a 4K result looks starved, return to the original and choose 1080p or 720p resolution mode instead. For a free Discord upload, the Discord compressor targets below the current 20 MB cap rather than spending every last byte.
Archive or personal library
Keep the original master if storage allows. For a smaller viewing copy, H.265 is worth testing on camera footage because playback is under your control. Do not delete the original based only on file size.
Common mistakes
Keeping 4K at any cost
Resolution is one property, not a quality score. If the available bitrate is too low, 1080p wins.
Targeting the exact upload limit
Our hard 4K jobs overshot by up to 10.2%. Leave headroom.
Compressing the compressed output again
Every lossy generation discards more information. Return to the original and make one stronger encode.
Converting HDR without checking tone mapping
A standard H.264 delivery copy may map HDR into SDR. Check highlights, skin, coloured lights and shadow detail on the destination screen.
Assuming H.265 fixes an impossible budget
A more efficient codec cannot make one minute of normal 4K camera footage look healthy at 2.5 Mbps. Reduce resolution, shorten the clip or allow a larger file.
FAQ
Can I compress 4K video without losing quality?
Not literally with normal lossy compression. You can often remove a large amount without a visible difference when the source has excess bitrate. The safe amount depends on motion, noise, duration, frame rate and where the result will be viewed.
Should I compress 4K to 1080p?
Yes when the destination does not display 4K or the target bitrate is too low. A well-funded 1080p encode usually looks cleaner than starved 4K at the same file size.
What bitrate should I use for 4K video?
There is no universal number. YouTube recommends 35–45 Mbps for 4K SDR uploads at 24–30 fps and 53–68 Mbps at 48–60 fps. A private archive, website clip or screen recording can have different needs. Start from the destination and test the hardest scene.
Is H.265 better for 4K than H.264?
It is often more space-efficient for camera footage, but it is slower and less universally playable. Use H.265 when you control playback; use H.264 for an unknown recipient or an upload workflow that asks for it.
Why is my 4K video still blurry after compression?
The bitrate may be too low for the resolution, especially on motion, water, grass, grain or 60 fps footage. Increase the target size or scale to 1080p. Keeping the 3840×2160 label does not restore discarded detail.
How small can a one-minute 4K video be?
Mathematically it can be any size. Practically, one minute in 100 MB leaves about 13.2 Mbps for video after 128 kbps audio, while 20 MB leaves only 2.54 Mbps. The second number is too constrained for normal 4K detail. Content complexity decides where the visible failure begins.
Why did a 10 MB target create an 11 MB file?
Target size is not byte-exact. Audio, container overhead, motion complexity and encoder rate control can move the result. Our measured 4K 10 MB jobs landed at 10.57 MB and 11.02 MB, which is why hard upload limits need headroom.
Can I compress a 4K iPhone video?
Yes. Keep the original HEVC/MOV master, create a delivery copy, preserve rotation, and inspect HDR-to-SDR handling if applicable. Use 1080p for ordinary sharing or keep 4K with a realistic bitrate for large-screen playback.
Sources and limitations
- First-party target-size jobs:
scripts/fc-target-size-results.jsonl, measured September 14, 2026. The table reports bytes, bitrate and dimensions; these runs did not capture VMAF. - YouTube recommended upload encoding settings, checked November 3, 2026.
- The measured rows come from one high-motion iPhone waves scene in two input encodings. They expose target-size behavior, not source diversity, and are not representative of every camera, phone or scene.
- File-size limits are usually decimal MB; operating systems may display binary MiB, so the same file can appear to have two different numbers.
Compress a 4K video, or compress any video when the destination matters more than preserving UHD dimensions.