How to compare av1 and h.265 in your browser
- Step 1Pick one representative source clip — Use a 20–60 second clip that represents your real content (motion, grain, detail). Both encoders read it locally via an off-screen
<video>; nothing uploads. - Step 2Decide the comparison axis: size or speed — Matched bitrate (pin the same kbps on both) → compare which looks better at the same size. Matched quality (eyeball-match by adjusting bitrate) → compare resulting file sizes. Don't compare auto-vs-auto, because the two have different per-pixel defaults (0.04 vs 0.06).
- Step 3Encode H.265 first — Open the H.265 encoder, set your Bitrate and the source's real Framerate, encode, note the output size and how long it took. H.265 hardware encode works on almost any modern GPU.
- Step 4Encode AV1 with the identical settings — Open the AV1 encoder, set the same Bitrate and Framerate, encode, note size and time. If your GPU lacks AV1 encode hardware, this step refuses — which is itself a decisive comparison result for your machine.
- Step 5Compare size, speed, and playback — At matched bitrate the files are similar in size — judge quality by eye. At matched quality, AV1 should be ~30% smaller. Also note encode time (both are hardware, usually near real-time) and that both outputs are video-only.
- Step 6Decide based on reach, not just the numbers — If AV1 is smaller/better AND your target audience's devices decode AV1, choose AV1. If you need to reach older devices or your encode hardware can't do AV1, H.265 wins on reach. For maximum compatibility, even H.264 via the transcoder has its place.
AV1 vs H.265 in JAD's encoders — head to head
Both run the same WebCodecs hardware pipeline. The differences that matter are the auto target, hardware-encode breadth, and decode reach.
| Attribute | AV1 encoder | H.265 encoder |
|---|---|---|
| Engine | WebCodecs hardware (av01.0.04M.08) | WebCodecs hardware (hev1.1.6.L120.B0) |
| Auto bitrate target | 0.04 bits/pixel | 0.06 bits/pixel |
| Controls | Bitrate (0–50000) + Framerate (0–120) | Bitrate (0–50000) + Framerate (0–120) |
| Hardware encode reach | RTX 40-series / Arc / Apple M3+ | NVENC / QuickSync / VideoToolbox / AMF (wide) |
| Decode reach | Modern devices, growing | Very wide incl. many older devices |
| Audio in output | No (video-only) | No (video-only) |
| Licensing | Royalty-free | Patent pools / fees |
| Tier required | Pro + Media | Pro + Media |
How to set up a fair test
The single most common comparison mistake is comparing the two auto targets, which differ by design. Pin settings to make it fair.
| Goal | AV1 setting | H.265 setting | Then compare |
|---|---|---|---|
| Quality at the same size | Pin Bitrate 4000 | Pin Bitrate 4000 | Which looks better |
| Size at the same quality | Adjust bitrate to match look | Adjust bitrate to match look | Which file is smaller |
| Encode speed | Same Bitrate + fps | Same Bitrate + fps | Which finished faster |
| Realistic default output | Bitrate 0 (auto 0.04) | Bitrate 0 (auto 0.06) | Out-of-box file sizes (not equal-quality) |
Cookbook
Run these as paired encodes (H.265 then AV1) with identical settings, then compare. Both encoders are video-only and no-upload.
Matched-bitrate quality test
Pin the same bitrate on both encoders and judge which looks better. This isolates the codec's efficiency at a fixed file size.
Clip : 1080p30, 40 s, action footage
H.265 : Bitrate 4000, Framerate 30 → 19.8 MB
AV1 : Bitrate 4000, Framerate 30 → 19.8 MB (same size)
Verdict: inspect both — AV1 usually holds detail better at
the same bitrate, especially in motionMatched-quality size test
Tune bitrates until both look equivalent, then compare sizes. This is the classic '~30% smaller' claim, measured on your clip.
Clip : 1080p30 talking-head H.265 : Bitrate 3000 → looks clean → 14.7 MB AV1 : Bitrate 2100 → looks equivalent → 10.3 MB Result : AV1 ~30% smaller at matched quality
Encode-speed test on your GPU
Both paths are hardware, but AV1 NVENC vs HEVC NVENC can differ. Time each at identical settings.
Clip : 4K30, 30 s H.265 : Bitrate 12000, fps 30 → encoded in ~32 s AV1 : Bitrate 12000, fps 30 → encoded in ~38 s Note : both near real-time on Ada; numbers vary by GPU
Default-vs-default (NOT a fair quality test)
Comparing the two auto modes shows out-of-box file sizes, but it's not equal quality because the per-pixel targets differ.
AV1 auto (0.04 b/px) 1080p ≈ 2.5 Mbps → smaller file
H.265 auto (0.06 b/px) 1080p ≈ 3.7 Mbps → larger file
Caveat : the AV1 file is smaller partly because its auto
target is lower — pin equal bitrate for fairnessHardware-support reality check
The most decisive comparison is sometimes whether AV1 encodes at all on your machine. If AV1 refuses, H.265 wins by default for you.
GPU : RTX 3070 (Ampere) H.265 : encodes fine (HEVC NVENC) AV1 : 'Hardware encoding for AV1 is not available' Verdict: on this machine, H.265 is the only HW option
Edge cases and what actually happens
Comparing the two auto modes
Unfair testAV1 auto is 0.04 bits/pixel and H.265 auto is 0.06 — different targets by design. Comparing auto-vs-auto shows out-of-box sizes, not equal-quality efficiency. To compare codecs fairly, pin the same bitrate on both (then judge quality) or match quality by eye (then compare size).
AV1 refuses on your hardware
Not availableIf your GPU can't hardware-encode AV1 (anything below RTX 40-series / Arc / Apple M3), the AV1 side throws 'Hardware encoding for AV1 is not available on this device.' That's a valid result: for your machine, H.265 is the practical choice. There's no software AV1 to compare against here.
Different framerate between the two encodes
Invalid comparisonIf you leave one encoder on auto framerate (30) and set the other to the real fps, the encodes have different frame counts and the size/quality comparison is meaningless. Set the same Framerate (the source's real fps) on both.
Both outputs are video-only
By designNeither encoder carries audio through the hardware path, so both comparison files are silent. That's fine for a codec test — just don't conclude 'audio quality' from these. For audio-bearing output use the transcoder.
Judging quality by file size alone
MisleadingSmaller isn't automatically better — a smaller file at the same bitrate setting means nothing without looking at it. Always view both outputs at 100% on real footage (motion, gradients, grain) before deciding. Bitrate is the lever; perceived quality is the judge.
Decode-reach not in the numbers
Deployment factorAV1 may win on size/quality but lose on audience reach if your viewers have older devices. H.265 (and H.264) decode on far more hardware. The 'winner' depends on who watches, not only the encode metrics.
Container differences
ExpectedBoth downloads are named with their codec suffix and may need a re-mux for strict players. That doesn't affect the size/quality comparison — judge the encoded stream, not the container label. Re-mux via the web optimizer if needed.
Tiny / already-compressed source
Noisy resultComparing codecs on a low-bitrate, already-compressed clip produces noisy, unrepresentative numbers because there's little detail left to differentiate. Test on a high-quality master for a meaningful comparison.
Encode-speed varies by GPU and content
ExpectedAV1 vs H.265 hardware encode speed differs by GPU generation and scene complexity. Both are usually near real-time on Ada/Arc/M3+, but don't extrapolate one machine's timing to another.
Both require Pro + Media
Tier gateBoth the AV1 and H.265 hardware encoders are gated to the Pro + Media tier. You need that plan to run the comparison. Per-file video ceiling is 100 GB streaming on that tier.
Frequently asked questions
Which is smaller, AV1 or H.265?
At matched quality, AV1 is typically ~30% smaller — and JAD's defaults reflect it (AV1 auto 0.04 bits/pixel vs H.265 0.06). The honest way to know for your footage is to encode both, match quality by eye, and compare the resulting file sizes.
How do I make the comparison fair?
Pin the same Bitrate and the same Framerate on both encoders, then compare quality at that fixed size. Or match quality by eye and compare sizes. Don't compare the two auto modes — their per-pixel targets differ by design (0.04 vs 0.06).
Why does the AV1 auto file come out smaller than H.265 auto?
Partly real efficiency, partly because AV1's auto target (0.04 bits/pixel) is lower than H.265's (0.06). For a true codec comparison, pin equal bitrates rather than trusting the differing defaults.
Which encodes faster?
Both are hardware encodes and usually near real-time on RTX 40-series / Arc / Apple M3+. AV1 NVENC vs HEVC NVENC speed differs by GPU and content — time each at identical settings on your own machine.
Can I compare audio quality too?
No — both encoders are video-only, so both comparison files are silent. This is a video-codec test only. For audio handling, use the transcoder.
What if AV1 won't encode on my computer?
Then for your hardware the comparison is already decided: H.265 is the practical choice. AV1 hardware encode needs RTX 40-series / Arc / Apple M3+; without it the AV1 tool refuses, and there's no software AV1 to fall back on.
Is smaller always better?
No. A smaller file at the same bitrate setting means nothing until you watch it. Always inspect both outputs on representative footage. And the deployment winner also depends on which codec your audience's devices can decode.
Why does H.265 still matter if AV1 is more efficient?
Reach. H.265 hardware encode is nearly universal and H.265/H.264 decode runs on far more devices, including older phones and TVs. AV1 wins on efficiency and licensing; H.265 wins on compatibility. Pick per audience.
Do both run without uploading my video?
Yes. Both encode locally in your browser on your GPU. Your clip never leaves the machine, which also makes the comparison private.
What about comparing against H.264?
Both AV1 and H.265 beat H.264 substantially on efficiency, but H.264 has the widest reach of all. For that comparison, encode H.264 via the transcoder and add it to your test.
Why did my comparison give weird numbers?
Usually one of: comparing auto-vs-auto, mismatched framerate between the two encodes, or testing on an already-compressed clip with little detail left. Pin identical Bitrate + Framerate and use a high-quality source.
What plan do I need to run both?
Pro + Media — both the AV1 and H.265 hardware encoders are gated to that tier. The per-file video ceiling there is 100 GB streaming.
Privacy first
Every JAD Video tool runs entirely in your browser via WebCodecs and FFmpeg (WebAssembly). Your video files never leave your device — verified by zero outbound network requests during processing.