The Honest Codec Comparison
H.265 and AV1 are genuinely better compression standards than H.264. That statement needs to come first, because the argument that follows is not that codec upgrades are wrong — they're right, on the right timeline. The argument is that they're not sufficient on their own to solve a cost problem that compounds with data growth.
Mature standard
H.265
Also known as HEVC
vs H.264 file size30–50% smaller
Encoding compute2–4× vs H.264
Decoder supportBroad — most devices
LicensingPatent-encumbered
Transition timelineMonths to years
Next generation
AV1
Open standard, royalty-free
vs H.265 file size30–40% smaller
Encoding compute5–10× vs H.265
Decoder supportGrowing — not universal
LicensingRoyalty-free
Transition timelineYears for large deployments
File Size Reduction: What Each Codec Actually Delivers
Per-generational compression improvements are real, but they come with important caveats about encoding conditions, content type, and quality targets. These are realistic ranges rather than peak benchmarks.
File size relative to H.264 baseline — same visual quality
H.264 (baseline) Reference point
H.265 / HEVC ~30–50% smaller than H.264
AV1 ~50–60% smaller than H.264
H.264 + MForja entropy conditioning 50%+ smaller than H.264
H.265 + MForja Available today with H.265 infrastructure
AV1 + MForja Maximum efficiency — lower bar is better
Lower bar = smaller file size = lower storage and CDN cost. Figures are representative ranges — actual results vary by content type and encoding configuration.
The chart shows something counterintuitive: H.264 with entropy conditioning delivers comparable file sizes to unconditioned AV1 — the codec two generations ahead. And it does so without any decoder changes, CDN reconfiguration, or transition risk.
The Real Cost of a Codec Transition
The compression gain from a codec upgrade is real. The cost and complexity of getting there is also real — and frequently underestimated at the planning stage.
| Transition requirement |
H.264 → H.265 |
H.265 → AV1 |
Add MForja (any codec) |
| Encoder upgrades required |
✓ |
✓ |
None |
| Decoder/player updates |
✓ Often required |
✓ Required |
None |
| CDN configuration changes |
Sometimes |
Sometimes |
None |
| ABR ladder rebuild |
✓ |
✓ |
None |
| QC and compatibility testing |
Extensive |
Extensive |
Lightweight |
| Typical deployment timeline |
12–36 months |
18–48 months |
Weeks |
| Additional encoding compute |
2–4× |
5–10× |
Trivial overhead |
The deployment timeline problem
A codec transition for a large streaming platform or surveillance operator is rarely completed in under a year. During that transition, the majority of your content continues to be encoded and delivered under the old codec. The efficiency gain from the new codec accrues slowly, at the tail end of a long transition, while infrastructure costs compound throughout. A codec-agnostic improvement delivers from day one — across every codec you're running, including both old and new as the transition proceeds.
Why Codec Upgrades Don't Break the Cost Curve
Even after a successful codec transition — all the testing done, all the decoders updated, all the CDN configurations verified — the underlying cost problem reasserts itself.
Here's why: a codec upgrade improves the compression algorithm. It makes the encoder smarter at compressing a given stream. But it doesn't change the stream itself. The informational complexity — the entropy — of the underlying video data is the same before and after the codec upgrade. The new codec handles that entropy more efficiently, but it's still working with the same raw informational burden.
And after the transition, data volumes continue to grow. A year after completing an AV1 migration, a platform delivering 20% more content than at transition time has already given back a significant portion of the efficiency gain. The cost curve resumes, just from a lower starting point.
"Codec upgrades provide a one-time benefit that diminishes over time as the new codec matures and content volumes continue to grow."
The cost curve after a codec transition
Before transition — H.264, 10PB at $20/TB/month
$200K/month
Transition complete — AV1, ~45% smaller files
~$110K/month
2 years later — data volumes doubled, back on growth curve
~$220K/month
Same period with entropy conditioning — data still smaller
~$110K/month
The codec transition bought two years of cost relief before data growth consumed the gain. Entropy conditioning continues to apply to every byte encoded — the gain doesn't erode as volumes grow because the file size reduction is proportional to data volume, not absolute.
What Entropy Conditioning Adds to Each Codec
Entropy conditioning and codec upgrades solve different parts of the same problem. A codec upgrade improves the compression algorithm. Entropy conditioning improves the input the algorithm receives. Both changes deliver independent efficiency gains that apply simultaneously.
H.264 alone
H.264
Baseline
Reference point
Still the dominant deployed format worldwide
AV1 — no conditioning
AV1
~50–60%
smaller than H.264
Requires full decoder/CDN transition
H.264 + MForja — today
H.264 + MForja
50%+
smaller than H.264
No transition. Available now. No downstream changes.
H.265 — no conditioning
H.265
~30–50%
smaller than H.264
Mature — most devices already support it
H.265 + MForja
H.265 + MForja
~65%+
smaller than H.264
Outperforms unconditioned AV1
AV1 + MForja
AV1 + MForja
~75%+
smaller than H.264
Maximum efficiency — gains compound at every generation
The practical implication: if you're running H.265 today and planning an AV1 migration, adding entropy conditioning now delivers a result comparable to completing that AV1 migration — immediately, without the transition complexity. Then when AV1 is ready to deploy at scale, entropy conditioning compounds with it too.
Which Codec Should You Choose?
The honest answer depends on where your infrastructure is today:
If you're running H.264: H.265 is a mature, well-supported upgrade worth pursuing. AV1 is the longer-term direction, but hardware decoder support — while growing — is not yet universal across all devices you may need to reach. A pragmatic path is H.265 now with entropy conditioning, AV1 when decoder support matures in your target device ecosystem.
If you're running H.265: AV1 is worth planning for, particularly if your encoding infrastructure can absorb the higher compute demands or you have access to hardware AV1 encoders. The royalty-free licensing is also a meaningful long-term consideration. Add entropy conditioning now — it delivers immediately and compounds when AV1 is deployed.
If you're running a mixed codec environment: This is where a codec-agnostic approach has the clearest immediate value. Entropy conditioning applies to all of them from a single integration point. You don't need to complete any transition to start benefiting.
50%+
H.264 + MForja — available now, no transition
65%+
H.265 + MForja — outperforms unconditioned AV1
75%+
AV1 + MForja — maximum compounding efficiency
Frequently Asked Questions
Is H.265 or AV1 better for compression?
AV1 delivers better compression — typically 30–40% smaller files at equivalent quality compared to H.265, and 50–60% smaller than H.264. However, AV1 requires broader decoder support and significantly higher encoding compute. H.265 remains the most widely deployed high-efficiency codec in production. Both deliver meaningful improvements over H.264.
How much does H.265 reduce file size compared to H.264?
H.265 delivers approximately 30–50% smaller file sizes compared to H.264 at equivalent visual quality, depending on content type and encoding configuration. Achieving this requires encoder upgrades and decoder compatibility verification across all playback endpoints.
How much does AV1 reduce file size compared to H.265?
AV1 delivers approximately 30–40% smaller file sizes compared to H.265 at equivalent quality — roughly 50–60% improvement over H.264. AV1 encoding is computationally more demanding, and hardware decoder support, while growing rapidly, is not yet universal across all playback endpoints.
Should I upgrade to H.265 or AV1?
Both are worth pursuing on their own timelines. H.265 is mature, widely supported, and delivers meaningful efficiency gains. AV1 is the direction the industry is moving, with growing hardware decoder support and royalty-free licensing. Entropy conditioning via MForja works with whichever codec you choose and compounds with both.
Why doesn't a codec upgrade alone solve video infrastructure cost problems?
Codec upgrades deliver a one-time compression improvement at the point of transition. After that, data volumes continue to grow and costs resume their upward trajectory. Additionally, codec transitions require significant investment — encoder upgrades, decoder compatibility verification, CDN reconfiguration, and testing cycles that take years at scale. Entropy conditioning operates below the codec layer, delivers immediately, and continues to apply as volumes grow.
Can entropy conditioning be used alongside an AV1 migration?
Yes — and this is the highest-value combination. Entropy conditioning can be deployed on your existing H.264 or H.265 infrastructure today, delivering immediate file size reductions. As your AV1 migration proceeds, entropy conditioning continues to apply — the gains compound at each stage of the transition.
Make Your Current Codec More Efficient Today
No migration required. 50%+ file size reduction on whichever codec you're running.