Should I start encoding with AV1?

[mtommy2008]

Hi, I wonder, should I start encoding with AV1, rather than H.265?
I read that AV1 compresses much better than H.265 and now it also supports HDR.

Which AV1-encoder should I use for archiving video?
AOM-AV1 or SVT-AV1? I hear that AOM-AV1 gives better video-quality than the faster SVT-AV1.

[Katie Boundary]

No, start with mpeg-1 and work your way up LOL.

But seriously, Xvid has an excellent balance of speed, compression efficiency, and compatibility with other programs and file formats (like AVI). I'd recommend starting with it.

[GeoffreyA]

It depends on your use case. AV1 has considerably better compression than H.265, but if you want to archive video, you may find that the picture, while excellent, comes out softer than expected. It leaves older codecs behind at lower bitrates and does quite well with anime. From a compatibility point of view, AV1 may not play on one's TV or other hardware devices.

AOM has marginally better quality than SVT-AV1 and is a lot slower. If you'd like to try AV1, perhaps experiment with the two encoders in FFmpeg. Quality is controlled by -crf as in x264/5, a lower value giving better quality. A lower -preset or -cpu-used leads to better efficiency but at the cost of speed; you will want to use values of 4 and 3, respectively, for a final encode.

ffmpeg -i INPUT.mp4 -c:v libsvtav1 -preset 6 -crf 25 -g 240 -svtav1-params tune=0 -c:a copy OUTPUT.mp4
ffmpeg -i INPUT.mp4 -c:v libaom-av1 -cpu-used 5 -crf 25 -g 240 -c:a copy OUTPUT.mp4

There is also the newer H.266/VVC. Fraunhofer's VVenC encoder is slightly ahead of AV1 and, arguably, a touch sharper; but its medium preset is impractically slow at present and has to be weighed against the minor gains over AV1. Use -qp to control quality.

ffmpeg -i INPUT.mp4 -c:v libvvenc -preset fast -qp 25 -period 10 -c:a copy OUTPUT.mp4

[Atlantis]

What I'm doing now is that I use H.265 for 2K and AV1 for 4K.

[Corn Xin]

Originally Posted by mtommy2008

Hi, I wonder, should I start encoding with AV1, rather than H.265?
I read that AV1 compresses much better than H.265 and now it also supports HDR.

Which AV1-encoder should I use for archiving video?
AOM-AV1 or SVT-AV1? I hear that AOM-AV1 gives better video-quality than the faster SVT-AV1.

Hey! AV1's compression and HDR support are tempting, but consider the trade-offs. If your target devices support AV1 and you have the processing power, go for it!

For archiving, I'd recommend AOM-AV1 for its quality benefits. It's slower, but worth it. If you're short on time, SVT-AV1 is a good alternative.

[dracore]

Yes, AV1 compression is very good and HDR support is a big plus.
If you want the best possible quality for your archive, I recommend the AOM-AV1.
It will be slower, but it's worth it.
The SVT-AV1 isn't bad either; it's just a good compromise between speed and quality.
Either way, it will be exciting to see how these encoders continue to develop and improve.

[Forteen88]

Originally Posted by GeoffreyA

ffmpeg -i INPUT.mp4 -c:v libsvtav1 -preset 6 -crf 25 -g 240 -svtav1-params tune=0 -c:a copy OUTPUT.mp4
ffmpeg -i INPUT.mp4 -c:v libaom-av1 -cpu-used 5 -crf 25 -g 240 -c:a copy OUTPUT.mp4

Sharpness seems to be an important option for keeping details in AV1 too,

The **"Sharpness"** option in SVT-AV1 via FFmpeg isn't just a basic sharpening filter—it plays a deeper role in how the encoder balances **detail preservation** and **compression efficiency**.

### What "Sharpness" Really Does in SVT-AV1

The `sharpness` parameter in SVT-AV1 affects **how aggressively the encoder smooths or retains fine details** during compression. Here's a breakdown:

- **Not a post-processing filter**: It doesn't apply sharpening to the final image like a typical image editor would.

FFMPEG SVT-AV! seems to set default "sharpness=0", while I have heard that the original (non-FFMPEG) SVT-AV1 keeps "sharpness=2" as default.
"sharpness=4" was good on the last source-video I tried it on.

[GeoffreyA]

Originally Posted by Forteen88

Sharpness seems to be an important option for keeping details in AV1 too,

FFMPEG SVT-AV! seems to set default "sharpness=0", while I have heard that the original (non-FFMPEG) SVT-AV1 keeps "sharpness=2" as default.
"sharpness=4" was good on the last source-video I tried it on.

Yes. The fork SVT-AV1-PSY has a lot more variables to adjust, and can preserve detail and grain tolerably well. There are psy settings and different tunes. Some features were merged into mainline SVT-AV1.

https://github.com/psy-ex/svt-av1-ps...ses/tag/v3.0.2

[Z2697]

Default sharpness is also 0 in mainline CLI, and it's not that important really, it controls in-loop deblocking filter, and a bit of rd process (why do both in one parameter tho?).

[Forteen88]

Originally Posted by GeoffreyA

Yes. The fork SVT-AV1-PSY has a lot more variables to adjust, and can preserve detail and grain tolerably well. There are psy settings and different tunes. Some features were merged into mainline SVT-AV1.
https://github.com/psy-ex/svt-av1-ps...ses/tag/v3.0.2

Yeah, SVT-AV1-PSY is great for grainy live-action source-videos, but for tutorials (like Visual Studio screen capture, not much grain) source-videos, plain SVT-AV1 is good.

Originally Posted by Z2697

Default sharpness is also 0 in mainline CLI

Oh, M$ Copilot-ChatBot is wrong as usual then Thanks.

[GeoffreyA]

Originally Posted by Forteen88

Yeah, SVT-AV1-PSY is great for grainy live-action source-videos, but for tutorials (like Visual Studio screen capture, not much grain) source-videos, plain SVT-AV1 is good.

SVT-AV1 should have a screen-content tune, which would help in that regard, but I'm not sure about it. Funny enough, I found that for screen content, or even live action, Intel's hardware AV1 encoder did well. In Resolve, I used the Intelligent Quality Control setting at 8, and the results were transparent and small.

[Forteen88]

Originally Posted by GeoffreyA

SVT-AV1 should have a screen-content tune, which would help in that regard, but I'm not sure about it. Funny enough, I found that for screen content, or even live action, Intel's hardware AV1 encoder did well. In Resolve, I used the Intelligent Quality Control setting at 8, and the results were transparent and small.

Yeah, screen-content tune would be great. Currently I use this SVT-AV1 commandline for screen-content often:

ffmpeg.exe -i "input_video.mp4" -c:v libsvtav1 -preset 2 -crf 24 -g 300 -svtav1-params lp=4:tune=0:lookahead=80:scd=1:scm=1:sharpness=2:enable-overlays=1 -vf "cas=strength=0.5" -metadata:s:v:0 encoder="SVT-AV1 v3.1.0-6-g380af557 (preset 2, crf 24, g 300, tune=0, lookahead=80, scd=1, scm=1, sharpness=2, enable-overlays=1, vf(cas=strength=0.5))" -c:a copy "output-av1.mkv"

[GeoffreyA]

Thanks. I might find it handy because I often encode screen content, but hadn't worked out a proper SVT-AV1 command line for that. A Google search reveals that scm=1 activates full screen-content mode.

[Forteen88]

Originally Posted by GeoffreyA

A Google search reveals that scm=1 activates full screen-content mode.

Yeah,

scm=1 — Screen Content Mode
Purpose: Optimizes encoding for screen content, such as desktop captures, presentations, or animations.

Effect: Enables tools and heuristics that improve compression efficiency for content with sharp edges, text, and limited color variation.

Use Case: Ideal for videos that are not natural scenes—think screencasts, software tutorials, or slideshows.