av1an duplicating some params

[motbob]

Windows, a33162f nightly.

C:\Users\sing4\Desktop\video\av1an\av1an.exe -i dandadan05_simple.vpy -e aom -v "--bit-depth=10 --input-bit-depth=10 --cpu-used=3 --end-usage=q --cq-level=21 --threads=2 --kf-max-dist=999 --kf-min-dist=999 --arnr-strength=0" --scenes dandadan05_scenes.log --extra-split 275 --photon-noise 1 --chroma-noise -w 6 -o dandadan05.ivf

results in

I set cpu-used=3, cq-level=21, and threads=2. You can see in the screenshot that the encoder is showing cpu-used=6, cq-level=30, and threads=8 and the desired cpu-used=3, cq-level=21, and threads=2. I don't know which is actually taking effect (probably the latter).

Using --no-defaults fixes the problem.

[emrakyz]

This is intended.
Latter takes effect.

Use --no-defaults if needed.

[motbob]

Respectfully, could you reevaluate whether this is desired behavior? The intuitive result of this is to make the user worry about whether their settings are being honored or not.

[emrakyz]

@motbob

Yes, it is a desired behavior. This is a power-user tool in the end, the user should know/understand eventually.

Additionally, it's shown in the code, explained in the docs & help page and there is a specific parameter called --no-defaults.

Most people who encode videos also know that encoders override previous parameters with later parameters. If not; then they will find out.

The defaults are there for the cases when the user doesn't input any of default parameters.

[nekotrix]

There is no problem since the latter are overwriting the former. The defaults themselves are questionable, but it's at least preferable than the even more questioning aomenc defaults (cpu-used=0 namely...). --no-defaults is a parameter created to disable this very behavior and nothing else.

[motbob]

I am surprised that it is so hard to convince you that an encoder displaying options that aren't actually taking effect is a bad thing. But whatever, I just ask that you remember back to this thread if you get confused feedback from users in the future.

[nekotrix]

I may be wrong on that, but I'm pretty sure this behavior has been in effect since the day I used Av1an for the first time in late 2021. Never really heard any issue stem from that.

[shssoichiro]

The concern here is that displaying both is likely to confuse users, correct? While indeed the behavior of all encoders supported by av1an is that the last parameter takes precedence if duplicates are specified, I can understand why users may be confused by this output.

This probably is worth resolving from a UX perspective.