[Question] Am I using hqdn3d vfilter correctly?

[9Oc]

I am doing sample encodes of movies with and without the hqdn3d de-noise filter with conservative temporal de-noise settings, but when doing sample-encodes the file size reduction gained on the samples with hqdn3d on vs. off seems absurdly high. I think that it may simply be using the default hqdn3d settings when no values are supplied which enables both spatial and temporal de-noising at medium values (found here: https://ffmpeg.org/ffmpeg-filters.html#hqdn3d).

My ab-av1 commands:
De-noise: ab-av1 sample-encode -e libx265 -i movie.mkv --vfilter "hqdn3d=0:0:3:3" --pix-format yuv420p10le --preset 6 --crf 20
No de-noise: ab-av1 sample-encode -e libx265 -i movie.mkv --pix-format yuv420p10le --preset 6 --crf 20

I also tried using the syntax found on the ffmpeg documentation like so (with the same result as the syntax above):
--vfilter "hqdn3d=luma_spatial=0:chroma_spatial=0:luma_tmp=3:chroma_tmp=3"

The difference in file size between these two samples sometimes ends up being over 60% less than the samples without hqdn3d enabled. Something else that I noticed is that when encoding the full movie with handbrake both with and without hqdn3d enabled with my desired settings for only temporal de-noising, the file size reduction is much more moderate (about 5-10% less) which is in line with what I would expect with very light temporal de-noising values.

So my question is, what am I doing wrong? Or is ab-av1 having some sort of issue with the hqdn3d filter?

Any help would be appreciated.