MEGUI AutoEncode Bit-rate setting bug?

[hasse69]

I am having some issues with the settings in the AutoEncode of MeGui.
If a manually calculate a desired bit-rate based on a given QF and enter the resulting kpbs instead of specifying a size, the result of the x264 encode still seems to reduce the final bit-rate with some 100kpbs due to audio being added/muxed into the stream.
It seems that MeGui still uses the size that is indirectly calculated by the specified bit-rate. But if I specify a bit-rate instead of a size it means that size is of no importance, but video bit-rate is. Then adding audio to the stream should not affect the bit-rate of the video.
I guess a workaround could be to first encode the video, then mux the video+audio streams. But I like the AutoEncode feature :(
Could someone please tell me if I am doing something wrong?

[Sharktooth]

you're doing it wrong.
if you need a certain quality, use a CRF preset. if you need to fit a filesize use a bitrate based preset.
it is so simple...

[hasse69]

Guess you are right! ;) I will try a CRF18 encode to get a rough estimation on how far off I am.

[atreides93]

I think 18 will produce huge files...the bitrate will be all over the place based on the video content.
So the overall size will be unpredictable. I used to play around with that, but didn't like how some results were so big...now I use an average bitrate of around 6500 for most movies at 1080p....seems to work fine.

[hasse69]

I used CRF18 because that is the recommendation in other threads for a pure compressibility test before doing an ABR. I guess this thread became somewhat off-topic. However, the fact remains, the AutoEncode function of MeGui does not seem to work properly if one specify a bit rate instead of a file size. The final bit rate is still reduced due to the fact that audio is being added. That was *not* what I expected when specifying a bit rate instead of a file size.

Edit. Btw, doing a comp. test using CRF18 seems to become exactly the same as a QF=0,18 ABR??
Is this a coincidence or could that actually be true? I tried on 3 different sources with different target resolution. If it is true then I do no see the point on doing a CRF18 test. This also puts even more interest in my original question.