American Pie, 1h35m, x3, CQ_VBR 11.132 (I managed to get the sample size within 50 bytes of the target ), no error margin.

Desired video size: 721,354,850
Actual video size: 714,300,328
Difference: 0.98%

Currently encoding K19 ( 2 hours, 18 minutes ) with the LBR. We'll see in 1 hour 47 minutes the result
So far it seems that nobody has missed the target

-kwag

Here's the result on K-19:

Predicted file size: 692,084
Final file size: 673,912

Diff: ~2.62 %

This was with .99 as factor. Maybe we can forget the %factor now and leave it at 1.0

-kwag

That's what I've been doing and I haven't gone over yet.

My latest results (no audio):

Predicted file size: 682,61mb
Final file size: 681,97mb
Difference: 0.09%

This was with 1.0 as factor.

So bye bye factor . Just leave the option on Predictor, etc., for any future anomalies

-kwag

just finished spiderman

predicted video 1489.153
actuall video 1490.128

used 1factor as well