How much does deshaking increase compression? Test.

[-Habanero-]

I wanted to see how much it would increase efficiency for x265 vs x264 and I assumed that it would help considerably for x264 but less with x265 because it's more modern and would compensate the motion better but I was wrong.

For the original shaky footage:
x264 CRF16 with maximum settings produced 2304 kb/s.
x265 CRF16.6 produced 1803 kb/s for the same SSIM. 28% reduction.

Video was always stabilized with Vdub deshaker.

x264 CRF16 produced 1612 kb/s. 43% reduction from shaky original.
x265 CRF16.6 produced 1266 kb/s. 42% reduction.

Video was then professionally stabilized to be 100% stationary.

x264 CRF16 produced 1662 kb/s. 39% reduction.
x265 CRF16.6 produced 971 kb/s. 86% reduction.

It seems x265 benefits more greatly from stabilization than x264. Why the substandard Vdub deshaker increased compression for x264 over professional methods is unclear.

[Skiller]

Why the substandard Vdub deshaker increased compression for x264 over professional methods is unclear.

Because Deshaker by Gunnar Thalin for VirtualDub is pretty darn good at motion tracking and has plenty of sophisticated options for border treatment.

[-Habanero-]

Because Deshaker by Gunnar Thalin for VirtualDub is pretty darn good at motion tracking and has plenty of sophisticated options for border treatment.

It's not better than professional stabilizers like mocha pro. The video I uploaded was the mocha-stabilized result. The output of the virtualdub deshaker had more movement yet was compressed more by x264.