I would expect the Sundara's pad / yoke design to seal better on a rig with a flat plate around the pinna than on a real human's head. The material facing the skin is leaky and won't be fully in contact with the skin on a real human (but will be with the rig's plate).
But I suspect that there may be quite a bit of leak (if not most of it) coming from the pad to earcup attachment given its design, and possibly quite a bit of sample variation coming from that area of the design (it's very poorly conceived to ensure a consistent degree of seal).
The two copies of the 2020 Sundara I had (or rather one and a half as one side didn't work out of the box on the second) and my HE400SE also drop off in the sub bass quite a bit on my head, similarly to headphones likes the HD650 for example, as can easily be attested by on-head measurements.
Below how the R and L channels of my own HE400SE (solid green traces) measure on my own head with blocked ear canal entrance microphones vs HD58X and HD650 (dotted traces), normalised across one octave centered at 500hz :
View attachment 171054
And here the first copy of the 2020 Sundara I received vs. the
same HD650 with open ear canals in concha microphones (at these frequencies the results are similar to blocked ear canal entrance microphones),
measured months earlier (I didn't have blocked ear canal mics at the time and I no longer have the Sundara), R and L channels :
View attachment 171064
Please don't compare both graphs between each others, but rather the HE400SE vs HD650 in the first and Sundara vs HD650 in the second in relative terms, as the mics compensation is a little bit different and I've changed measurement practices over time.
So for me both performed more or less in the vein of the HD58X/HD650.
Resolve seemed to experience a much better bass response with his Sundara on-head than I did.