The fact that the majority of this thread has now shifted focus to the "bumps before rotation" (an observed but unpredicted bump in the first ~22 seconds) rather than addressing Marik's core video claims is, in my honest view, somewhat telling.

I fully accept the high level of scrutiny applied to my contributions - the frustum roll from pitch, 1:1 bank derotation, cloud-motion triangulation, elevation tables, and the GoFast side-by-side control test.

That scrutiny is appropriate for any claim challenging the refined glare model. What stands out, however, is the clear disparity with the treatment of LBF's "pilot comfort" rotation model. That model was accepted and promptly implemented into Sitrec, even though the same GoFast side-by-side test demonstrated that the refined glare formula produces a physically impossible straight water path, while the natural frustum + bank method correctly reproduces the curved path seen in the actual footage.

Most notably, I have now asked the same specific question for months, both here and in the refinements thread:

In the Gimbal footage, the background clouds are clearly observed moving through the frame at an angle (not level/horizontal. "Micks puzzle"). Given that the default position must be that the artificial horizon and pod orientation data are accurate and reliable - a fundamental requirement for pilot safety and situational awareness - how does the refined glare model first justify overriding that default? Specifically, how do we determine that the angled background motion is not simply natural motion caused by normal pod elevation change and tracking, but instead evidence of an additional derotation mechanism?

This is the exact issue I have repeatedly highlighted in my posts.

The refined glare model requires this extra roll/ derotation mechanism, yet no one has pointed to any specific, demonstrable part of the actual ATFLIR footage that shows it in use - as opposed to the well-understood pod elevation dynamics.

There has been no valid, evidence-based reason provided for implementing LBF's model into Sitrec while leaving this central question (and the GoFast control-test failure) unaddressed. I'm not asking for agreement - I'm asking for a clear, specific explanation tied to the footage itself.

Until that is provided, the claim that the extra mechanism is "well demonstrated" remains an assertion rather than a demonstrated fact.

I welcome any direct response on this point.