Added: This answer was based on advice I received that the SU27 had auto-ejection capability.
In a comment xxx states that the SU27 does not have auto-eject capability.
I have not yet obtained supporting factual references for either point of view. I will update this answer when I know more.
If the SU27 does NOT have auto eject capability then the crew members were extremely brave, regardless of what responsibility they did or did not have for the crash. I viewed a video of the crash and did a frame by frame examination from 1st contact with the ground by a wing until ejection.
The video speed was 25 frames per second.
The video was never "super clear" and became increasingly less so as the crash progressed (apparently due to the videographer attempting to not die).
From the time the wing 1st touched ground 4 frames (0.16S) elapsed until certain canopy motion was observable. It may have been starting to move for 1 to 2 frames before this.
The canopy was well clear by frame 8 (0.32 seconds post contact) and one crew member ejected at frame 11 (0.44 s post contact). The aircraft was moving out of frame but it appeared essentially certain that a prior ejection had not occurred so the 2nd ejection probably occurred some time after frame 11 (after 0.44s post contact).
While ejection initiation would have occurred at some time before frame 4 when canopy motion was first noticeable it would almost certainly have been a very short period prior, so it is very likely that the aircraft wing had touched the ground at or before ejection initiation.
That's brave.
Also stupid, if you want to live.
But very brave.
_______ End of addition. ______________________________________
Yes.
This is a frame grab from a video about a Ukrainian airshow disaster when an SU27 crashed into the crowd with many fatalities. Both crew members ejected essentially "on impact" and I'm told that the ejections were automatically triggered. Both survived.
The ejection shown here obviously had to be preceded by a trigger, canopy eject, fire, ... sequence which I'd expect to last at least some "tenths of a second", so the actual decision point is uncertain.
