I'm well aware what triggers log truncation in simple vs full/bulk recovery model (checkpoint vs log backup). I'm also aware of the conditions that determine what VLFs will actually be cleared (if any at all), once the clearing process is triggered. Those clearing conditions are also different, depending on recovery model, is there mirroring, transaction replication, full/diff bacup in progress etc.

Checkpoint does not require any log records - crash recovery process does. But the checkpoint (better to say: the lack of it) can prevent VLF to be cleared (when clearing process is triggered by log backup).