A “ghost playhead" which ALWAYS marks the timeline location of the currently displayed frame
(Perhaps "current view indicator" is a better name than "ghost playhead.")
Here is the problem: The (cyan, user-controlled) playhead does NOT necessarily correlate to the frame that is being shown in the program or source monitor.
Sometimes, an old frame continues to be displayed, even though the user clicked elsewhere on the timeline (or "time ruler," as Premiere calls it.)
Depending upon computer specs, network bandwidth and latency, video codec, and other factors, it can sometimes take several seconds for the chosen frame to actually load and be displayed.
In the meantime, users need some kind of indication as to which frame is currently being displayed. This would be "the ghost playhead," which could perhaps be colored red.
If the normal playhead IS NOT on the same frame as the ghost playhead, it should be a different color, perhaps purple. This could be called the "Pending playhead."
If the normal playhead and ghost playhead ARE on the same frame, they should combine to form a different color - probably cyan as it is now.
Here's a video explanation:
Here's a link to a closely related feature request: https://adobe-video.uservoice.com/forums/911233-premiere-pro/suggestions/42186322--i-frame-first-fast-seeking-for-inter-frame-compr
Pierre Louis Beranek commented
Ghost frames would be overkill and a waste of programming effort IMO. Perhaps a better solution might be a playhead that changes color once the correct frame is loaded?
As for your FR for "I-frame first fast seeking" I agree 100% that this should be added!
David Johnson commented
As a pro, this is probably the simplest and most brilliant idea on the forum for improvement. Just show me both on the timeline where I want to choose, and where the program is trying to render for preview ... at the same time. Without this, I have to wait, to know if where I have picked is accurate ... and that time is variable. Giving me more information at this moment is critical for efficient use. I have exactly the same problem as this poster, and this should be fixed.