Comparing loads and gameplay from XandoToaster's ps1 run: https://www.youtube.com/watch?v=aAeC-13wEyA

To Dayo's ps2 run: https://www.youtube.com/watch?v=pJ5KkN5WBgU

To John's duckstation run: https://www.youtube.com/watch?v=jpXcisqO-3Y

Toaster's run has music turned on

John's run has music turned off

Times written in frames at 60fps, to be summed at the end

After doing the any% comparison, it seemed that the gameplay differences in dragon times were negligable. They were not considered here.

As a result, this comparison isn't perfect, but it is close, and it simplifies the work needed significantly.

Load order is written following Toaster's route

Loads into levels start when the hud / sparx disappear

Loads into levels end when the background starts to move

Loads leaving levels through a menu start when the menu disappears

Loads leaving levels through a vortex start when the text appears at the top of the screen

Loads leaving levels end when the homeworld begins to appear again

An exception was made for Dark Hollow: Toaster left through the vortex, so john and dayo's loads leaving the level began when the text appeared at the top of the screen

Dragon loads start when the dragon pedastal lights up (which is the same frame the hud disappears)

Dragon loads end when the black bar at the top begins to recede

Dragon loads when done properly should be broken into the intro and outro, but past experience has shown that gameplay differences in dragon touch and cutscene skipping generally only make a difference of a couple frames per dragon. Any egregious differences in gameplay should have a note next to them.

That said, most but not all examples of sds beating fds on a dragon load are due to these very minor gameplay differences

Credits skip timing starts when Spyro's shadow appears on the final ledge in the Gnasty chase

Credits skip timing ends when Gnorc Gnexus begins to appear again

Death abuse timing starts when the hud disappears

Death abuse timing ends when the level begins to appear again

Note from Toaster: This appears to confirm a suspicion I've had that turning off the music makes death abuses marginally faster. John's deaths were consistently faster than dayo's despite not using fds.

IMPORTANT NOTE: Individual consoles can have their loads vary greatly. This spreadsheet should not be used as a final say on how much time fds saves. This is only a little bit of data in a much larger set we should be building.

NOTE: John hit Boris pretty far from the optimal spot here, which should account for most of this timeloss

NOTE: Toaster managed to hid Todor from the opposite side, losing ~1 second