This brief README describes how to reproduce the experiments I performed in order to determine the ACT interval time:

1. Create modified versions of SPR/ACT files that display the frame ID and adjust the display time
2. I used the poring sprite as it is very simple and just iterates through the frames in idle mode
3. Pupa might also work but it's not as straight-forward, so I elected to not use it after all
4. You can use the provided debug-sprites (the one without suffix use 128 time intervals for best results)
5. Note that you must adjust the camera (see video) since I didn't bother replacing all look directions
6. The timing in the ACT file needs to be long enough to see a difference, but fast enough to actually be displayed
7. Replace poring.spr and poring.act inside the data,grf (I used GRF Editor to do this)
8. You should see the displayed frame ingame (if at the right camera angle), and it will change very slowly in idle mode

Then, I recorded the video using OBS - make sure it's at 60 FPS and that the client also doesn't lag (needs 60 FPS as well).

How to generate frames:

You can step through the video with "e" in VLC player. I tried this initially but it's error-prone to manually count frames.

So instead, I recommend to use ffmpeg and dump the frames temporarily to disk:

mkdir frames
ffmpeg -i poring-test.mkv frames/poring-test_%04d.png

This will need several gigabytes (5 to 10) of disk space! Most of the frames are irrelevant junk, so feel free to delete them.

--

Relevant files:

* debug-sprites.zip: The modified debug sprites that use longer display time and show the current frame (in IDLE mode)
* poring-test.mkv: The video I used to dump the sprites from - you only need frames 106 to 300 or so
* ACT_FrameTimes.ods: The spreadsheet that holds the research notes and computed frame times