Fair enough! I tested with Galaxian, and see 3.7 halflives per 1/60th of a second, or a halflife of 4.5ms. This means that the correct constant (with this TV) for the expression I gave above is k=1/13. This is low enough that on the NES, the effect will only ever be visible when transitioning to black pixels. So here's an animated gif, simulating same:
- anim.gif (15.52 KiB) Viewed 7287 times
It's
really subtle, especially without any of the phosphor size blur. You'll also want to enlarge it; I can't see anything when it's at 100dpi.
And after manually capturing 17 frames from FCEUX, here's how I processed it:
Code: Select all
for i in `seq 0 16`; do
pnmarith -maximum $i.ppm previous.ppm > n$i.ppm;
ppmtogif n$i.ppm > n$i.gif;
pnmgamma -ungamma 2.2 n$i.ppm | ppmdim .077 | pnmgamma 2.2 > previous.ppm;
done
pnmcat -tb n*.ppm | ppmtogif > all.gif
gifsicle --use-colormap all.gif -O3 -V -o anim.gif -d2 n{0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}.gif
Note that I'm fixing up the gamma; if I hadn't, it'd be even harder to see.
I suspect that standard 24-bit displays are not actually deep enough to show this in a compelling way; after two refreshes (1/13)² even full scale content is just 1 LSB.