Snap ML Contest

You'd have to generate .XML files, figure out how the chart files are structured, re-generate those.

Then you need to somehow create interesting and unique songs from those pictures, perhaps an existing AI exists to do that, not sure if there are any APIs for that though.

The hardest part is actually making it work, if you want anything interesting to happen you'd also have to make it compatible with Haxe for it to actually generate new code, then compile it, and anything else you need for the FNF engine, overall it seems like a massive undertaking.

Edit: Thinking about it for a bit longer, how do you plan to make the charts from the songs and have them be accurate? And all from a single image, to remind myself, wow.

i have a sorta idea involving colors to generate everything. ill explain it later

so kinda like the old fnf chart

kinda

Let's make some kind of prototype as a PoC.

So, it finds alot of colors in the image, and uses the color numbers to generate custom notes, charts and music. Thats bassically the jist of it.

@sir_kitten2 can we have a week more to make it?

yeah this is kinda hard tbh, also @scratchmodification's cloud blocks are resetting so im missing a day

I'm gonna try to make our prototype.

joining is closed.

What are they?

What's that?

Supervised, unsupervised, and reinforcement.

Just posted a News article talking about this.

@everyoneworkingonthiscontest you don't seem to be doing anything. I will extend the deadline to April 10th.

thank you!

So i guess by supervised and unsupervised you mean moderated and unmoderated, but what does reinforcement mean in this context?

Yes thank you! We needed more time.

wait, what if i make people agree to some terms that i am not responcable for ANYTHING that is in there?

For me, that's probably actually harder than reinforcement.