What can you make on
your own card?

Why open weights

The tools got good because
people gave them away.

Wan, LTX-Video, HunyuanVideo, Flux, Qwen — every model you'd use here exists because somebody released the weights instead of renting them to you.

The custom node you rely on was written by one person on a weekend and posted for nothing. The trick that finally got your character consistent across shots came from a stranger in a Discord who had no reason at all to tell you. That's a remarkable way for a field to work, and it's how this one has worked so far.

It's also fragile. Nearly all the money in AI video is going into models you can only rent, and the good workflows keep getting posted into threads that are unsearchable a month later. Nobody's paying to document any of it, because there's nothing to sell you at the end.

So no, Veo, Kling, Sora, Runway, Luma and Midjourney aren't allowed here — and that's not us picking a fight with them. They're good. Sending a job off to a datacenter genuinely gets you more power than a card in your desk, and pretending otherwise would be silly. We're not trying to prove they're bad, and we're not competing with them.

We're interested in the other question, which nobody seems to be asking: what's actually possible on the machine you already have? There are neat things happening locally right now — character consistency out of quantised models, sound generated on the same box that made the picture, tricks people invented at 2am and posted to a Discord. Almost none of it gets shown off properly, because the showcase always goes to whatever had the biggest render farm behind it.

Open source thrives on demonstration. People adopt the free tools when they see somebody do something great with them — not when someone argues they're philosophically preferable. So the rule isn't a protest. It's a focus. Everything here runs on tools anybody can download and use for nothing, and the point is to find out how far that actually goes.

Let's prove it's further than people think.

Who's doing this

We're two brothers
who make animations.

We've been in this scene a while — thousands of renders between us, most of them bad, a handful we're properly pleased with.

It all runs locally, on cards we originally bought to play games on. We know the ComfyUI graph held together with custom nodes because we've got one. We know the feeling of leaving something going overnight and coming down in the morning to find out whether it worked.

And the thing that kept nagging at us: we'd see something genuinely great scroll past — someone's character work holding across eight shots on a 3060, a sequence that obviously took forty attempts to land — and it'd pick up a dozen upvotes and vanish. There's a huge amount of good work in this scene that nobody ever really gets to see, and no occasion built around looking at it.

So: what if there were somewhere that made a bit of a fuss about it? Not a platform, not a startup, nothing to sign up for beyond an email. A challenge, a screening, an archive, and a small pot of money to get people moving.

About that $500

It's small on purpose, and we think that's a feature. Small enough that nobody's faking a render log to steal it, and small enough that the whole thing stays about the films rather than the prize. Nobody's turning up here for the payday.

But it isn't nothing, either. If you're building something on your own, $500 is a card upgrade, or a month of not thinking about it. That felt like the right size — enough to matter to somebody indie, not enough to distort anything.

Honestly, we'd run this for nothing. We just figured if we're asking people to spend a few weekends on it, we should put something on the table to get it started. Why not, really.

Fancy judging instead?

Some people would rather
watch than make.

If that's you, we'd genuinely love the help. Sitting with fifty films, working out which ones actually landed and arguing the case for them, is its own kind of fun — and it's a real contribution.

We could go and borrow some recognisable names for the panel. We'd rather it came from people who use these tools. So there's no invitation-only tier: send us a couple of links and a line about what you make or what you're into, and that's the whole application. No CV.

You can't judge a round and enter the same one, obviously. Past that we're not precious about it — judge one and enter the next, judge every one, dip in when you fancy it. Whatever suits you.

And if you think there's a better way to run a panel, tell us that too. We've made a lot of this up. Plenty of it will turn out to be wrong, and we'd much rather hear a good idea early than defend a bad one later.

Whoever ends up on it goes up on this site by name before signups open, so entrants know who's deciding. There's a separate community vote as well — the panel isn't the only voice in the room.

Put your hand up →

Not the only game in town

There are bigger ones.
Go and enter those too.

There's a whole calendar of AI film competitions running right now, several with prize pools that make ours look like pocket change. We keep a list.

It seemed daft not to. If you've made something good, you should be entering it everywhere it fits — and we'd rather point you at the ones with $65,000 on the table than pretend they don't exist. Some of them are excellent.

See what else is open →

And afterwards, we'll publish the complaints

After every round we put out a retrospective — what we paid and when, what people told us, and what we're changing because of it. Including the criticism, unedited.

We can promise that credibly because we're small — at this size we'll be reading everything people send us anyway.