sunnyba

I used gemini for the general art as well!

nice! I was personally surprised seeing scores roll in that were lower than mine, since I had played the level so many times while debugging haha.

Best in the system now is 17.1s and it must have been a near perfect run.

Thank you, I appreciate you sharing that!

thanks for the traffic everyone -- definitely hitting rate limits on gemini already so if it doesn't work please give it a try later in the day or try the example.

Apologies for the inconvenience! If anyone on the gemini team can help out, please lmk (email in bio).

Thanks! A few friends have starting using the generated digis as profile pictures because of how recognizable they are. I was definitely surprised at how good the models had gotten.

Thanks! Like another user commented below, I think the appeal of the pixel art sprites is that at a lower level of detail it can feel really accurate. 10 different real life red jackets can all end up as the same pixel art representation but each person would recognize it as their jacket! It feels like a form of compression.

On prompting, you can get most of the way there in AI studio on Gemini 2.0 Flash (Image Generation) Experimental by uploading a picture and asking for "a high quality detailed pixel art sprite of this character." Most of the backend annoyance here was iterating to improve prompt adherence (characters not facing the same way, outfits changing between frames, etc).

Thanks! There are some content filtering rules via Gemini, but if you got an error on Trump I'm guessing it's an issue with the image generation rate limit getting overloaded.

thanks for the kind words!

Saw some more granular charts on this for the different options here as well for those interested:

https://unusualwhales.substack.com/p/unusual-options-trading...

Really creative game! Great work. Echoing the suggestions in the other comments to provide some clue on the number of letters.

Is there a simple way to download the portion of the data that was breached for just our own account?

I've switched providers and deleted my LastPass account after this last breach but getting to see that would make it a lot easier to understand risks.

Implemented a history button after completing the game that should make it easier for you to share!

this is a great idea, let me see if I can make that easier!

Both great ideas. I'll look at the calendar-like feature.

Stats would be cool, I don't currently collect them but will think about adding them. Thanks!

interesting, thanks for flagging -- this looks to be a parsing issue when ChatGPT occasionally returns guesses in a format different than the one instructed to. I'll give this some more thought on how to fix...

Some technical notes. Even for something like this, the OpenAI API can be expensive - there's a few decisions I made that keep costs lower.

1. I cache guesses and responses in Redis when I first see them, so I don't need to hit the API for duplicate guesses (which are common when lots of people are guessing the same movies).

2. Emoji order doesn't matter (I apply a sorting to guesses).

3. Guesses are not path dependent, with each guess being treated as completely new. This is understandably annoying when GPT guesses the same incorrect movie multiple times, but it drastically improves the cache-ability and keeps me from hitting my API limits.

Happy to answer any questions here! If you'd like to reach out to me my email is in my bio.

country flags are enabled - the search displays the Italian flag if I type in "italy". However if I type "italian", it doesn't match to the flag unfortunately.

Ah that's annoying! I've swapped it out -- thanks for catching it.

thanks!

When I soft launched it 3 weeks ago I was doing TV shows, but 3x shows per day quickly exhausted my initial list, so I switched to movies over the past few days

It's free to guess any movie -- the prompt is like "guess the movie represented by these 3 emojis", no filtering at all.

It'll guess some crazy obscure stuff once in a while!

Thank you! Sadly I removed the thumbs up/down emojis and okay hand emoji -- when these were used in clues, GPT 3.5 tended to complain that the emojis were too generic and refuse to guess.

nice! I don't limit the search space in the system prompt to GPT, though each daily puzzle is using popular movies. I think for the top ~100 movies it'll probably do a fine job.

If this game goes on for a full year though, I imagine it's going to really struggle unless the movie is about something really unique.

Thank you!

I ended up trying to funnel people towards the emoji picker so that I could: 1) prevent unknown emojis outside the training window of GPT 3.5. As an example, ChatGPT said that this one was a box of falafel.

2) exclude certain emojis that had a high likelihood of leading to GPT refusing to guess (it would claim were too generic).

That's right, they're not available.

From testing as I developed this, I noticed that certain emojis often lead to complaints from GPT 3.5 about how the combinations were too generic. After some really basic testing I ended up removing the 3 worst offenders from the emoji picker: ['+1','-1','okay_hand'].

Should be back up, sorry about that! For anyone else hitting this you may need to refresh and try again in a few minutes -- the OpenAI API sometimes complains but I don't handle this so gracefully yet.

I asked ChatGPT for lists of movies and TV shows that it knows about and used those, to avoid getting into situations with movies published after the training window

Thanks a lot!

I disabled skin tones since they're encoded in an odd way, but it's on my future feature list to re-add them if people are enjoying it enough.

That's really strange! I should probably add some logic to strip whitespaces for that validation, thank you!

thanks a lot! I came across yours as well, great idea to publish.

Thank you!