With your text as an input, the AI system will select a relevant meme and overlay your text with hopes to give you and others a quick laugh. We hope you and your friends enjoy creating memes.
@nathan_reidy decided to share the code publicly (GitHub link below) in case anyone was curious to learn how it works in detail. Basically there are two steps for the AI, (1) a semantic search for a meme type with the closes description and then (2) a few-shot completion to parse the words into pieces to be overlayed.
https://github.com/joshbickett/m...
Funny concept Josh, but I'm always getting the same meme image when I generate it (the first one in the video of the lonely guy in a party). Is the same happening to anyone else?
@pedrobh thanks for this feedback. I’ve heard this feedback from others. Other images should be showing up more frequently. I will improve this with the next version.
Report
Hey, @josh_bickett1
Loved the concept, its LIT!!!
Is there a trial version without signing up?
If not, I would strongly recommend allowing users to make at least 1-3 memes to get hooked before they sign up.
The reason I am asking is that I get the "Make Meme" button greyed out.
Do correct me incase I am doing something wrong.
Thanks in advance
@heffendi signing up is required in order to make the memes.
I am going to look into allowing 1 or 2 memes before sign up, but knowing the customer is important for advanced AI systems to ensure they are used appropriately
Report
@josh_bickett1 Awesome! I will definitely be signing up to explore more.
Cheers!
Report
Can you explain or outline the high level process? I am assuming memes are labeled with different features and then a pattern match based on the text input. Text is then divided and overlayed on image? I am curious!
@aakash_barot Yes, you have it mostly correct. Each meme template is labeled with a description. The user's input is checked against all meme description and finds the closest one with a semantic search. After that, a completion with few shot examples is used to parse the user's input into text to overlay on the image. The code is available on Github in case you want to take a closer look!
github.com/joshbickett/makememe_ai
In the last year powerful AI systems became available to developers and I started wondering what could be built with these systems. A friend mentioned automating meme making and that captured my imagination.
There are many questions about the future of AI, but one of my favorites is - will it have a sense of humor?
Find out - makememe.ai beta is open to new users
Report
Love the idea! Huge fan of imgflip and I use it a lot for my Twitter content + slides soon 😁.
Instead of just giving us one image, it would be nice to be able to shuffle a few alternatives.
Copy.ai does that really well.
@sofya_narbut Thanks for trying it out. Could you imagine yourself returning to the website next time you have a meme in mind to create or was it just fun for a one time use?
Report
@josh_bickett1 I rarely make memes, but I might do it again to mess around together with friends
Talk to AI Human
Personal Assistant by HyperWrite
The Newsroom
Personal Assistant by HyperWrite
Personal Assistant by HyperWrite
Personal Assistant by HyperWrite
Personal Assistant by HyperWrite
Personal Assistant by HyperWrite
Personal Assistant by HyperWrite