Inspiration

i keep seeing so much bland, lifeless ai content everywhere—stuff that technically works, but just doesn’t make you feel anything. the idea was to flip that on its head: could we help people easily create visuals that actually capture the vibe they want, using both ai and a bit of human touch? that’s where mood meter started.

What it does

mood meter lets you pick the feeling you want your design to give off, be it through text, voice, or camera (a bit of fun). then, it instantly suggests matching colors, fonts, and layouts. you can also run it on an existing design to see if your work says happy when you meant calm, and it’ll give easy tips to tweak the vibe.

effective sales and marketing is always about driving emotion (did you watch mad men? they would've loved this), and social media is the hot place to grab attention, evoke emotion, and drive sales.

How we built it

built with javascript using the adobe express add-on sdk, spectrum web components. we tied in sentiment analysis (for text), simple face and audio analysis (for camera/voice) and mapped those to color palettes and font choices.

currently, all the results are done via manual mapping and rules. this is obviously not scalable, sustainable or suitable for production! however, this will be done with an LLM API very easily! i'm really excited to see what results could come of that and some learnings from those analyses. designers could really hone their content based on psychology as opposed to their subjective decisions.

Challenges we ran into

  • time constraints (as ever!)
  • cache problems with adobe express. sometimes refreshing wasn't enough and had to disconnect or refresh

What we learned

  • it's actually really easy to build an add-on, and that local server connection is sooo easy to use! 10/10 dev experience. can we expect hotloading at some point?

What's next for mood meter

  • leverage AI to use trained data to go deeper into sentiment analysis
  • more nuanced emotions
  • improved design

Built With

Share this project:

Updates