Artscroll

An art doomscrolling platform designed for discoverability and education of new art collectors.

Comment

Inspiration

During our first meeting, we were all doomscrolling. We then realized the potential to use this trend as a way to entice/educate new art collectors.

What it does

Artscroll is a website where users can doomscroll through short-videos showcasing art. These videos are narrated and can zoom in on relevant features to support the explanation. Artists can simply upload an image of their art and fill in relevant details (name, price, genre, style, where to purchase, etc.), and a video will be automatically generated.

How we built it

We used Manus AI to build and deploy the website. A Kaggle WikiArt (https://www.kaggle.com/datasets/steubk/wikiart) database is fed into the system, a sample description for one painting is provided, then the system automatically generated all the videos (including the script).

Challenges we ran into

At first, many functions did not work. The UI was also not user-friendly and was not suitable for presenting art.

Accomplishments that we're proud of

The videos are successfully generated and proper, with the auto zoom-function.

What we learned

How to efficiently incorporate Manus AI into our workflows to create a working prototype, and how to feed a dataset into the website during development.

What's next for Artscroll

Personalized Taste Profiling Learn user preferences from scrolling behavior to recommend artists, styles, and price ranges they’ll actually like.

Museum & Artist Partnerships Collaborate with museums, galleries, and artists to feature curated works, exhibitions, and verified collecting info.

Deeper Interactions Likes, saves, swipe signals, collections, and comparison feeds to make art discovery more addictive and social.

Higher-Quality AI Narration More natural, expressive text-to-speech with better pacing, tone, and emotional emphasis for storytelling.

More Context-Aware Zoom Zooming can work better by actually analyzing the content of the painting, not working solely based on the description.

Built With

Share this project:

Updates