AI E-Ink Gallery Engagement Study

ISEF Category: Technology Enhances the Arts

Subcategory: Display Technology  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

What if a wall frame could make a new painting every day from the weather, the sunrise, and the sounds in the room? That sounds like a tiny art museum with a mood ring inside. Your project asks a real question, do people keep looking when the art feels alive?

What Is It?

This project studies a display that makes generative art, then measures how people react to it over time. The frame uses an AI model to create one new image each day. The image can depend on inputs like weather, sunrise color, and indoor sound level. Think of it like a playlist, but for pictures. Instead of random art, the system gets a daily mood score from its inputs and turns that into a new visual piece.

The key idea is not just making art. It is testing whether a display feels more engaging when it changes in a meaningful way. A fixed-rotation baseline shows the same set of images on repeat. Your variable display adds novelty and context. You can then compare things like time spent looking, how often people stop near the frame, whether they remember the artwork, and whether they describe it as more alive, calming, or interesting.

Why This Is a Good Topic

This is a strong science fair topic because you can test it with real viewers and real data. You can measure engagement, compare two display styles, and use statistics to see whether any difference looks real. The project also connects to a real problem in display design, how to make digital art feel fresh instead of ignored. You can learn human-centered testing, basic machine learning concepts, and experimental design without needing to invent a new model from scratch.

Research Questions

• How does an AI-generated daily image affect viewer dwell time compared with a fixed image rotation?
• What is the effect of weather-linked image changes on how often viewers stop to look at the frame?
• Does pairing sunrise color data with generative art change viewer ratings of calmness or mood?
• To what extent does indoor ambient sound influence how viewers describe the artwork after repeated exposure?
• Which display mode produces better short-term recall of the artwork, AI-generated daily art or fixed-rotation baseline art?
• How does novelty over 30 days affect engagement after the first viewing week?

Basic Materials

• E-ink display or e-paper frame with a small controller computer.
• Laptop for prompt writing, data logging, and analysis.
• Smartphone or tablet for photographing the display and recording response data.
• Weather source from a public service such as NOAA or a local weather station.
• Light sensor or color reference card for sunrise color capture.
• Microphone app or sound meter for indoor ambient sound measurement.
• Consent form and survey sheets for viewer feedback.
• Timer or visitor log for recording dwell time.
• Spreadsheet software for organizing daily data.
• Printed fixed-rotation baseline images.

Advanced Materials

• University-grade e-ink display panel with driver board.
• Small edge computer or microcomputer with enough storage for daily image generation.
• Local LLM access for prompt generation and image selection logic.
• Image generation pipeline using SD-Turbo or a similar diffusion model.
• Calibrated color sensor or spectrometer for sunrise color sampling.
• Sound level meter with exportable data.
• Computer vision camera for automated engagement tracking, if approved.
• Secure database for storing daily prompts, images, and viewer responses.
• Statistical software for mixed-effects analysis.
• Control artwork set matched for style and brightness.

Software & Tools

• Python: Handles data collection, image organization, and statistical analysis.
• ImageJ: Measures brightness, contrast, and color features in generated images.
• R: Runs engagement comparisons and mixed-effects models.
• Google Sheets: Tracks daily weather, sound, and viewer response data.
• PubMed: Finds peer-reviewed papers on visual attention, digital art, and human-computer interaction.

Experiment Steps

1. Define the one engagement measure you will trust most, such as dwell time, recall, or stop rate.
2. Choose your control condition, then match it to the AI display on brightness, size, and viewing setting.
3. Decide how daily inputs like weather, sunrise color, and sound will become image prompts or style rules.
4. Plan a repeatable viewer test schedule so each exposure happens under similar conditions.
5. Build a data table that links each daily image to its inputs, its prompt, and the response metrics.
6. Choose the analysis plan before you start, so you can compare trends across the full 30-day run.

Common Pitfalls

• Changing room lighting between sessions, which makes the display seem more or less engaging for the wrong reason.
• Letting the AI images vary in brightness or clutter too much, which confounds art style with visibility.
• Using casual visitor comments as your main data, which makes the results hard to compare across days.
• Forgetting to match the baseline rotation to the AI version in image quality, color range, or novelty level.
• Mixing up exposure frequency with interest, which happens when frequent viewers report less excitement simply because they have seen the frame before.

What Makes This Competitive

A stronger project would separate art novelty from real engagement with careful controls. You could test more than one engagement metric, then use a stronger statistical method than a simple average comparison. You could also compare different input sources, such as weather only versus weather plus sound, to see which context adds the most value. That kind of design shows real insight into both display technology and human behavior.

Project Variations

• Compare weather-conditioned art with art conditioned only on sunrise color to isolate which input matters most.
• Test whether viewers respond differently to abstract generative paintings versus landscape-like outputs on the same e-ink frame.
• Measure engagement in a classroom, lobby, or gallery-like hallway to see how location changes the effect.

Learn More

• NOAA Climate Data Online: Find local weather records and daily conditions for your display inputs, then search the NOAA site for station data.
• NASA Earth Observatory: Read about color, light, and environmental image data, then search the NASA site for articles on atmospheric color and visualization.
• NIH PubMed: Search for review articles on visual attention, novelty, and human-computer interaction.
• MIT OpenCourseWare: Look for free materials on machine learning, computer vision, and design methods.
• Journal of the Society for Information Display: Search the journal for papers on e-paper, display perception, and viewing behavior.