Skip to main content
Enterprise AI Analysis: Artworks Reimagined: Exploring Human-AI Co-Creation through Body Prompting

Enterprise AI Analysis

Artworks Reimagined: Exploring Human-AI Co-Creation through Body Prompting

Jonas Oppenlaender, University of Oulu, Finland

Hannah Johnston, Carleton University, Canada

Johanna Maria Silvennoinen, University of Jyvaskyla, Finland

Helena Barranha, University of Lisbon, Portugal and NOVA University Lisbon, Portugal

Executive Impact & Key Findings

This study introduces body prompting as an innovative human-AI co-creation method for generative art installations, offering a more engaging and accessible alternative to traditional text-based inputs in public settings.

0 Total Images Generated
0 Avg. Generation Time
0 High User Pleasantness
0 Participant-Driven Narrative Shifts

Deep Analysis & Enterprise Applications

Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.

Challenges & Goals
System Architecture
User Experience & Strategies
Ethics & Recommendations

Mapping Challenges to Design Goals & Technical Solutions

The study identified key challenges in text-to-image generation for public settings and formulated design goals addressed by specific technical implementations, ensuring a more engaging and accessible experience.

Challenge Design Goal Technical Implementation
C1 (Output control): Limited by language. D1 (Enhance user control): Body movements guide image creation.
  • ControlNet translates body poses.
  • OpenPose extracts skeletal key points.
  • Style transfer & automated prompt refinement.
C2 (Literacy): Requires English literacy & prompt engineering expertise. D2 (Reduce language dependency): Body prompting removes need for textual skills.
  • Webcam captures user photos.
  • System detects poses.
  • CLIP-Interrogator generates text prompts from artworks.
  • Manual negative prompts improve output quality.
C3 (Collaboration & social creativity): Primarily solitary creation. D3 (Facilitate social creativity & collaboration): Encourages group participation.
  • Separate applications for prompting & viewing.
  • Separate display prevents blocking.
  • 60-second reset maintains flow.
  • Privacy screens on laptops offer alternative viewing.
C4 (Expressivity): Static, non-intuitive input in public settings. D4 (Support dynamic & expressive interactions): Real-time feedback creates engaging experience.
  • Real-time camera feed enables precise pose capture.
  • Linear application flow guides users.
  • Real-ESRGAN upscales images for public display.
  • Interface displays participant poses alongside images.
  • Random artwork shuffling ensures variety.
Implicit (Interaction Context): Public performance inhibition. D5 (Accommodate varied interaction contexts): Public & private booth options.
  • Privacy screens on laptops offer alternative viewing options.

Interactive User Journey & Core AI Pipeline

The Artworks Reimagined installation guides participants through a structured yet intuitive process, leveraging advanced AI models to transform traditional art through body prompts. This flow ensures a seamless co-creative experience from consent to outcome viewing.

Enterprise Process Flow: User Journey

1. Start & Consent
2. Select Artwork
3. Pose (Body Prompt)
4. End Screen
5. Interview & View Outcome

Technical Pipeline for Body Prompting

The system's backend utilizes a sophisticated architecture to interpret user poses and generate reimagined artworks:

  • Pose Detection (OpenPose): Captures user body poses from a webcam feed and extracts key points to form a skeletal representation, ensuring anonymity by not capturing facial expressions.
  • Image Generation (T2I-Adapter with Stable Diffusion 1.5 & ControlNet): The skeletal pose, along with style information extracted from the selected source artwork and an automated text prompt (from CLIP-Interrogator), are fed into the T2I-Adapter. ControlNet ensures the generated image aligns with the user's pose.
  • Automated Prompting (CLIP-Interrogator): Analyzes the source artwork to generate a descriptive text prompt, guiding the AI on style and content.
  • Negative Prompting: A manually designed negative prompt prevents undesirable generative glitches and ensures high-quality, safe outputs.
  • Upscaling (Real-ESRGAN & GFPGAN): Generated images are upscaled for high-resolution public display, with GFPGAN specifically enhancing faces.
  • Cloud Infrastructure (AWS & Replicate.com): The system is deployed on AWS S3 buckets for web applications and leverages Replicate.com's API for running machine learning models, ensuring scalability and robust performance.

Participant Engagement & Creative Strategies

Body prompting was overwhelmingly perceived as a highly engaging and fun experience, allowing participants to explore creative expression through physical interaction.

0 Avg. Pleasantness Score
0 AI-Introduced Surprises (Hallucinations)

Identified Posing Strategies

Participants employed three distinct strategies when engaging with the body prompting system:

  • Re-creation (Imitation): Approximately 30% of participants aimed to mimic the source artwork's pose, seeking harmony and a close resemblance. Motivations included adapting to the artwork's mood or simply wanting to replicate what they saw.
  • Reimagination (Reinterpretation): Another third of participants sought to contrast the source artwork, creating a novel image that significantly diverged from the original. This often stemmed from a desire to experiment, create funny or weird pictures, or simply "create a different atmosphere."
  • Casual Interaction (No Specific Reason): The remaining third of participants posed without deep thought, often going with intuition or assuming a "normal" (neutral) pose. Curiosity about the AI's reaction to their pose was a common driver here.

Despite minor discomforts related to the countdown timer or being photographed in public, the overall experience was positive, fostering creative exploration and playful interactions.

Ethical Considerations and Design Recommendations

Implementing generative AI in public settings requires careful consideration of ethical implications and user experience. The study provides insights and recommendations for practitioners and researchers.

Ethical Considerations in Public AI Installations:

  • Consent and Data Privacy: Implemented anonymization protocols, storing only detected poses, not original photos. Transparent policies and secure data handling are crucial.
  • Body Shaming and Self-esteem: Focused on pose detection (OpenPose) rather than raw images to avoid negative self-perception, presenting results neutrally.
  • Emotional and Psychological Impact: A debriefing process was in place to address unexpected emotional reactions, particularly for children who expected their likeness in the generated images.

Recommendations for Future Body Prompting Systems:

  • Multi-display setup & User Experience: Provide immediate feedback on the same screen, avoid bottlenecks, and clearly explain public/private options.
  • Manage Expectations: Clarify the installation's purpose (e.g., reinterpreting art vs. "camera booth"). Include facial expressions and hand gestures for more image control.
  • Enable Expressiveness: Integrate user controls for mood/style alongside body prompts (e.g., sliders, knobs) to enhance user control over outcomes.
  • Enhancing AI Co-creativity: Embrace hallucinations and surprising AI artifacts playfully, balancing user control with AI agency to foster collaboration.
  • Interaction with Installation: Address "first-click problem" with clear, visual instructions, especially for private booths.
  • Takeaways & Souvenirs: Offer means for visitors to take home their generated images (e.g., QR codes, print-outs) to extend social sharing.
  • Nudity & Inappropriate Imagery: Implement strong guardrails (e.g., careful artwork selection, negative prompts) to prevent unsafe generations.
  • Materials: Provide diverse source artworks (figurative, abstract, varying complexity) that relate to the human figure and can accommodate different posing strategies and group sizes.
  • Personality Traits: Public staging is generally preferred over private booths for its social and performative aspects in event settings.

Calculate Your Potential AI Impact

Estimate the tangible benefits of integrating advanced AI solutions into your enterprise operations. This calculator provides a preliminary ROI based on industry benchmarks and operational parameters.

Estimated Annual Savings $0
Annual Hours Reclaimed 0

Your AI Implementation Roadmap

A strategic phased approach ensures successful integration and maximum impact of generative AI within your organization. We guide you through every step, from initial assessment to ongoing optimization.

Discovery & Strategy

Comprehensive assessment of current workflows, identifying high-impact AI opportunities and defining a tailored implementation strategy.

Pilot & Prototyping

Develop and test initial AI models on a smaller scale, gathering feedback and refining the solution for optimal performance and user acceptance.

Full-Scale Integration

Seamless deployment of the AI solution across relevant departments, ensuring robust infrastructure and comprehensive training for your team.

Optimization & Scaling

Continuous monitoring, performance tuning, and expansion of AI capabilities to new areas, maximizing long-term ROI and competitive advantage.

Ready to Reimagine Your Enterprise with AI?

Unlock the full potential of human-AI co-creation. Schedule a free, no-obligation strategy session with our experts to explore how these insights can transform your business.

Ready to Get Started?

Book Your Free Consultation.

Let's Discuss Your AI Strategy!

Lets Discuss Your Needs


AI Consultation Booking