Shadowplay is an interactive and immersive visual installation that has been exhibited at venues including Shambala Festival, Light Up Lancaster, Windermere Science Festival, Supermassive, and EMF Camp.
At the heart of Shadowplay is an AI image generator that creates a stream of moving images. Audiences interact with and control the stream using their bodies, light and shadows.
Is Shadowplay a research, an artwork, or something else?
Shadowplay is a series of artworks and experiences that evolves to fit different settings, from parties and festivals, to galleries or educational events.
We are continually building new collaborations and experimenting with Shadowplay in many different contexts across music, theatre, wellbeing, and education. We love collaborations! If you have an idea then please get in touch!
Shadowplay is also a Design Research project that has three areas of exploration:
- The creative possibilities of realtime AI image generation
- The challenges and opportunities that generative AI creates for interaction design
- The applications of realtime AI image generation
How does it work and what is PromptTank?
Shadowplay brings together several pieces of technology, which we explain in detail in this research paper.
At its heart we have a bespoke user interface called PromptTank, a flexible and intuitive way to interact with generative AI that adopts an approach we have coined Holistic Prompt Craft.
PromptTank has been built and refined over the course of many different installations, experiments and collaborations with creative practitioners. An earlier UI we created, PromptJ, followed the metaphor of a DJ mixing desk with crossfading between prompt ‘channels’. When using this early system at Light Up Lancaster and in collaboration with Klanghaus and with Supermassive, we found that adjusting pixels in the video being fed into the model gave as much control as text prompting.
Partly inspired by this insight, PromptTank was built as a flexible interface which allows the user to decide which controls they want to use over both the image and text prompts. A collaboration with Theodore Koterwas led us to add a ‘snapshot’ system for storing prompts for a live performance. An extensive collaboration with Loura Whitham in preparation for our Shambala 2025 and 2026 installations led to precise control over textures in the image prompt, plus many developments and enhancements to the PromptTank UI. Collaboration with Keith Saha and the Portraits of Dementia project led us to integrate live voice control over text prompts for theatrical performance.
Right now PromptTank is not publicly available, but if you’re interested in using it, developing it further, or want to know more, then please get in touch!
Image of PromptTank in use
Who is behind Shadowplay and how do we get in touch?
Shadowplay is led by Dr Joseph Lindley and Dr Roger Whitham. If you’d like to contact us to discuss Shadowplay, then the best way is via email using:
j.lindley@lancaster.ac.uk and r.whitham@lancaster.ac.uk
We would like to acknowledge and thank the many people who have collaborated with us over the past two years, fed into the development of PromptTank and put it to use in events and activities. These include Theodore Koterwas, Keith Saha, Jenni MacRitchie, George Harris, Julie Brown, Elsie Ledger, Marc Barto, Klanghaus and Supermassive amongst many others. Special thanks to Loura Whitham for her ongoing collaboration with us to design and refine the PromptTank UI, and for originating the Textures system within PromptTank.
We are very grateful for extensive advice and expert support from members of the Lancaster University School of Arts technical team: Simon Attwood, Mark Bentele, Alex Flahive and Beth Pete.
The work is funded by Design Research Works, which in turn is a UKRI Future Leaders Fellowship.