Wood Wide Web: Learning to Personify Forests

The Wood Wide Web is a hybrid interactive installation, which will be bringing ancient endangered trees from India and the UK to life through the use of skeletal tracking and AI. The sacred forests get personified and tell their stories, evoking empathy in humans across the globe to inspire more care for the planet. 

Artists: Kanchan Joneja, Kristina Pulejkova
With support from Anupam Mahajan & Cameron Naylor

Introduction: The Seed

In a world where we have moved much beyond the world wide web era by stretching the limits of technology through AI and more, we have also pushed the boundaries of our planet and its fragile ecosystems. In such a context of interconnectedness — both digital and physical — human activity shows up unannounced and has an adverse environmental impact. What if we could tap into the secret conversations of the trees? What if we could be our nature? What if… ? 

Wood Wide Web is a term coined by Dr. Suzanne Simard, a forest ecologist from the University of British Columbia, who describes the relationships between vegetal life via an underground, mycelium network. Our project Wood Wide Web intends to be an act of embodiment and empathy, an effort to make visible and communicate the unseen troubles and trials of endangered tree ecosystems to humans. These elder giants are the ambassadors of their own stories, as the human becomes the tree; making us embody nature. This closeness can help us understand the way nature communicates and feel what they feel, with the hope to illicit an environment of care.

What if we could tap into the secret conversations of the trees?

Sensitivity towards nature is what brought the team together to work on this project. Anupam, Cameron, Kanchan and Kristina are creative practitioners who have tinkered with plants, technology, art and storytelling in their own unique ways. While Anupam enjoys experimenting with interaction design and cares deeply about biodiverse ecosystems; Cameron has generated electrical signals and sounds from plants; Kanchan loves to build narratives and experiences that engage people to care about the planet; and Kristina explores how the use of technology might lead to more significant forms of sustainability in human-nature relationships. 

Initial brainstorming session before the pitch

The Process

In order to translate the concept into reality, the team started working on a few aspects, to begin with:

1. Data collection from sacred forests and woodlands in India and the UK from secondary sources and from the field. We sought the help of local partner organizations (Farmers for Forests, and Kew Gardens) in the form of text, interviews, images, sounds and any stories or folklore that bring out the cultural and environmental significance of these species within larger ecosystems and communities 
2. Defining the human experience which could then help guide the character design, the flow of interaction, and the nature of outputs to be generated from the AI models (especially the tone and messaging) 
3. Visualization of the space and curation of visual references of the characters (virtual 3D models) according to the feelings that Wood Wide Web intends to evoke
4. Generating sample texts using available data from the AI model, and creating voiceovers allowing each tree to speak its story.  

Since the ancient trees (along with animals and birds) in India and UK are extremely old and native, these are personified as ‘elder giants’ in our artwork, bestowing wisdom upon the groves with which they live. When personified, these take the form of aged-looking trees that move, talk and tell stories of their erasure, and mourn the loss of their existence. We intend for this to evoke a feeling of sorrow in humans and a sense of interconnectedness with nature through empathy. 

Resources for textual data for ancient trees, India and UK

AI-Generated Text Tests

We created two models using data from UK ancient woodlands, and Indian sacred groves. The datasets consist of Indian and British folktales and mythology surrounding ancient trees, testimonies from people working on the ground and British and Indian poetry and storytelling concerning specific tree types. We will be using the generated text as a script for the narration of ancient trees. Below is an example from the UK trees narrative. 

Example of AI-generated tests using the GPT2 ML model 

When personified, these take the form of aged-looking trees that move, talk and tell stories of their erasure, and mourn the loss of their existence. We intend for this to evoke a feeling of sorrow in humans and a sense of interconnectedness with nature through empathy. 

We are currently underway in generating the script for the characters of trees. As soon as we finalise appropriate texts for the characters, we will proceed to create voiceovers for each character, using Vocaloid and other AI-generated voice types. 

In terms of the AI-generated textures for the characters, we are currently in the process of collating data from various digital and physical resources. The dataset will consist of treebark samples from ancient and sacred trees from across India and the UK. For trees from the UK, we are gathering samples from oak and yew trees, and are currently in the process of identifying representatives for the India trees. Kristina will be visiting Kew Gardens in London to photograph treebark samples from Kew’s Economic Botany Collection, specifically the barks of oak and yew. We are in touch with Farmers for Forests in India to get some samples and indications on what the best candidates might be for the Indian ancient trees. The next step in this process is to collate the data and create images using the StyleGAN2 Machine Learning model, which will then be used to make latent videos, to create an ever-shifting, dynamic texture for our tree characters.

Visitor Interaction & Flow  

We expect each visitor to interact for an average of 1 minute, which is of course variable depending on the visitor’s interest. The flow is envisioned as follows: 

5s-10s | Scene 1: Enters to a screen with tree sleeping/ dormant – wakes up as soon as a visitor in front

10s-30s | Scene 2: Visitor moves hands and body – screen gets brighter, lights up, other Non-Player Characters (NPC) wake up

30-50s | Scene 3: Voiceover ensues with AI-generated audio playing 

50-60s | Scene 4: Visitor moves away, everything dims down, and dies again; visitor leaves feeling sorrow and empathy for the trees.

We studied what spectrum of emotions we aim to induce in the audience through the installation and curated the experience around it. The installation aims to produce sorrow regarding the ongoing loss of biodiversity, empathy through the personified ancient tree characters facing extinction, and curiosity to learn more about the visuals and sounds they experienced as part of the installation.

Character Designs:
We studied different references to shortlist a few interesting types for our character designs. This helped us prepare a design brief for 3D artists. 

Site Visualisation:

We studied the venue options and created a rough layout of the interaction space. The projection screen will house the main centrepiece that will reflect the movement of the user, along with some NPC tree characters. 

The installation aims to produce sorrow regarding the ongoing loss of biodiversity, empathy through the personified ancient tree characters facing extinction, and curiosity to learn more about the visuals and sounds they experienced as part of the installation.

Soundscape:

The soundscape element serves as an aid in defining an audible context of the forest and its ecosystem. Sound, as an empathetic listening process, is therefore used to emphasize the themes of curiosity in the listener, through an unseen landscape beyond the boundaries of the projection.

Through archival audio, as well as on-site recording, the soundscape serves as a sonic collage of endangered animals, and the habitat in which they reside. The song of the Forest Owlet, Song Thrush, and the now-extinct Kaua’i bird punctuate the soundscape, highlighting the sounds that are presently under extreme threat and will not survive if we continue to eradicate their habitat.

The backdrop, a shifting landscape of sacred groves, and the whistling and rustling of endangered trees serve as a sonic backdrop for the installation, helping in personifying the scene, and allowing audience members to re-engage with nature

Song Thrush Birdsong Kauai ‘O’o 

Sacred Groves of India: https://youtu.be/DuKSUIt3K1A

Mycelium Video Effects: 

Mycelium are incredibly tiny “threads” of the greater fungal organism that wrap around or bore into tree roots. Taken together, mycelium composes what’s called a ‘mycorrhizal network’, which connects individual plants and trees together to transfer water, nitrogen, carbon and other minerals.

The installation will have a visualisation of a virtual mycelium network when the centrepiece of the artwork is activated. We explored different game video effects options to incorporate the visualisation of the network in the virtual space.

Mycelial connection render in Unity

CREDITS

Created and Designed By: Kanchan Joneja, Anupam Mahajan, Kristina Pulejkova & Cameron Naylor

3D development and interaction Design: Anupam Mahajan

VFX: Apratim Verma

Text Generation: Kristina Pulejkova & Kanchan Joneja

Narrative building and user experience: Kanchan Joneja

Programming: KristinaSound Design: Cameron Naylor