Living in Switzerland, I cherish time spent in nature. The majestic Alps, winding rivers, and serene lakes are perfect places for me to find rest and inspiration. I wanted to express my gratitude for the many moments I’ve spent so far immersed in the beauty of the mountains with a data art piece that combines my greatest passions: photography, creative coding, and AI.
A moment of wonder: Photography of the Matterhorn
The piece’s starting point was my photograph of the iconic Swiss mountain, the Matterhorn, viewed from Schwarzsee above Zermatt, captured in late autumn (Fig. 1). That day, the mountain stood silent under the early snow, bathed in sunlight. The stillness, the softness of the snow, and the power of the alpine peaks came together in a moment of awe and peace, a memory deeply cherished.
A dialogue with the machine: Diffusive reinterpretation
What draws me to machine learning is a fascination with how simple building blocks, based on mathematical concepts, can be assembled into increasingly complex architectures capable of learning and mastering intricate tasks. I’m especially captivated by the elegant mechanics of generative AI, particularly diffusion models. These models, when used in text-to-image mode, begin with pure noise, and guided by a prompt, gradually transform it into coherent imagery through a step-by-step denoising process. It feels like a mathematical meditation on emergence: a journey from randomness to form.
In my project, however, I explored text-guided image-to-image mode, steering the generation using both the photograph and a simple text prompt:
“Artistic wavy version with subtle pink and violet tones.”
The model began by adding noise to the photograph and then iteratively removed it until a reimagined vision emerged (Fig. 2). To make this transformation feel alive, I also captured the denoising steps and wove 30 interpolated frames between them, revealing a seamless, almost dreamlike unfolding from the original photograph to its AI-crafted interpretation (Fig. 3).
A celebration of the nature: Diffusive Alpine Metamorphosis
Inspired by the concept of diffusing particles from my physics lectures, I sought to express the breathtaking beauty of the mountain and its artistic reinterpretation through a joyful, dynamic spectacle. This led me to code an animation where each image disintegrates into spheres, which subtly recolor during diffusion and regather to form the subsequent image in the denoising sequence. The spheres become messengers of transition: first chaotic and dynamic, then gradually converging into defined forms. Later, I added music to amplify the light and vibrant atmosphere supporting the piece’s emotional core. Figure 4 presents the resulting animation—I warmly invite you to witness the transformation as it unfolds.
A harmony: Nature, machine’s ‘imagination’ and human creativity
At its core, this piece began with a single photograph—a data point constituting a structured array of color values, capturing a fleeting alpine moment. Through the diffusion model, this data underwent a transformation: noise was algorithmically added and then removed, guided by the prompt, resulting in new visual interpretations rooted in probabilistic learning. The final piece emerged as a collaboration between nature, machine, and human creativity; bringing together the serenity of the original landscape, the generative potential of machine learning, and a personal creative vision expressed through photography, prompt design, code, and music.
As generative AI becomes more intertwined with data analysis and visualization workflows, it’s reshaping the boundaries between factual and creative representations. We’re already seeing AI generate basic charts from datasets, enhance visual metaphors, suggest stylistic treatments for dashboards, and assist in code generation for custom visualizations. In this broader context, projects like this one explore how to reveal the internal workings of a model, its interpretation of data, and the layered steps of its computational imagination. This expansion of the designer’s palette, through AI-augmented visual storytelling, offers new ways to make meaning, provoke curiosity, and connect audiences to the stories behind the numbers.
The “Diffusive Alpine Metamorphosis” piece is part of my broader journey at Data Immersion, where I create data-driven artworks that honor nature’s beauty and life’s vibrancy. I warmly invite you to visit the Data Immersion blog to explore more work inspired by these themes.
For an extended, more technical article about this data art piece, please visit Generative AI for Data Art & Visualization: Diffusion Model.
Sylwia Nowakowska
Sylwia Nowakowska is a physicist with a Ph.D. in Nanoscience. She has more than 12 years of experience in Research & Development, spanning Physics, Material Science, and Artificial Intelligence. Sylwia has always found joy crafting aesthetic data visualizations, whether for summarizing experiments, presentations, or academic publications. She finds it incredibly satisfying to see complex information become clear and accessible, meaningful, and beautifully represented. This passion led her to found Data Immersion, a platform where she shares her enthusiasm for Data Art & Visualization. When she’s not immersed in data, you can find her immersed in water, enjoying swimming, or in the beauty of Swiss mountains, which she captures through her lens.
