Interpreting Reality: A Visual Journey Through the Catwalk Property with AI

Danne Woo
5 min readMay 30, 2023

--

As I prepared to traverse the expansive Catwalk property, I wasn’t alone. Accompanying me was an unexpected ally: Mid Journey, a text-to-image generative AI platform. The venture wasn’t to frame traditional photos but to engage Mid Journey with narrative sketches of my surroundings, hoping it would rekindle the vistas that unfolded before me. My choice of Mid Journey over other platforms like OpenAI’s Dalle, Stable Diffusion, or Adobe’s Firefly was reinforced by comparative tests that proved Mid Journey’s superior realism in image generation.

The Tower View

My journey initiated from the tower, my residence and workspace during this period. The tower offers a breathtaking view of the Hudson River and the architectural marvel of the Rip Van Winkle bridge. My verbal portrait of the tranquil vastness of the river, the distant yet awe-inspiring presence of the bridge, and the high-altitude perspective from the tower provided raw material for Mid Journey. After several iterations, the platform approximated the magnificence of the scene remarkably well, embodying the sense of awe invoked from such a viewpoint.

Prompt: Generate an image showcasing a grand bridge that transitions from the foreground to the background. The bridge should be part of a mountainous landscape in the Hudson Valley, New York and go over a river. Tree line view. Trees in foreground. Early summer. Afternoon light. Rule of thirds. Photograph. ––aspect 7:4

The Catslair Fire Pit

Next, I moved to the fire pit at the Catslair house. My description fed the AI details of the extinguished fire pit, an echo of the 6-foot bonfire from the night before, now reduced to ash. The task was complex for Mid Journey as it was prompted to comprehend a fire pit without a fire. This challenge was overcome by leveraging the text weights function offered by Mid Journey. It allowed me to assign greater weight to certain keywords or phrases, and conversely, negative values meant reducing the emphasis on the respective word or phrase. Hence, applying “Fire::-0.5” ensured that the final image generated would be less likely to include any sign of fire. The image returned by Mid Journey successfully mirrored the serene tranquility of the scene.

Prompt: Lush green forest background. 4 mesh brown outdoor chairs, facing away surrounding an old metal camp fire pit that is not lit and filled with ash. No fire. Grassy area surrounded by trees. Pile of logs to the right next to a flower pot. Early summer. Mid day light. Photograph::1 Fire::-0.5 ––aspect 7:4

Hudson River from the Catslair

From there, I sought to capture a glimpse of the Hudson River seen through the trees by the Catslair house. I provided Mid Journey with a narrative of the silhouette of trees framing the view and the serene aura of the scene. It required several attempts for the AI to accurately capture the unique interplay of wilderness and tranquility that makes this view so captivating.

Prompt: Leaves, branches and bushes in the foreground, framing a river and a hill covered in trees, clear blue sky, mid day light, early summer, photograph
––aspect 7:4

The Lonely Meadow Tree

The fourth setting took us to a solitary tree standing regally in the meadow between the main house and the Catslair house. My description focused on the tree’s solitary elegance, its wild surroundings, and somewhat sparse canopy. Mid Journey, after numerous iterations, rendered an image that exuded the quiet strength symbolized by this lone tree.

Prompt: A singular tree alone in the center of the frame standing in the middle of chest high grass and yellow flowers, forested in the distant background, early summer, afternoon light, symmetrical, sparse leaves, butterflies, photograph
––aspect 7:4

Hammock Skyward View

After a considerable walk, I found solace in a hammock, my eyes focused upward. I described the setting, the uncommon perspective, and the sunlight’s dance through the leaves. Mid Journey’s iterations brought about an unexpectedly delightful output — not merely a rendition of the view but a translation of the trees’ grandeur.

Prompt: From a worms perspective in upstate New York, looking up at two large pine trees, clear blue sky, mid afternoon, sun shining down on you, photograph ––aspect 7:4

Garden Shed from the Hammock

Lastly, from my resting spot in the hammock, I directed my attention to the garden shed to my right. I furnished Mid Journey’s neural network with the quaint charm of the structure and its rustic appeal. After several modifications, Mid Journey skillfully approximated the understated beauty of the shed.

Prompt: A garden shed in the distance, light green siding, horizontal siding, lights on inside, hose, lattice, vines, flower pots, in the middle of the woods, wooded, french doors open, grass, dandy lions, early summer, mid afternoon light, photograph ––aspect 7:4

As the exploration concluded, I was awestruck by how accurately Mid Journey had translated my verbal descriptions into vivid visuals. Although it was quite a learning lesson on how to best write the prompts and use the settings Mid Journey provides it was quite impressive with its ability to understand my prompts. It was a testament to the power of AI, not just in terms of technical prowess, but as a partner in creative exploration. This exercise, at its core, proved that AI has the potential to augment our experiences, enabling us to reimagine reality and share it in ways previously unimagined.

--

--

Danne Woo
Danne Woo

Written by Danne Woo

Founder of @datavisualinfo, Professor at @QC_news, @meddemfund/@fordfoundation Fellow at @colorofchange and @itp_nyu alum. #datadork #designer #programmer

No responses yet