AI visualization in "30 seconds":
How bluegardens is harnessing Google's nano-banana for gardening and landscaping
When doubts become visualizations
A pool for 75,000 euros, but the customer can hardly imagine the result. Three clicks later, he sees a plausible, photorealistic visualization. In many cases, this is now possible within 30 seconds under good conditions and with an optimized workflow.
At bluegardens, we constantly monitor new AI technologies, test them systematically and only adopt those that bring real added value to horticulture and pool construction. After intensive testing, we have adopted Google’s “nano-banana” (
What is "nano-banana" / Gemini 2.5 Flash Image?
Google officially refers to the model as Gemini 2.5 Flash Image internally also nano-banana internally. It was launched in August 2025 and is available via the Gemini API , Google AI Studio and Vertex AI available(developers.googleblog.com).
Functions according to Google
Multi-Image Fusion: Combination of several image sources – e.g. an existing garden photo and a separate pool shot to create a new scene.
Consistency across processing steps: Objects or materials retain their shape and position across multiple variations.
Targeted transformations: Changes can of course be instructed (“add wooden deck”, “change lighting mood”).
Semantic knowledge of the world: The model understands real materials, proportions and lighting, which makes images appear more plausible.
SynthID watermark: All created or edited images contain an invisible watermark that identifies them as AI images.
The bluegardens integration: AI solution for landscaping
Universal AI models such as Gemini 2.5 are powerful, but not tailored to the technical language and material systems used in landscaping.
Our prompt bridge translates terms, product data and planning information into structured prompts.
How it works
Input: “carropools overflow pool”.
Automatic completion by bluegardens AI: dimensions, materials, light, shadow, surroundings.
Transfer to Gemini 2.5 / nano-banana.
Quality control (manual or automatic).
Our database with pool and GaLaBau terms and product libraries (covers, stones, foils, etc.) ensures that terms are interpreted correctly. A human-in-the-loop test guarantees quality and legal certainty.
From garden to vision
On the left, the initial state: a simple terrace, functional but uninspiring.
On the right, the result: a modern garden with a Carropools overflow pool, light-colored natural stones, structured planting and an elegant pergola.
A family of builders planned precisely this conversion and used the AI solution from bluegardens for it.
Example: Carropools overflow pool with natural stone terrace
The family wanted a pool with a calm water level, flush with the new terrace made of light-colored limestone. The overflow was to be circumferential and the expansion tank was to be concealed.
This is how the visualization works:
The description is entered into the bluegardens AI solution:
“Carropools overflow pool 6.00 × 3.00 metres, flush with the edge of the terrace, light natural stone, afternoon light, view of the tree.”
The AI automatically adds technical and visual details, from the gutter geometry, mirror height and light reflections to the shadows cast, and transfers the data to Google’s Gemini 2.5.
Photorealistic variants with precise joints and edge guidance are created within a few minutes. After a brief visual inspection, the designs can be used directly in consultations or for decision-making.
The result:
Realistic visualizations that help customers see their ideas before the first sod is turned and significantly speed up planning discussions.
bluegardens shows how modern AI technology makes garden and pool construction more tangible, efficient and inspiring.
Law and transparency
All images created with Gemini contain an invisible SynthID watermark from Google. For use in customer projects, a visible note such as “Visualization with AI, for orientation purposes” is recommended.
KI images are intended as a creative supplement, not as a substitute for technical drawings, building applications or structural verifications. The responsibility remains with the specialist planners: each illustration should be carefully checked before implementation.
Vision 2026: bluegarden’s garden planner with AI
bluegardens is working on the next generation of digital planning tools. In future, the garden planner will be directly linked to Gemini 2.5. Dimensions, materials and costs will be transferred automatically. An AR function will make it possible to experience the planned pool to scale in your own garden. AI-supported video renderings are also in preparation.
The goal is clear: to combine planning, visualization and calculation in one integrated system – precise, clear and practical.
Visualization as a sales advantage
The bluegardens AI solution in conjunction with Google’s Gemini 2.5 opens up new possibilities for garden and pool planning. Photorealistic images are created within minutes, bringing designs to life and speeding up decision-making processes.
Customers can see at a glance what they can expect and make decisions much more quickly.
bluegardens ensures that new technologies do not remain in the laboratory, but have a real impact in everyday pool and landscaping work.
1. is photorealistic visualization really that fast?
Under optimal conditions, a design can be created in around 30 seconds. Realistically, it usually takes three to five minutes – depending on the subject and level of detail.
2 How real are the materials and lighting moods?
Standard materials such as wood, stone or water are reproduced very accurately. A small amount of post-processing may be necessary for special surfaces.
3. what does a visualization with Gemini 2.5 cost?
On average around four cents per picture – depending on the package and the current exchange rate.
4 Who owns the rights to an AI image?
Images generated purely by AI have no automatic copyright protection. However, as soon as a human intervenes creatively, a protectable work can be created.
5. is the use of such images legally unobjectionable?
Yes, as long as they are marked as AI visualizations and do not claim to be technical drawings.
6. can the AI also create variants of a project?
Yes, and even with a high degree of recognizability. Multi Image Fusion ensures that shapes, colors and perspectives remain consistent.
7 Where are the limits of the system?
The finest reflections or complex textures may still deviate slightly. However, Google is continuously improving the model.
8. how realistic is a Carropools overflow pool?
Visually very accurate – with a harmonious water level, edges and channels. Only technical components such as the expansion tank or hydraulics appear schematic.