AI Visualization for Craft Businesses: Show projects before they are built
Customers want to see what they are getting. Whether it’s a pool in the garden, a skylight in an old building, or a PV system on a terraced house, anyone who can visually represent a project sells better. With AI-supported image editing, craft businesses can create realistic project visualizations from customer photos. The tool for this is called Nano Banana.
What is Nano Banana?
Nano Banana is Google’s image generation and editing model within the Gemini app. It was introduced in September 2025 and supplemented in November 2025 by Nano Banana Pro. The special feature: Unlike conventional image generators, Nano Banana understands the context of an image. It recognizes objects, perspectives and spatial relationships.
For craft businesses, this means: An uploaded photo of a customer’s garden can be specifically edited. Elements can be added, removed or changed, while the rest of the image is preserved.
The basic version is available free of charge in the Gemini app. For more extensive use, there are paid subscriptions with higher quotas.
A practical example: From lawn to three design variants
A typical scenario from consulting practice. The starting image shows a German residential garden: simple lawn, privet hedge as privacy screen, covered garden furniture on the terrace. Functional, but without a clear design idea.
With Nano Banana, three different visualizations were created from this one photo, each with its own character:
Variant 1: Pool with pergola A rectangular pool with dark lining, surrounded by large-format light concrete slabs. A modern slatted pergola in front of the house creates a covered terrace area. The planting focuses on ornamental grasses and compact shrubs along the property line.
Variant 2: Wooden deck with lounge area No pool, instead a spacious wooden deck with lounge furniture. A curved gravel path leads through the garden, past raised beds and perennial plantings. The focus is on quality of stay and natural design.
Variant 3: Pergola with manicured lawn A slatted pergola covers the dining area with wooden furniture. The lawn is preserved, but is structured by clear lawn edges and ornamental grass beds. A restrained variant for customers who do not want a complete redesign.
All three variants show the same house, the same perspective, the same neighboring buildings. Only the garden design differs. This is exactly what makes the value for the customer discussion: Instead of abstract descriptions, the customer sees concrete alternatives and can make a more informed decision.
Nano Banana alone is not enough
Here is the crucial point: Nano Banana is a tool for image editing, not a planning tool. For a convincing project visualization, the interaction of several AI solutions is needed.
Step 1: Analysis and concept Before an image is edited, it must be clear what is to be shown. Language models such as ChatGPT or Claude can help here. They can develop concrete design ideas from customer requests, suggest materials or describe styles.
Step 2: Prompt creation Nano Banana works with text instructions. The more precise the description, the better the result. A prompt like “Add a pool” delivers different results than “Add a rectangular pool with anthracite-colored lining, surrounded by light gray concrete slabs in 60x60cm format, with a wooden deck made of thermo ash as a transition to the terrace.”
Step 3: Iterative editing The first result is rarely correct. Nano Banana makes it possible to refine in several passes: Adjust lighting, correct perspective, add details. This process requires experience and an understanding of how the model reacts to instructions.
Step 4: Post-processing For professional results, classic image editing can be useful. This includes color corrections, sharpening and the insertion of logos and watermarks.
Use cases for different trades
Gardening and landscaping Pool integration, terrace design, planting concepts, lighting planning, privacy solutions. The example of the bluegardens AI solution for garden and pool planning shows what an industry-specific implementation with its own prompt bridge can look like.
Roofer Visualization of roof windows, simulation of PV systems on the existing roof, representation of different roof coverings
Carpenter Carports, conservatories, patio roofs in the context of the existing building
Painter and facade builder Color concepts for facades, before-and-after representations for renovations
Carpenter and joiner Visualize furniture in the room, show built-in cabinets in the customer context
What Nano Banana can and cannot do
Strengths:
- Targeted editing of individual image areas while preserving the rest
- Understanding of spatial relationships and perspectives
- Combination of multiple images into one scene (up to 14 reference images with Nano Banana Pro)
- Text display directly in the image (for labels or dimensions)
- Adjustment of lighting and time of day
Limits:
- No technically exact plans or scales
- Sometimes difficulties with fine details or small faces
- Results are visualizations, not construction drawings
- Unnatural transitions can occur in complex scenes
- The model occasionally makes mistakes with proportions
Important: All images created by Nano Banana carry an invisible SynthID watermark. This can be used to prove that an image was generated by AI.
The workflow in practice
- Take initial photo: A good photo of the current state is the basis. Pay attention to sufficient lighting and a perspective that shows the project well.
- Develop concept: Clarify with the customer what should be shown. If necessary, use a language model to structure ideas.
- Formulate prompt: Describe precisely what should be changed. Details on materials, colors and style improve the result.
- Edit in Gemini: Upload the photo and enter the prompt. Select “Create images” in the Tools menu.
- Check and refine result: Do perspective, proportions and style match? If necessary, adjust the prompt or edit specific areas.
- Document: Save intermediate steps. This allows you to repeat the process for similar projects.
Costs and availability
Nano Banana is available in the free version of the Gemini app. The daily quota is limited. For regular use, Google offers various subscriptions:
- Google AI Plus: Extended quota
- Google AI Pro: Higher quota, access to Nano Banana Pro
- Google AI Ultra: Highest quota, all functions
For developers and companies, there is access via the Gemini API and Vertex AI.
A tool in the toolbox
Nano Banana is not a substitute for expertise or planning skills. It is a tool that helps craft businesses make their ideas visible. The best results are achieved when different AI solutions work together: language models for concept and prompt, image generators for visualization, classic software for fine-tuning.
The entry is low-threshold. A Google account and the Gemini app are sufficient for initial attempts. Good results require practice, but for companies that regularly visualize projects, the familiarization is worthwhile: not as a replacement for planning tools, but as a supplement for the customer discussion.
Frequently asked questions
How long does a visualization take? Under optimal conditions, a draft can be created in under a minute. Realistically, it usually takes three to five minutes, depending on the motif, level of detail and number of correction loops.
What does the use cost? The basic version is free. For paid subscriptions, the costs are about USD 0.04 per generated image. The exact prices vary depending on the package and region.
How realistic do materials look? Standard materials such as wood, natural stone or water are reproduced very accurately. For special products or brand-specific surfaces, post-processing may be useful.
Who owns the rights to AI-generated images? Purely AI-generated images do not have automatic copyright protection in most jurisdictions. As soon as a person intervenes creatively, a protectable work can arise. The use is harmless, provided images are marked as AI visualization.
Can I create multiple variants of a project? Yes. Nano Banana is particularly strong in consistency over several editing steps. Shapes, colors and perspectives remain recognizable, even if materials or details are varied.
Where are the limits? Fine reflections, complex textures or very small details may differ. The model generates visualizations, not technical drawings. For building applications, statics or dimensions, specialist plans are still required.
Post Tags :
Share :
AI visualization, Nano Banana, Gemini image editing, craft digitization, project visualization, AI image generation, landscaping AI, Google Gemini, AI for craftsmen, customer consulting