Latest news
D5 Render: sky, HDRI and weather settings – step-by-step guideChaos Vantage Scene States: Managing Day-Night and Material Variations from a Single SceneHow to enable realistic caustics in Corona Renderer and D5 Render: a guideArchitectural render videos with Runway Gen-4: a camera movement prompt guideSpeeding Up Presentation Prep with Enscape Visual Settings PresetsBuilding your own render workflow with Veras 5.3 NODES: a step-by-step guideHow to Generate Music for Architectural Presentation Videos with Lyria 3.5 Clip PreviewToonflow: open-source AI platform for short drama and video productionInstant NeRF: Training 3D Scenes from Photos in Minutes — Setup and StepsUsing Gaussian Splat in V-Ray: a step-by-step guide for Maya
TutorialsGemini APIMari

Architectural Image Analysis with the Gemini API: 5 Prompts and Steps

The Gemini API supports multimodal inputs and examples for understanding 3D scenes. This guide outlines a workflow for architectural image analysis, from API key setup to writing prompts.

A scale model on a desk in a design studio where an architectural image is being reviewedAI image
Representative image, generated with AI.Image: 3dsınıfı / FCA AI

In brief

  1. To get started, you need a Google account and an API key created through Google AI Studio.
  2. The Gemini API can be used via REST or with SDKs for Python, Go, Node.js, Java, and C#.
  3. The prompts in this guide focus on evaluating spatial layouts, materials, lighting, camera framing, and visual consistency.
  4. There is an example of using the API to understand 3D scenes; results should not replace design decisions and should be checked against the source image.

What will you learn?

This guide covers how to ask the Gemini API clear, well-scoped questions about an architectural image or 3D scene. The goal is not to have the image designed automatically, but to describe spatial relationships, create checkpoints before a presentation, and capture more structured notes about a scene.

Google’s Gemini API Cookbook lists basic prompt creation and multimodal input among its introductory topics. It also includes an example focused on understanding 3D scenes. The prompts below are original examples adapted for architecture and visualization workflows, not copied from that documentation. The model’s response should not be treated as a verified record of actual dimensions or design decisions shown in the image.

Requirements

To get started, you need a Google account and an API key created through Google AI Studio. The Gemini API can be called directly as a REST API, and official SDKs are available for Python, Go, Node.js, Java, and C#. The Cookbook recommends starting with its authentication and “Get Started” guides.

This guide does not require a specific software version. The client or code you use to add images to an API request must support multimodal input. Keep your API key secure; before using project or client images, check your organization’s data-sharing rules. Since usage costs vary depending on the model selected and the terms of use, review the current pricing for the relevant model before you begin.

Architectural image analysis, step by step

  1. Choose the input to analyze. Start with a single interior render, façade image, or screenshot of a 3D scene. Narrow the analysis question to fit the image’s scope: for example, ask only for feedback on the impression of circulation or the distribution of light. Don’t expect the model to provide measurements that aren’t visible in the image as if they were certain.

  2. Specify the task and its limits in your prompt. Rather than making a broad request such as “Analyze the image,” state which elements to examine and how to structure the response. Include the image in your API call as multimodal input.

  3. Ask for spatial relationships to be described. This example is designed to turn visible relationships in an interior render into structured notes.

Prompt
Bu iç mekân görselini mimari sunum kontrolü için incele. Görülebilen ana mekânları, açıklıkların konumunu ve mobilyaların birbirleriyle ilişkisini ayrı maddeler hâlinde tarif et. Görselde açıkça seçilemeyen bilgileri tahmin etme; belirsiz noktaları ayrıca belirt.

Use the response as an observation checklist that the project team can review alongside the image. Distinguishing between “visible” and “uncertain” helps reduce the risk of treating inferences as design data.

  1. Use separate prompts to review materials and lighting. Rather than asking for many different evaluations in one request, focus on a specific aspect of the image. This makes it easier to sort the resulting notes by render revision.

Prompt
Bu mimari görselde seçilebilen yüzey ve malzeme özelliklerini tarif et. Renk, parlaklık ve doku izlenimlerini gözleme dayalı olarak ayır; gerçek malzeme türünü görselden kesin belirleyemiyorsan bunu açıkça yaz. Görselde olmayan ürün veya malzeme adı ekleme.

This prompt does not create a materials list; it aims only to classify impressions that can be perceived in the image. Use project documents and samples to make actual material selections.

Prompt
Görseldeki ışık ve gölge dağılımını sunum kontrolü açısından değerlendir. Aydınlık ve karanlık bölgeleri, öne çıkan yansımaları ve okunması zorlaşan alanları ayrı başlıklarda belirt. Işık kaynağının yönü kesin seçilemiyorsa tahmin yürütme.

Treat this output as a checklist for reviewing the scene again, not as an instruction to change render settings directly.

  1. Review the camera and framing. Especially for presentation images, you can ask which design elements the framing emphasizes and which areas it leaves out.

Prompt
Bu mimari görselin kadrajını incele. Görselde en çok dikkat çeken üç unsuru ve kadraj nedeniyle daha az görünür kalan alanları belirt. Perspektif veya gerçek mekân ölçüsü hakkında kesin sonuç çıkarma; yalnızca görüntüde görülenleri değerlendir.
  1. Ask for a consistency check in the final round. Rather than treating previous responses as project facts, ask the model to generate questions for the team to check against the image.

Prompt
Bu görsel için tasarım ekibinin doğrulaması gereken beş kontrol sorusu hazırla. Sorular; mekânsal ilişki, malzeme okunabilirliği, ışık ve kadrajla ilgili olsun. Görselde doğrulanamayan bir durumu gerçekmiş gibi ifade etme.

Common mistakes

  • Treating dimensions or technical specifications inferred from an image as certain: A perspective image alone does not provide verified measurements. Refer to project sources for dimensions, material codes, or product names.

  • Packing too many tasks into one prompt: Split spatial analysis, material interpretation, and framing critique into separate prompts. This makes it easier to compare responses.

  • Leaving uncertainty out of the prompt: If you don’t want the model to guess, tell it to identify unknowns and not invent information that cannot be made out in the image.

  • Treating the response as design approval: The output may help with internal review and discussion, but it does not replace project drawings, the model, or review by the responsible designer.

  • Failing to check cost and access terms: Creating an API key is one of the initial steps; review the relevant access and pricing information before deciding which model to use.

Implications for architecture and visualization workflows

This approach can be useful for architecture firms, interior designers, and archviz teams looking to prepare a quick visual checklist before a render or presentation. Students can also practice distinguishing which elements in an image are clearly legible from those that remain open to interpretation. The source includes an example of understanding more complex 3D content; consult the relevant API guide for details about supported data formats and implementation.

In a workflow, review API access, key security, and cost terms before considering hardware. Since the sources do not specify local hardware requirements, this guide cannot recommend a graphics card or workstation. For production decisions, evaluate the image, model output, and project documents together.

Next steps

Start with a single image and one evaluation goal. Compare the results with the image as a team and keep a record of which prompts are useful. Then move on to the Gemini API Cookbook’s authentication and getting-started guides, its multimodal input documentation, and its 3D scene example. Teams working with code can use the REST API or one of the official SDKs.

Source and license

Google’s Gemini API Cookbook is published under the CC BY 4.0 license. This guide has been adapted into Turkish based on the source’s information about the getting-started path, multimodal input, and 3D scene example. The prompts are original examples created for architecture and visualization workflows.

Sources

1 source
A(
ai.google.dev (CC BY 4.0)ai.google.dev/gemini-api/cookbook
Summary

Source texts are not republished; short quotes are marked, everything else is our own summary and commentary.

3dsınıfı’s take
3dEditor’s assessment

For architecture and archviz teams, the most practical approach is to position the model as an assistant that lists potentially overlooked points, not as a tool that makes design decisions. Reviewing space, materials, and lighting with separate prompts can make feedback within the team more structured.

For firms and students in Turkey, it’s important to assess API costs, key security, and the terms for sharing client images before getting started. The sources do not specify local hardware requirements, so there’s no basis to conclude that a new workstation investment is necessary. Outputs should not replace the verification of dimensions or materials.

Frequently asked questions

What do I need for architectural image analysis with the Gemini API?

You need a Google account and an API key created through Google AI Studio. The API can be used via REST or with SDKs for Python, Go, Node.js, Java, and C#.

Can the Gemini API analyze 3D scenes?

The Gemini API Cookbook includes an example focused on understanding 3D scenes. Refer to the relevant API guide for implementation details.

Does the Gemini API provide exact measurements when analyzing architectural images?

The prompts in this guide are designed to describe visible elements; they do not provide exact measurements or material information. Verify such details against project sources.

Comments and the forum are in Turkish.Join the discussion
+

Related news