Is it realistically possible to generate a…
0 reactions
Is it realistically possible to generate a production-ready SketchUp model from photos + measurements using AI?
Hi everyone,
I recently got a freelance project, and before accepting the full development, I’ve been asked to do a feasibility study to determine whether the main idea is actually technically achievable.
The client works in carpentry/interior fitting. The basic workflow would be:
* Someone goes to the customer's site and takes several photos of the room/area with a phone.
* They manually take accurate measurements and enter them into a separate file, probably Excel.
* Both the photos and measurements are uploaded to an application.
* The application should understand the existing space and generate a **3D model in/for SketchUp**.
* Eventually, it should also be able to add the new furniture/carpentry element into the scene and use the real measurements for its dimensions.
The measurements are important here: I'm **not expecting AI to accurately determine dimensions just from the photos**. The photos would mainly be used to understand the layout, objects, walls, orientation, etc., while the measurements provided separately would be the source of truth for dimensions.
The part I'm unsure about is the jump from **photos + structured measurements → usable SketchUp geometry**.
I've been looking into using computer vision / multimodal AI to analyze the images, then generating structured scene information that could be passed to SketchUp through its Ruby API or a custom extension.
My main questions are:
1. Is this technically realistic with current AI/computer vision technology?
2. How reliable is AI at reconstructing the geometry/layout of an indoor space from several photos when accurate measurements are also provided separately?
3. Can this realistically produce a **production-ready SketchUp model**, or should I expect AI to only generate an approximation that a human would then need to correct?
i d appreciate any advice and thank you in advance .
#dev #programming #technology
source
Hi everyone,
I recently got a freelance project, and before accepting the full development, I’ve been asked to do a feasibility study to determine whether the main idea is actually technically achievable.
The client works in carpentry/interior fitting. The basic workflow would be:
* Someone goes to the customer's site and takes several photos of the room/area with a phone.
* They manually take accurate measurements and enter them into a separate file, probably Excel.
* Both the photos and measurements are uploaded to an application.
* The application should understand the existing space and generate a **3D model in/for SketchUp**.
* Eventually, it should also be able to add the new furniture/carpentry element into the scene and use the real measurements for its dimensions.
The measurements are important here: I'm **not expecting AI to accurately determine dimensions just from the photos**. The photos would mainly be used to understand the layout, objects, walls, orientation, etc., while the measurements provided separately would be the source of truth for dimensions.
The part I'm unsure about is the jump from **photos + structured measurements → usable SketchUp geometry**.
I've been looking into using computer vision / multimodal AI to analyze the images, then generating structured scene information that could be passed to SketchUp through its Ruby API or a custom extension.
My main questions are:
1. Is this technically realistic with current AI/computer vision technology?
2. How reliable is AI at reconstructing the geometry/layout of an indoor space from several photos when accurate measurements are also provided separately?
3. Can this realistically produce a **production-ready SketchUp model**, or should I expect AI to only generate an approximation that a human would then need to correct?
i d appreciate any advice and thank you in advance .
#dev #programming #technology
source
| 3,000 mlx | |
| 4 | |
| 0 |