Yesterday, with the help of a tutorial I managed to create a geometry nodes in Blender, with which to generate a pattern of panels of different widths, randomly arranged one next to the other.
I started from the idea in this post about panels:
Today, with the help of Gemini AI I created an extension for SketchUp, ms Panel It, which generates a similar pattern. The extension still requires some corrections to improve the way it works.
ms Panel it extension
Hi! Looks like very exiting. How I can test this plugin?
For now, the extension is still a work in progress and has not been published, I need to solve a few problems. Thanks!
From this post about ‘precise control of camera’
ms Camera extension
You can control the following camera settings (see photo/video):
creator: Gemini AI for ms
In test, unpublished.
ms Camera extension
Really cool
Thanks! Yes, it’s useful when you want to more easily render a large-scale construction like temples in SketchUp.
The Dacian Fortress of Căpâlna - hypothetical 3D virtual reconstruction
It was built 2000 years ago, in the second half of the 1st century BCE.
3D model that I made in SketchUp 8 and now rendered using Gemini AI (Nano Banana 2).
The Dacian fortress of Ardeu - hypothetical 3D virtual reconstruction
It was built 2000 years ago, in the 1st century BCE.
3D model that I made in SketchUp 8 and now rendered using Gemini AI (Nano Banana 2).
Im surprised at how consistent the images are between the changes of the seasons. I thought this consistency was a failing of AI but its obviously made improvements. Looks great.
Thanks!
After giving Gemini AI the base image, a SketchUp scene, and a description of all the elements in the model, I started with a summer image. Then, I just asked it, one by one, to keep the base and composition (rocks, vegetation, hills) and to change the season and time of day.
But it is necessary to be as clear as possible and not to overlap the requests.
When I told it to position the four guards along the entire wall, it got a little confused and inserted a lot instead.
It also had a problem when I told it to render two images - foggy morning and winter, it collaged them into a single image.
But it manages to capture very well what I specified and even complete it with well-thought-out things (the entourage, for example).
And there’s no limit to the number of words I could write to it.
A Dacian medical kit from 2000 years ago - 3D virtual reconstruction.
Discovered in a dwelling located on a plateau at Sarmizegetusa Regia (GPS) - the capital of Dacia (Romania today).
In that location, grouped together, very close to each other, were found: a bronze tweezers, an iron knife blade (scalpel) with bronze guards, five very small fired clay jars and a plate made of pressed volcanic ash. All of these were contained in a wooden box, destroyed at the time of discovery, leaving behind, apart from the charcoal, the neatly crafted iron handle and a few fragments of the bronze belt.
3D model that I made in SketchUp and now rendered using Gemini AI (Nano Banana 2).
SketchUp photos
The scalpel and tweezers, after 2000 years, at the National Museum of History of Romania (Thanks, Cătălin Borangic)
The Dacian fortress of Șimleu Silvaniei - hypothetical 3D virtual reconstruction
It was built 2000 years ago, in the 1st century BCE.
3D model that I made in SketchUp and now rendered using Gemini AI (Nano Banana 2).
Remove material from nested groups/components in SketchUp (2017-2026)
Hi @eric-s, first of all, thanks for the new video!
Based on your video, I asked Gemini to create two images for me, the first based on your sketch, and the second based on a sketch I made based on the one you created.
I used the same text for both images.
These are the results:
Hi @Mihai.s They look impressive. Would you be kind enough to share the text description used. Being very new to AI rendering it very clear that the wording of the description to the AI engine is very important.
Thank you
Hi Patrick,
Yes, with the caveat that in the discussion with Gemini I used Romanian. But you have the image I used and if you want you can test with the English translation, I think it would generate a similar image.
Hello Gemini, please render a photo-realistic image based on the new sketch attached.
The fence is made of vertical wooden slats. The 5 stylized trees have that spiral shape. In front and behind the wooden fence are flowers of various colors and bushes. That sign with text is created from several irregular parallelepiped-shaped boulders, stacked one on top of the other, and on them is attached a wooden plate with the text ‘cambrian’ carved into it and highlighted by varnishing/painting it in the hollow area. At the other end of the plate, in the back, it is attached to a curved piece of metal the color of the wood. The entire sign is affected by the passage of time and weather. A few fluffy clouds are in the sky.
The sketch also shows a silhouette of a man so that you can have the scale of the entire image.
Thanks!
Thanks! 
Thank you for this Mihai.s much appreciated.
I’ll have a go later today.
