Archicad AI Visualizer
Discussions about Archicad AI Visualizer.

This group is dedicated to the experimental AI Visualizer for Archicad 27, which will reach End of Service on December 31, 2024.

The new AI Visualizer in Archicad 28 is now fully cloud-based, with advanced controls like creativity and upscaling. Click here for further details.

SOLVED!

Does AI Visualizer understand dimensions?

PV1972
Contributor

Hi!

I haven't tested AI Visualizer yet. I'm curious to know if it understands the dimensions of the model. The basic model in ArchiCAD is made in real life dimensions, of course. So let's say I have a a "box" 6...7 meters in height. Does AI Visualizer understand that and make (quite propably) 2-story building? Or can it be anything?

First I thought it would understand it, but now while writing this I came to opposite thought. 🙂

Thanks!

1 ACCEPTED SOLUTION

Accepted Solutions
Solution
David Collins
Advocate

The AI Visualizer team talked about this in their webinar.

The visualizer only sees and understands a flat image, so it never gets any dimensional information from ArchiCAD’s 3d model. Instead, it tries to interpret clues both from the image and the prompt. An “office building” will get more stories than a “house.”  I’ve found that placing a person somewhere in the image can help the Visualizer better understand the scale.

David Collins

Win10 64bit Intel i7 6700 3.40 Ghz, 32 Gb RAM, GeForce RTX 3070
AC 27.0 (4001 INT FULL)

View solution in original post

4 REPLIES 4
Karl Ottenstein
Moderator

Yes.  Simply look at the examples posted by people and you can see that the model size is understood.

One of the forum moderators
AC 28 USA and earlier   •   macOS Sequoia 15.2, MacBook Pro M2 Max 12CPU/30GPU cores, 32GB

Oh yes, the Gallery. Didn't notice that. Thanks!

Solution
David Collins
Advocate

The AI Visualizer team talked about this in their webinar.

The visualizer only sees and understands a flat image, so it never gets any dimensional information from ArchiCAD’s 3d model. Instead, it tries to interpret clues both from the image and the prompt. An “office building” will get more stories than a “house.”  I’ve found that placing a person somewhere in the image can help the Visualizer better understand the scale.

David Collins

Win10 64bit Intel i7 6700 3.40 Ghz, 32 Gb RAM, GeForce RTX 3070
AC 27.0 (4001 INT FULL)

Thanks for clarifying, @David Collins   I totally misunderstood what info the visualizer worked from... as I didn't attend the webinar (and have no interest given all of the machine installation hacks required to test this preview technology).  Given all of the open AI image models that just work with flat images, I just assumed that Graphisoft was doing something special/advanced that took the actual model into account.  Oh well.

One of the forum moderators
AC 28 USA and earlier   •   macOS Sequoia 15.2, MacBook Pro M2 Max 12CPU/30GPU cores, 32GB

Didn't find the answer?

Check other topics in this Forum

Back to Forum

Read the latest accepted solutions!

Accepted Solutions

Start a new conversation!