AI Visualizer in TP28

[Aixtreme]

Hi,

the AI Visualizer in TP 28 is not working at all and a lot of features/options are gone. Is this the final version or can we expect more options in the interface like: Picture size aso.?

Best regards,

Frank

[Kenneth Solvik]

Hi Frank,

We are now in the final development cycle for the AI Visualizer. Our current focus is on the ease of use, prompts, explained more in detail in this post: https://community.graphisoft.com/t5/Graphisoft-Technology-Preview/Optimizing-Architectural-Prompts-f.... The post also includes a session on how the community can best influence the improvement process. In this context it’s worth mentioning that improvement process includes QA, such as bug fixing, malfunctioning, or “not working” as you described it.

Would greatly appreciate if you can provide more details, on what is not working and if you have suggestions for improvements. I will see that your feedback is added to our backlog for review and addressed either in AC28 or in the next version of the AI Visualizer.

Thank you!

[Aixtreme]

Hi,

well here are my experiences so far:

1. I opened a file that I made with AC27 Edu Version. I cannot open it, because of the lack of the lic information. When I try my old lic info and open it, the AI vis is not working

2. See the screenshot. I draw two slabs and 4 walls and this is the result. So what does this one slider mean? It needs to be much more details and of course a slider of how much to take the original construction. And by the way: the view angle is not correct...

There are good ideas of the additional options that are needed in the final version. Else I would suggest to put the size of the results by myself, especially when "upscale" is not clear.

Most important: for the edu version it should work without payment.

Best regards

Prof. Frank Lube

[Aixtreme]

Here are some more screenshots with 0, 20 and 40% creativity...

[Barry Kelly]

If I have no model at all, I still get a result.

Try giving your 3D a bit more of a perspective view and maybe some more detail in the model itself.

It doesn't actually use the 3D model, just the 2D image you see on screen of your 3D model, so it needs to get depth and perspective from the rotation of the model.

And more description in the prompt.

If you want a front elevation view, add this to the prompt.

It would be good if there could be some pre-populated but editable prompts to choose from.

Barry.

[Kenneth Solvik]

Thanks Frank! Your input is reported. 

Best regards, Kenneth