Archicad AI Visualiser - is there a purpose for it?

[Barry Kelly]

As the title asks, is there?

First let me say I don't have the equipment necessary to run it, so I have not jumped on the AI train.

When you get it running, what exactly does it do and what value will it be?

From what I can see, it can do some pretty renders in many different styles.

I assume the more accurate the model you feed it, the better the results will be.

But even then, in many cases it seems to distort the base model into something that is not constructionally possible.

Some of the roofs and glass curtain walls I have see are laughable.

Let's hope it is true AI and is learning from its mistakes.

But who is telling it where it is going wrong?

I hope GS employees are not tasked with this, there is some many more important things they can be doing in my opinion.

And so, you can get images of unlimited options for your fantasy design.

But that is all they are, just images.

There is no BIM component to help you document what you see.

It doesn't even tweak the geometry you feed it from my understanding.

Maybe it can be tweaked as an AI rendering engine, where it more faithfully keeps the geometry you feed it.

I see this as like asking a chat AI to write your university thesis.

Sure, it can be done and it may give you some ideas.

But you wouldn't really want to use it as it is.

Barry.

[scottjm]

I am yet to try it too, mainly cause of all the forum post saying ‘it’s not working, can’t connect to server’ etc. 

from my understanding basically all it does is take the view from the 3d window and pass that to the stable diffusion engine along with your prompt and the settings. It’s turkey an image based exercise. If you were to snip tool the 3d window and pass it to any of the numerous image generator AI tools you would be able to achieve a similar result. (Correct me if I’m wrong). 

the only way I can see it implemented into our office is to include an image of a design concept of building there are yet to design in a master plan report. Really just a ‘hey this building on this part of the site could look something like this, but you aren’t paying us to design it just’. 

i think at this point adding the words AI to your product are what the words blockchain were a few years ago. It’s irrelevant whether you are actually implementing real AI integration at all (or blockchain for that matter) you just had to say the words in a press release or marketing material. If you didn’t, your stock price would tank. 

[scoz]

Please understand that my English is not good enough.
I've read the above.
I'm not an advocate for GS, I'm not a spokesperson for GS
But I'm using AC well enough for my work and I think it's a really good program.
There are a lot of opinions about AI-Visualizer
Stable Diffusion is the basic image generator and it's called TEXT TO IMAGE
I think we should acknowledge that there is a premise.
I want you to know that the images generated by the system or patience you use vary widely, and the images generated are unpredictable.
Sometimes I enjoy the moment of waiting for an unpredictable image.
I looked around all the AI programs listed above.
They are all great and good programs.
These are different form factor programs from AI-Visualizer, so I think the simplest solution is to purchase and use the program if you want it.
It is a separate matter that GS provides new AI tools
It's kind of funny to keep talking about the effectiveness of AI-Visualizer in this forum.
AC itself is a good tool, and if you don't want a solution to AI, it's suitable for you
I think it's a good way to find another AI-TOOL.

[Jeno Barta]

Hello Everyone! 
First of all thank you for taking the time and effort to give us feedback on AI Visualizer, and also Graphisoft as a company.

I would like to address some of the concerns you have raised in this topic:

"Why are we doing this, instead of addressing Archicad's main problems, and developing the features our users want?" - The main thing to understand here, is that it not works like, all the GS employees jump on a project like AI Visualizer, and we do not develop anything else besides it. The truth is that AI Visualizer is developed by a fairly small team, who did not take away any resources from the main projects. So please do not think that if you do not get some feature, or fix in a main version, or an update it's because AI Visualizer. It's a fairly small innovation project, released as an add-on. 

"It's a bad rendering engine." - We did not made, and marketed AI Visualizer as a rendering solution. When we started to experiment with visual generative AIs, it's was clear from the start that AIs cannot render. Rendering needs to keep all the features of the planned building perfectly, and AI cannot do that (Now). But we saw the potential in inspiration, and made/marketed this way.
AI Visualizer is creating concepts from mass models, with the intent to give you ideas, and to be one inspirational source among others. Thats basically it.  

I understand that the name of the solution could be misleading as "visualisation" is a synonim for rendering, but we tried to be clear on this in our marketing efforts. 

"Others doing the AI thing way better then Graphisoft, so we should not do anything." - Sure, this is a fairly new technology, and we are experimenting with it too. There are a lots of ways to go with AI. There are machine learning algorythms for data analysis, deep neural networks, with foundationl models for a lots of general purpose, etc. We could develop our own, buy / rent one, or finetune and already existing foundational model. There are a lots of opportunities and we try to find the ones those fit our goals the best. 

Sure there are a lots of startups, businesses who create amazing AI based solutions for space optimization, feasibillity studies, and there are a lot more who use visual generative AI (mostly based on Stable Diffusion aswell) like us with AI visualizer, and were able to add a lots of features to their solutions. But the main thing to understand here, and i would to reference back to my first point, we do a lots of other things, and AI Visualizer is a fairly small add-on project. 

We do not think like, "if we cannot be the first and best let's just don't do anything". We loved to experiment with Visual AIs, and we think it's a really cool tool for inspiration, for our users. It's not perfect, but it's a nice, free additon for our users. 

"It's a marketing gimmick" - We as a company try to follow the latest technological achivements, and incorporate them in our products. Sometimes we hit, and sometimes we miss, but we try to be there when something happens in the tech world. Sure there are a lots of very complex usages of AI, but as I said we are developing solutions on a lots of fronts, and trying to hit a balance where to incorporate them, and where not. AI Visualizer is a nice lightweight start to show that we are investigating AI, and yes it has its healthy message part aswell. But we do not aim AI Visualizer as just a marketing gimmick. 

So as a summary, please do not think that because of AI Visualizer, parts of Archicad did not developed. Also we tried our best to communicate the AI Visualizer is a source of inspiration in the early stages of design. It's free, it does not require any server connection, and it can give you ideas if you would. 

Lastly we do not aim to compete with startups that exclusively work on solutions like this. We are experimenting with AI, and try to give our users the best from it. 

Thanks again for your feedbacks!

[torben_wadlinger]

GS took the wrong route with the AI-implementation.  Have a look at SketchUp, where Trimble did it the right way: just install a simple, small Add-On (instead of 23 Gig of stuff that might work) and do the AI at the company-server for no cost. That's it.

[Jp1138]

"for no cost" - for now. The use of a local solution is the best thing about this GS experiment, as far as I am concerned. Any server based solution is going to be a new subscription sooner than later, and no mention of the privacy issues of sending data away.

[torben_wadlinger]

I am 100% in local solutions! But when I have to choose between 23 Gig of weired stuff installed onto my machine and lets say 1 MB just to try it - I'll go with 1 MB.

[Jeno Barta]

@torben_wadlinger @Jp1138 
Yes, renting servers for a visual generative AI's computational needs is very expensive, building your own server park for that is even more expensive. 

We chosse the path to give the whole technology to our users, and we also put our twist to its top. I do not talk about the interface, i talk about the underlying structure of AI Visualizer. When it comes to foundational models, there are a Lots of paramteres you can fiddle with. We put a lot of effort to optimize these parameters for architectural inspiration uses, and channel them to just a couple of parameters on the screen. 

We are absolutely aware that its not easy to install, and pretty resource hungry, but for our first experimental feature, we choose this way. 

[scoz]

I did some work with AI-Visualizer again today.
I found this tool very useful again.
Surprisingly, this tool is also free.
Running this AI-Visualizer makes me study a lot.
Has anyone wondered why AI-Visualizer never imports 3D TEXT?
I've found the way and I'm using it properly when I need it.
I'm an individual who has nothing to do with GS, but I don't understand how much complaints about AI-Visualizer are discussed in this forum.
Again, if you don't like it, you don't use it.
Below are some things I worked on today.
Isn't this good enough?

[MASz]

I definitely feel like it is a marketing gimmick. It just feels like Graphisoft riding the AI wave to seem relevant. This whole approach has not impressed me at all.

I would say the future is real-time visualisation in Unreal Engine. It has been key to a lot of my projects. Clients love it. It goes beyond visualisation and is a design tool as well.  The problem is still the inherent low poly counts in Archicad, from curves (walls, slabs, complex profiles etc) to some embarrassing looking objects. I've laboured this many times over the years and there is still minimal progress and no indication that curves in 3d will be ever addressed.

And yes, I get that it's not a direct comparison because the AI is supposed to be for "inspiration". Well, I have many feelings towards AI but there are few instances where I would be looking to AI for inspiration. I am not sure who would actually be doing that.

It's all well and good to say it's not taking resources off main projects, but again all I see is that there are these long standing issues that do not get addressed while new, problematic and underdeveloped features are thrown into the mix.

Is it even free? Our subscription fees went up significantly without explanation. Not necessarily blaming the AI visualiser for that, but it doesn't exactly create an environment where I am feeling warm and fuzzy about it.

[torben_wadlinger]

You meant well, but you did it badly. The AI tool thus joins the long list of typical Graphisoft developments. When will you finally realise that before you go public with things like this, you should perhaps do a beta test BEFOREhand. Or simply consider using AI differently to your competitors.

I'm stunned every time I read statements like this.

At the latest when I realise that a considerable amount of work with containers etc. has to be installed in the background and the installation package is a whopping 23 gigabytes in size (which also has to be sent over a line) - at the latest then someone in product development has to jump up and shout "Stop"!

How is this visualiser supposed to continue? Is it finished now, will it stay that way? Is there any further development? Or is it just another one-shot thing?

By the way: I once rebuilt your system for fun on an old HP-Z workstation with an NVIDIA card: Ubuntu, Stable Diffusion, ControlNet, WebUI and a prompt extension for better prompts. Nothing to do with a server farm. That's humbug.

So, if you have fun with something like that, then go for it. But solve the important things first. You're playing with our money here.