I have been developing a tool for creating fast, precise and FLEXIBLE architecture. With a workflow that makes sense in real world pipelines. It’s been sitting on my harddrive for 4 month untouched because of work. I want to develop it more and share or release it. In the coming years this tech is going to boom. There are some building generators out there. And construction companies are doing stuff that are purpose built… But a CREATIVE tool with high detail? I don’t see any tools out there trying to do it. I want Blender users to be at the forefront of it. I can’t be the only one who sees what’s possible.
7 Likes
Welcome to BA, and nice work 
I think you’re not the only one riding the wave of the future here- have you seen this thread?
Or this one?
You might be able to pick up some tricks to make your work even better
1 Like
Hey ! Looks like a cool project !
While trying a bit myself I find that it’s not easy to make a one fit for all tool.
I tried making architecture with lots of details, but then it was too much geometry if I wanted to try to make a little village or part of a city.
Maybe the best tool would be an addon that helps organize things a bit better, and for sure having some geo-node tools on top of it.
Buildify looks really cool, there is less details like inner structures, walls, but that’s why you can put a lot more buildings together !
Good luck !
4 Likes
Thanks for you comment my friend. Sadly I my sloppily put together video has failed to explain my vision. But I do hope it generates discussion!
I don’t care about big cities and diversity. That’s uninteresting. Buildify and the other are not my competition. That’s precisely why I showed them. Scale and diversity is not primarily what this tool is about. In fact even the creation of buildings is not central to the tool’s function. Those are all secondary and “trivial” features. You could even call them seperate tools.
This tool’s primary function is architectural creativity. Unifying an “architectural language”. Recreate existing styles or make entirely new ones. You’re absolutely right, it is difficult. Defining the relationships between different architecture. THEN, ontop of that, organzing the huge huge array of parameters into intuitive settings that aren’t half a mile long, without losing articulation at the user end.
The technical aspect is largely trivial. It’s the PLANNING. My real work here is in the groundwork. As programmers will understand, there’s many ways to build a tool, but improper definitions of your parameters will kill flexibility later on. It’s all about sitting down and looking at as many types of architecture and anticipating. The fewer corners you cut, the more freedom the end user will have.
Poor planning means you end up with something like Spore creator. Which no hate, I love, and was insanely ambitious and impressive, but was restricted by planning (time) AND technology. Architecture though? It’s entirely possible. And infinitely less complex than trying to define a unified “language of biology”. 
2 Likes
Capturing the entirety of human architecture and breaking it down into logical attributes is a worthy goal for sure, but it’s quite broad… possibly too broad, I don’t know if it’s truly possible to make one generator that covers millions of architectural styles over thousands of years and hundreds of cultures. There’s a reason that existing generators tend to be more focused- even if you can successfully implement all of this, your user interface will be unmanageable. You’ll find it difficult to distill this many parameters down into any less than dozens of not hundreds of front facing controls.
I don’t mean to discourage you, but as someone who has worked on some very ambitious failed software projects, I have found that the easiest way to create a unfinishable, unusable product is to have a broad scope of goals. Think about how Google Docs, Sheets, Slides, and Drives all have just one function and operate under limited parameters. They could easily be combined- it’s been tried before, the all-in-one office software, but none of that all-in-office software exists anymore 
or creative cloud- why has it endured where so many others have failed (sadly)? Because each app does something very specific and does it well, rather than doing a million things not very well. Breaking this into focused subprojects will help you a lot.
Ok, unasked for advice over, take it or leave it at your pleasure
2 Likes
Not at all brother, I appreciate it. Criticism is my map and naysaying is my fuel. Your criticism is EXACTLY what I’m after. My goal here is produce discussion, connections and support.
I don’t neccesarily have preconceptions on the form this work takes. As a deliverable packages, yes they will definitely be divided. Certainly NOT developed sequentially. Since I’m still deep in the trenches, conceptualizing, those divisions don’t exist yet. This early work ensures that those divisions make the most sense for the foundations. I believe it’s a mistake to assume those divisions exist as things like “houses vs castles vs huts vs whatever”. Mind you, my scope isn’t strictly “all architecture” but I can clarify it later… If you’re interested. But I’m not backtracking, the grand scope is absurdly big.
My plan from here is to follow what structural framing tell me. Finding a branching where they fundamentally divide, testing those branches against my foundations in my mind and doing super dirty tests to ensure the principles are sound. BUT, picking the biggest branch and completely IGNORING the others. I continue this branching process until I find two sufficiently similar, yet distinct, small branches which are enough for me to tackle. Those two branches will be the basis for a TRUE proof of concept showing the tool move along that spectrum.
Maybe there’s a word for this way of working but I don’t know. But it will enable me to organically pick a scope that I can focus on to create a POLISHED tool built ontop of a truly versatile foundation. From there, I will rely on the power of the open source gods, as well as the experience I’ve gained. Alllll those ignored branches can be tackled one by one.
Yes, my second biggest concern is UI. Not neccesarily the NUMBER of parameters but managing them in a CLEAN way. Object properties, vertex groups and drivers alone aren’t going to be sufficient likely. For my polished proof of concept, I MIIIIGHT get away with it. I suspect that at some point my work will have to be rewritten in C or python. But that doesn’t make my work obsolete. Noding is a beautiful way to mindmap.
Has my audacity inspired you or made you scoff? <3 Thanks again for writing.
2 Likes
Ok, it’s interesting to see later how your idea take shape !
When you work on something that doesn’t exist yet, you have to be cautious either because you
might have an idea really innovative, or it’s just that everyone who tried it failed for similar reasons.
There are a lot of human generator, but not an universal insect of animal creator.
You mentioned Spore Creator (which I didn’t know about) but it seem to have a limited scope in order to work and therefore is bound to a shape language simple enough to allow a lot of variety but within the same style.
I guess having an universal architecture generator would find a lot of interest , so if it’s not done yet, unless you really know how to do it (and then you should really keep it for yourself first) it might just be that it’s too complex to be done , and then you can take blender as an universal generator 
Good luck anyway, I’ll keep a eye on what you’re doing now that I’m teased, but my expectations are really low at this stage ! But I’d like to be proven wrong !
2 Likes
When I posted in October, I was at my wit’s end. But the interest received (and a new, better paying job) pulled me through. Maybe I talked myself into it too. In fact, large strides were made by Christmas. A big, proper update should come by May. In the meantime, some UI/UX. This monster is unwieldy and I cannot rush straight to python. There’s a tad more than just recursive inputs involved. The UX will pay big dividends too. Luckily Blender’s UI is pretty easy to reproduce in Figma. Including the outliner data in the prototype is a really great way to work.
1 Like
I would love to see the tool in action. The video is no longer available. My project involves a 3D representation of a small town… but there are a little over a hundred houses. I’m just looking at different tools that might be able to help me build these houses. The problem is, I don’t need them to be procedural. They would have to be a 1 - 1 replica … as close as possible. And right now I’m doing it by modeling. And I just suck at it. The terrain is imported from the osm addon - open street map… and I’m using a road generator as well as Boutique for grass, trees. I’m keeping an eye on this as it is very exciting to hopefully get something like this. I can only imagine the nightmare it takes to achieve this. Unbelievable… salut to you Sir.