We made a foray into designing and building a brand new house and let me tell you, my fellow HNers, the state of affairs with the design part is ABYSMAL, even in a reasonably modern architect firm.
Basically you first design, for look and feel, then you calculate technical details, then you build. Or rather the architect does all three, but you would normally want to "chip in" during the design phase - change the layout, move things around, etc. That's where things get bloody awful.
I wasn't expecting full VR tours of the place (though it is doable [1]), but I did expect to be able to walk through the place in 3D and see how it will look like in the winter morning and summer afternoon. Hell no. What are you smoking?
Floor plans. Planimetric facades. Roll your eyes to the ceiling and imagine. 3D renders that looks like from the early 00s take hours to produce, so they aren't done often. That's _still_ renders, mind you. 3D walk-abouts are possible, but they look like butt and they are viewed as the final step of the process, when everything has been done and done.
And this is with firms where principles are in their 40s and generally computer literate. It just seems that the industry is very conservative with its tooling choices AND these tools are also ridiculously expensive, further stifling any desire to switch. So they are stuck with using technical design software like ArchiCAD (good for laying out piping, calculating stress, air flow, etc.) for visual design. And the resulting process is sooooo slow, small changes and iterations take hours if not days to complete, so it takes weeks to converge to a general design of the house.
I designed/built my own house too[1], giving sketches to an architect and working with him. Some major takeaways for others who have yet to build:
* No one will care about your project as much as you and this includes the architect :(
* Few(???) architects and zero builders seem to understand light/wind/circulation like people apparently intuitively did 200 years ago. You might be better off with the average curious engineer designing the house than the average architect. Everyone is astonishingly lazy, even mansions around here have terrible light/circulation problems.
* VR is no match for IRL! It's best to actually get a measuring tape and stake out the dimensions of the house on the land. An even better, get chalk and go to a big flat parking lot and draw out the entire floor plan (compass-correct) and walk inside, looking at the sun (keeping in mind what season). This worked well for me, I should have done more of it. I didn't build a porch though, which is where getting more exacting summer/winter sun angles would matter.
I actually got the chalk idea from the excellent movie The Founder, where the creators of McDonalds try out different kitchen workflow models this way.
* Draw and sketch a huge amount. Sketch facades freeform, with the ruler, etc. The fidgeting will help you discover things.
* Read a bunch of books about old houses. Maybe most approachably: Get Your House Right and A Pattern Language
* Get a pinterest board for you + spouse. We used ours heavily and it was great for keeping track of interior detail decisions.
[1] It's a fundamentally simple house, basically a box that maximizes light and airflow and wood burning heat, so I probably had a much easier task than you. Some of the original drawings: https://twitter.com/simonsarris/status/1183150002002112512
Drawing your plan on the floor sounds a good idea, but isn't there an effect where outdoor surfaces look much smaller than the same surface indoor? How do you account for that?
Actually that seemed to be a big problem of computer models to me too. When I was a kid I modeled the houses I was living in using Counterstrike Source's Hammer editor, and they always felt SO SMALL to walk around "inside" when they were correct to scale, even though one of them was a fairly big 1800's house. So either way you simulate things you have funny perceptions.
And during construction you are kind of thrown about by the perception of the size. When framing is done, it feels huge. Once all the plaster is on the walls, it feels extremely "small" at first because you were so used to the big openness of the frame, then finally as more finishes are applied it just feels correct at last.
It's hard to describe exactly what feeling happens. But suffice to say it's never easy to get "right" feeling simulations of a structure until you put up literal walls.
For this reason, if you want to build a 5x7 bathroom or 11x13 bedroom or whatever, the best possible thing you can do is find one that already exists. Or find a 11x11 room or something and try to imagine it with 2 more feet. Then you can carry these mental rooms around with you.
You also move incredibly, absurdly fast. Play half-life 2 in VR and you will get very motion sick, but you will also realize how insanely fast you are moving.
This has been my problem. Planning with a floor plan is not remotely the same as being there. What I think will work on a floor plan doesn't when I actually try it.
VR would work for me because the sense of space exists in VR where as it does not in floor plans nor on 3D renders views on 2D screens.
I really enjoy the discussion that emerges here. It is rare to find so many viewpoints and constructive criticism on the way architecture services function today. As an architect I am taking notes.
You mention circulation, so I wanted to share a service I’m currently working on to check floor plans for how good the circulation is. Sunlight is coming soon too. I would appreciate your thoughts on it.
At the moment it simulates the most common scenarios in a home and the paths in the house that correspond to them, for example, getting back home with the groceries bags. Then it tells you if those paths can be improved and suggests how.
I use computer hardware everyday. Does this allow me to give relevant opinions on what are the best processes to build it? Some folks are thinking about it for decades.
Architectural rendering is overrated, it only exists as a way to sell the project to clumsy end users, investors and city planners. The same house but with a different sky, flying ducks, evergreen trees, people dancing and ferraris will look an order of magnitude better. The time wasted on this could be used to better the house itself. Don't forget also that what seems clearly an anachronism is not: 10 years ago computer hardware and software were not nearly there for the "3d revolution" you are asking for.
For a trivial example, the exact same house plan rotated by 90 degrees will have substantially different important qualities because of how the sun affects both light and temperature. There needs to be some way not only to communicate this to end users, but also take it into account during the planning and design process. Perhaps rendering is not the right tool for that and is overrated, but what methods do you suggest and use to understand and analyze these factors?
Those factors are always considered on a macro level when an architect makes a project. Only when they are critical, they are analysed on a micro level with specialized software. In most cases you can get very far with some rules of thumb:
North: indirect light, the coldest orientation (northern hemisphere of course)
East: morning light, good for bedrooms
South: lots of direct light, the warmer orientation
West: afternoon light, good for living rooms or even kitchens
I just want to understand if I will have the sun shining in my eye on summer mornings when I'm lying in bed, how the hallway will look at night, what the view will be from the kitchen and into the living space, etc. Just trying to place _myself_ in the house and see what I should fix before it's built.
You wouldn't buy a car after merely looking at its 2D projections and a sketchy looking rendering, so I don't think that asking to see how a house would look and feel before committing to building it is such an unreasonable thing to do.
> Architectural rendering is overrated, it only exists as a way to sell the project to clumsy end users, investors and city planners.
Right but isn't it overrated because it only exists as a way to sell the project? Surely the software could be improved to the point that you can use it to make decisions?
I can recommend ChiefArchitect. So far it has been great. It is within reach for a technical person to figure out, especially with all the training videos available online for free.
It is also powerful enough to do a lot of the details as well, which we are doing ourselves and then sending back to the architect for review.
I just finished modeling my house in Chief Architect, and I'm very impressed with it. There are things I don't love, especially the price as a plain homeowner (and the lack of plumbing / HVAC modeling), but it's an impressive tool, and the online tutorials are excellent.
One important note if anyone does try it, the system of measurement Can Not be changed for a file. I started mine in metric, because I generally prefer metric and it was a nice way to measure using a laser measuring tool. But I live in an imperial world (US), where all stores and contractors work in imperial units and there has been tens of times already that I wish my plan was also imperial. And Chief Architect can't convert the file. I can display imperial units, but the base units are always metric.
Just an FYI, I was able to make a rough design of our small house using the free, opensource sweethome3d program, convert it to webvr using obj2vr, and then virtually walk through it.
This is a blatant plug for something that's not ready and still in alpha: freecad (www.freecadweb.org)
It's an open source parametric modeler and the export to (still) renderers like appleseed, luxrender,... and import to blender is still a work in progress.
I'm hoping someone reads this and thinks "Hey, I can contribute to this" or at least file a bug/feature report.
The funny thing is, even these cumbersome tools are so much better than what the software industry has.
Small changes take only hours to complete? It takes only weeks to converge to a general design? Most software project managers would kill for that level of efficiency and predictability.
(A bunch of pictures made in Sketch is not the design. You can surely do that in a few days, but it’s essentially the same as making a bunch of facade drawings and pretending that’s the architectural design.)
Where it was so archaic(least in UI/usability) and extremely expensive that AutoDesk has found success following the Unity/Unreal model. Offering a "free" CAD/CAM software with Fusion 360. With plenty of Youtube presence and advertising with it, letting it grow.
However the problem with Architecture and Building itself, is that building homes and structures isn't really a hobby like Game Dev, 3D Modeling or Manufacturing(is becoming).
for thirty years (or more) here in California (home of exciting computing technology) professional Architects have been piling up as dead bodies on beaches and on the side of the road -- ok, not really, but almost. Every brilliant artistic and scholarly 16 year old wants to be an architect, then school admissions, then graduate programs, then debt and no work (for those that make it that far). Major Architecture programs at a certain top-10 school have a "Suicide Prevention" poster in the main hall. Money in the United States is aimed at fast and profitable, not design and long-lasting. Hundreds of large houses in Los Angeles, perfectly good, torn down to be replaced by larger interior-square-footage boxes, as fast as possible. McMansions in the extent Bay Area are made of sheetrock and Home Depot materials. Am I exaggerating? sadly, no.. How can innovative software be designed and profitable in that environment ?
>I wasn't expecting full VR tours of the place (though it is doable [1]), but I did expect to be able to walk through the place in 3D and see how it will look like in the winter morning and summer afternoon. Hell no. What are you smoking?
Architecture is a competitive industry. If you want a 3d model you can walkthrough it's not the end of the world - even Trimble sketchup pro has a built in walkthrough tool. You can also export most models to something like cinema4d for a video walkthrough. The models you expect to walkthrough to get some idea of light will require new texture work, lighting, and lots of small details that they otherwise have no reason to add. But those all cost time and money they probably had no idea you expected from them.
The thing is that you probably didn't make your expectations clear up front. If you did architects would quote you a price with that included. But theyd also say something like x revisions and then it's $125/hour so you dont waste their time.
It's a pretty reasonable if you have specific non standard expectations. They'll be happy to mark up the value of having someone else do it and charge it to you.
I asked all firms we spoke with how they work. None even had an option of making tweaks to the project with real-time 3D feedback. That is - change something and see how it looks in 3D on the second monitor, move the camera around, change time of the day, turn lights on and off, then try it differently, check again, try third, etc. Aim for dozens of "revisions" in a scope of one hour - that's not even remotely possible.
The existing model - get a draft, send in a list of tweaks, get a revision, send more comment, get another revision - is ridiculously inefficient and excruciatingly slow. What can be decided in a couple of hours takes days if not weeks. Coming to the architect's office and doing the same by sitting aide by side with then doesn't help much, it's still not a real-time and they get frustrated because that's not how they work.
The problem, as I mentioned, is both in architect software lacking (what's been available in game engines for years!) and the willingness to adopt the new tech and workflows on architect's side.
I mean, with all existing technologies I should be able to see my house-to-be in near exact realistic detail before we OK the project. This is NOT possible and this is supremely frustrating.
>None even had an option of making tweaks to the project with real-time 3D feedback. That is - change something and see how it looks in 3D on the second monitor, move the camera around, change time of the day, turn lights on and off, then try it differently, check again, try third, etc. Aim for dozens of "revisions" in a scope of one hour - that's not even remotely possible.
It's totally possible. 99.999% of clients aren't willing to pay for it, and architects aren't interested in having you do a dozen revisions no matter how much effort you think that is (an hour lol!).
I can tell you that simple walkthrough stuff is simple in sketchup (time of year/day built in), but most house builds don't require a detailed sketchup model with lighting real textures, rendering, etc.
You also have to remember that they are interviewing you as a client. If your expectation is 12 revisions in an hour as a reasonable, then there's very little they can do to make themselves competitive and I can understand why they wouldn't offer a likely unreasonable client an atypical contract.
You're going to them to get a proposal/quote. I can understand not being willing to offer you something you're clearly not willing to pay for. And from their perspective there's a lot of risk involved in taking you if they have someone else who isn't asking for atypical free extras knocking on their door.
Changes being "hard work" is a matter of a tool not being well adapted to the job, which is a problem with the tool, not the process. If I am playing with a floor plan, then certain changes, like moving walls and openings should be as easy to do as possible. If they aren't, it's the wrong tool for the job ... which is exactly what I mentioned in the first post - that the architects routinely use heavy-lifting ArchiCAD to do design phase sketching.
I think you're trying to describe bim software but you don't know what it is. This already exists. The most common software for it at the moment is Revit.
There are a lot of technical reasons why this may not be possible. For example, modern real-time lighting often uses precomputed data such as lightmaps to speed up/avoid real time calculations. For static geometry and lighting much of the work can be done 'offline' before runtime however it is not unheard of for this offline process to take hours (e.g. computing high resolution lightmaps for large environments).
There has been huge advances in this area recently. Real-time ray-tracing and lighting is Nvidia’s primary marketing push right now with their RTX cards. Unity/Unreal/Blender can all produce very high fidelity real time lighting while editing. It’s great technology.
Architects at the institutional scale (pharma, universities, government) have been doing integrated whole-building modeling for at least a decade. The industry keyword is "BIM" = "building information management". Static renders of building walkthroughs have been around for a while, but firms have been demonstrating weekly interactive walkthroughs over video conference for at least the past 5 years. They also do pretty accurate fluid dynamics modeling for smoke clearance in fires, ventilation etc. which takes into account interior surfaces and sunlight. You probably just need to find a better firm-- Architects that do private residences are (stereotypically) not the best and probably just don't want to go to the effort.
It's strange what you report, if you can set aside actual renderings a basic tool like Sketchup can do most you are asking for, it is easy to learn at least basic use (way easier than many professional cad's).
It was once a free (and standalone) program (it was for some time bought by google) though now I believe the free version is only "online" (and the other ones have become largely Saas):
There are products like Enscape that can deliver first person (or VR) walkthroughs at the push of a button. Sure, getting a perfectly correct lighting simulation cannot yet be done in realtime, but good approximations are possible.
Enscape looked great on paper but the amount of anti-aliasing makes everything look blurry - like first-gen Unreal VR titles before they fixed their Forward Renderer.
I understand they can't use MSAA because they use a deferred rendering path (which they need to support large numbers of dynamic lights) but whatever AA they use looks bad.
Some people might not notice it but it ruins it for me. I'd almost rather have aliasing in VR than "Vaseline on the lens"
There’s a bright future for better design tools but it was very interesting to talk to architects about them. Most are hugely conservative and want tools to fit their current workflow rather than anything that might be too innovative. That includes how much clients are able to get into the design process.
It could be worse. I've seen 2D architectural drawings that could have been M.C. Escher works. You at least got a design that was physically possible in three dimensions.
That's because software sucks badly. Architects are used to elegant and simple tools that work all the time: pen, paper and few other things. It's not just CAD software that sucks, it's the interfaces like mouse and keyboard that are slowing down design on computer. And yes, there are other systems but as you say they're not cheap so student architects won't learn on those.
Drafting & design are just hard problems. Starting with the mental models and human cognition. All the way thru workflows, use cases, and artifacts produced.
As a kid, I made the transition from drafting boards to CADD. My efforts to mitigate the poor UI, with my own wares, launched my career as a UI designer. Started as an enthusiastic technophile, left a bitter techno skeptic.
Just one small example. One research team funded by the Army Corps of Engineers were tasked with determining why their mechanical engineers using CADD were no more productive or accurate than when using manual drafting boards. The major finding was that whereas manually drafting had developed effective strategies over the millennia, eg work from left to right (for right handed drafters), CADD software had poor, missing, inconsistent strategies, eg matrix of verbs & nouns showed holes in the UI coverage.
To the best of my knowledge, these considerations are still unaddressed.
Here's the first hit for that research. It's been forever since I've thought about this stuff, so don't know for certain if anyone built on top of these findings.
FWIW, I was on Bentley Systems' user steering committee for MicroStation at the time. While one of their junior software devs was doing some really cool new stuff, the rest of their hierarchy fully embodied the belligerent arrogance of the willfully ignorant.
Alas you really only see those lovely 3D renders upon large projects in which the developer wants to woo in investors and sway local planning with how lovely it looks.
Do they map out the soundscape of the area from the impact or even the lighting levels for the building and cast shadows upon the area - heck no. These models are always upon lovely sunny days, and hey - look at those trees in the picture - ain't we nice people to do that kinda level.
But yes, consumer level would love that, for me - things like sound proofing and light impact from the seasonal variations would be more use for me. Be nice to know that the kitchen window wouldn't get much sun and then only in the summer as it's facing North West and will only catch the glancing sun in the summer first thing and last thing. Or indeed the factor that the prevailing wind will mean that as the kitchen window opens at an angle to the west, which is the same as the prevailing wind - seeing all your kitchen smells permeate the entire property and a penchant to set of the smoke alarm if you ever do any toast or worse - blow out the gas cooker hobs and all that entails. Small details, but they all add up.
But really erks me is the level of sound proofing, sure they bung in some loft insulation and double glazing and say that's cool. But most properties these days in out modern times have no/little code to cover noise pollution and as you view properties during the day, have little idea how they are at night. Heck I live in a flat and if I place an ear against any outer wall, I can clearly hear the downstairs neibours as the property was built without any cavity filling and the owner refuses to install any as their architect says it's not needed and how it is the standard way. Yet happily ignores that you can literally follow a standard level of conversation from neibours. Because there are no real standards about noise pollution. Moral being - never buy or rent a flat when they just put up a concrete frame, breeze blocked the inner leaf with brick outer leaf as for noise travel/pollution - they are horrible.
Indeed in my case the heating system (hot water radiator affair) has pipes going into the wall (all nice and hidden), but if say the downstairs neibours hit or bump into their radiator - the sound resonates up the cavity and has sympathy with the pipe and radiator - making it sound like my radiator was hit or knocked. A most unnerving and freaky experience, but at least one in which I worked out why my radiators at night made odd knocking sounds as if somebody hit them.
So whilst a VR walkthrough would be great, for me, better soundproof and frequency propagation of sound mapped per room along with seasonal lighting levels, would be way more useful and whilst for changes, the whole VR aspect would be great. It is and will be mostly used for initial selling and utilised by investor types who may well never even see the property they brought.
Real crux is that to submit plans/changes to your local planning - none of them demand or need a 3D model or walkthru and with that, as they don't have to do it, that has prevailed to the stage that you just don't get it.
But great opportunities to shake up the market here.
I'm working on that process now. Digital solutions are failing me, so I'm building and revising the floorplans and structure in Lego. It doesn't give a 3d walkthrough either, but is so much faster than anything I've found on a computer.
You mean an industry where you make something once and sell it infinitely vs an industry where you build a specific thing and have to make all your money during design and construction that one time?
Do small margins necessarily mean there is no chance for disruption? I think structural engineering and architecture both have voids that tech can fill. Alot of the work is tedious, repetitive, and begging to be automated.
The Finite Element Analysis software is widely available. Constructability, judgement, client relations, dealing with budget, proposals, and rfis, etc are totally different.
I think you've picked some creative fields but haven't actually identified voids to fill. What are you offering that S-Frame or Revit isn't? What drudge work in particular do you think you can tackle that no one else has?
If you can identify some specific voids to fill, go ahead. but I think those voids you identified are likely your unfamiliarity with the industry.
But if you've got something, go for it. Just research first, because right now it sounds pretty nebulous.
Unity is in a continual state of building out "shiny new toys"; libraries and modules and sibling software that work as intended for 1 or 2 release cycles before breaking and being discarded. I've seen this time and again with Unity and I think it will hurt their stock going forward as they try to convince more industries to use their tools.
Just in the past two years I've seen them develop, hype, and then quietly drop: their VR editor, AR preview tool, Octane Renderer, Substance integration, UNet, ML integration.
It's obvious they have an incredibly difficult time managing their tools when their platform is fundamentally shattered between multiple releases, versions, render pipelines, and now DOTS vs normal workflow. And it's basically impossible for other companies to manage Unity SDKs with a ground that moves that much.
And it's basically impossible for other companies to manage Unity SDKs with a ground that moves that much.
Many successful video game companies use Unity to make their games. It's basically impossible to use except compared to the alternatives, which are just as impossible or more so.
It's pretty impossible to make an sdk or plugin that's runs on every version of Unity. I haven't seen it, anyway.
Studios have the same issue. They pick a version to ship with and work around whatever known bugs there are. Upgrading is a huge event and it's never painless.
Only because they keep it on the version of the engine they started on. Just ask any game developer using Unity about whether they tried to update the engine mid-project and how that went if they did ;p
In my perspective that is a good sign -- they are constantly trying new ways to grow, and dynamically reallocating their resources to markets where the product features gain traction.
IMHO it's unrealistic to expect long-term commitment for maintenance / compatibility from ${ANY} new software, unless there's a 5-year enterprise contract with a SLA. Maintenance is difficult work, and only rewarding if the software provides real-world usefulness.
Bought Unity on sale with plans to use it later. Returned to a prototype two years later and both the product and licensed changed. No way to use the original license or version since their old registration flow broke. So it's upgrade... or crack it?
There is current a war between UE4/Epic and Unity in the AEC market (Architecture, Engineering and Construction.) You need to understand this in that context.
The first move happened with Autodesk releasing Stringray and tying it tightly to 3DS Max. Stingray was not super popular (from what I hear) but its business motivations were real -- there was real-time visualization needs in AEC. UE4 and Unity then started to build out in this area.
I haven't actually seen numbers to back up that there is a viable market here, but there is belief there is a market by all involved parties (Autodesk with Stingray, Epic and Unity.)
Where did the author get the info that this is open source? I haven't seen Unity state anywhere that this tool we be open source.
I could see them giving source code to people who pay for some enterprise license, but I doubt it will be open source. At least not the part that extracts data from BIM software.
Nobody seems to be bothered by this kind of argument when it comes to music publishing industry being disrupted by new technology, why is this any different?
The world is not here to ensure your business model can continue uninterrupted in perpetuity. The world moves forwards (well, mostly forwards), and it is up to you to adapt your business model accordingly.
Unlike others here (based on the downvotes), I don't think your concerns are entirely unfounded. Google is a classic example of a company that scorches the earth with their freeware releases - remember Kiko?
But ultimately, whatever is best for the consumer needs to win out. If freeware knocks out commercial competitors then stagnates (like Internet Explorer), that's bad, and analogous to dumping which is not legal.
If it's just better product and continues to be, like Google Calendar, the world is simply better off.
Right, same goes for google. Remember folks, free software and free products means you are the product. And in the process other companies and tools are being put of business while the masses cheers for yet another free tool, forgetting these are made by monopolies that use them as marketing.
You're confusing google (an advertising company) for unity (a product store).
They're not trying to acquire users for marketing, they're trying to acquire users into their ecosystem and bring more dependence on their tools, their asset store products, and make the unity ecosystem worth more.
It's a land grab. Ditto for VFX industry that they and Epic are trying to get into.
Hopefully it's not the same kind of land grab, where a large company will "nick" successful products off of their asset store and release them "free of charge".
Pretty much all commercial archviz oriented 3D rendering software (where I've spent the last 10 years) continues to feel the heat from such efforts, including Blender and all the support it gets these days from AMD and Nvidia.
Don't understand why you're getting so harshly downvoted.
I dont really understand it either. Whats is happening is that some giants give out “free” software, kill their competition and then monetise in other ways (personal data, or cross selling other products), while in some cases they make use of open source products to sell services. Either way the only win is a lousy 9-5 job for programmers who write the freeware, or loss of jobs for those developed the code commercially. Basically open source for non mission critical software is digging one’s own grave.
Didn't vote but I simply don't see how it is a bad thing when free software competes with commercial software. Something that used to cost money can now be done for free, sounds great to me.
Basically you first design, for look and feel, then you calculate technical details, then you build. Or rather the architect does all three, but you would normally want to "chip in" during the design phase - change the layout, move things around, etc. That's where things get bloody awful.
I wasn't expecting full VR tours of the place (though it is doable [1]), but I did expect to be able to walk through the place in 3D and see how it will look like in the winter morning and summer afternoon. Hell no. What are you smoking?
Floor plans. Planimetric facades. Roll your eyes to the ceiling and imagine. 3D renders that looks like from the early 00s take hours to produce, so they aren't done often. That's _still_ renders, mind you. 3D walk-abouts are possible, but they look like butt and they are viewed as the final step of the process, when everything has been done and done.
And this is with firms where principles are in their 40s and generally computer literate. It just seems that the industry is very conservative with its tooling choices AND these tools are also ridiculously expensive, further stifling any desire to switch. So they are stuck with using technical design software like ArchiCAD (good for laying out piping, calculating stress, air flow, etc.) for visual design. And the resulting process is sooooo slow, small changes and iterations take hours if not days to complete, so it takes weeks to converge to a general design of the house.
Painful, painful process. Ugh. Caveat emptor.
[1] https://www.benoitdereau.com/