Nice. I'm remodeling a place now. Observations from trying the flow:
Fell off the flow when I had to rasterize our plans, when I tried to add multiple rooms, and again during checkout. PDF support should be a must, ideally multiple room support. Making sure checkout works is also a must.
Path was Navigate to Roometron.com=>create account=>create new project=>go to upload my plans=>couldn't select my plans=>saw site didn't support pdf but did webp=>google "pdf to webp" converter=>found some suspicious sites to upload pdf's to=>went to chatgpt to write a script to convert pdf to webp=>saw the script was converting to png anyways=>went back to the roometron site and saw png was supported=>brew install poppler webp =>`pdftocairo -png plans.pdf page`=>upload page 1=>can't find where to upload the rest of the pages for multiple rooms (also my plans have multiple rooms per page)=>exit out=>seeing as this is a "shown hn" decide I'll try again and push through with just the first page of plans only=>see I need my square footage for the first page=>see its only listed for the floor=>manually measure my square footage for the uploaded page=>convert to square meters=>go to checkout=> enter in my credit card info=>can't scroll down to click "complete purchase" because of a rendering error presumably because the info was autopopulated=>quit the credit card flow=>attempt the paypal flow despite absolutely loathing paypal=>sign into paypal=>receive a payment request for $50 not $17=>think this doesn't look right=>quit the flow=>try to find a different payment option=>exiting the payment flow deletes my project=>completely give up here.
This is why releasing as early as possible to get customer feedback is key. The app clearly gives enough value to pay, but there's a problem in the payment flow that's a showstopper. That's the sort of thing you need to understand before you start scaling.
This is the opposite, fixing payment flow is something can be completely done without customer feedback. It’s q&a work.
You would want to test the response of customers on your product as well as your acquisition strategy, this is something you can’t fully predict, but still can get close if you spend a lot of time in that market
No one is going to bother with your side project if the first chance they attempt to use it, they simply can't. They'll say to themselves "oh well, I guess it's still a work in progress" and move on and forget about it. Posting a Show HN project is like an elevator pitch, you've got to capture the interest of your target audience quickly and be able to deliver at least a viable product. Not supporting PDF isn't a huge issue, that's just a future feature but not even allowing someone to use it is a failure even if the tech is sound.
This thing looks cool and if I was in real estate or apartment management, I'd probably be interested in trying it out. But if I can't, I'm just going to forget about it since I've probably already got a solution for 3D renders. At the very least, the emails are collected so OP can send out a blast saying it works now and hopefully some people give it another shot.
Sorry for making you struggle, I will try to make the order process more straightforward.
Please let me know about the refund, I guess I did message you on gmail.
FYI I also bailed when the price went up from 17$ to 50$ after I had entered my credit card.
Also it was very unclear to me what is the minimum quality of the floor plans on my side (i.e. is what I have good enough for your system?).
I'm Nick. This is Roometron, my first project that I managed to launch by myself while being a web developer and having a full-time job.
I started working on it in 2019, so it took almost 5 years to deliver a beta version now.
It was both a fun and challenging journey. I didn't expect things to go so slowly, but anyway, I'm happy to announce it's finally launched.
What is Roometron?
Roometron is a tool that converts floor plans into 3D apartments. It is VR-ready, highly performant, fast, efficient, and affordable due to its incredible automation.
I'm probably not in the target audience, but still, it looks lovely, I've enjoyed exploring the 3d part.
As a suggestion, a full screen button for the 3d scene would be nice, as I don't have VR, just a regular monitor. And if you're feeling adventurous, you could maybe even add WASD support for greater immersion. But that's a bit of work, as you would probably also have to include a physics lib for dealing with collisions.
Agree - WSAD or just cursor keys would make it really 'walking' - currently it's more like 'click to jump' :)
Also, could you tweak you algorithm to have the geometry deform less at the edges of the viewport? At the moment it's a little extreme and distracting.
I created an account with the intention of kicking the tires, but I have hit wall after wall and I never actually got to the cool 3D part.
If you take one thing from this message, I strongly urge you to make sure that new accounts start with a sample project in the list which people can play with to see what to expect.
In my case, I don't have a floor plan or know my square meters. I went onto Google images and borrowed a plan to upload. The TL;DR is that it seems like I have to pay $33 to see what this place would look like in 3D, and that's when I bounced. I'm curious, but I'm not that curious.
After the sample project, I hope that you will add a 2-3 minute onboarding video to show people how to get started. Getting dropped into any interface with only a vague idea of what is supposed to happen is a guaranteed way to lose 90% of your visitors to unnecessary friction.
Make it stupid simple to get started. Give people a look at the outcome so that they know what to covet.
I didn't really expect such a curiosity and my initial idea is to make this tool beneficial for real estate agencies and builders and only then try to make it useful for a wide audience :) Thanks for you interest and feedback.
Feature request: In the top/overview view I would like to be able to tap on each room and then the pivot point should move to the center of that room. After a double tap or tap on the top view button the pivot point should move back to the apartment center point. This will help zooming easily into each room from above.
Hi Nick, I have nothing more to say but congratulations, the whole project is put together really well, one can see that this thing was crafted with love.
I hope it does well in the market and you can profit off your work in here.
Damn neat, congrats! Some minor feedbacks to get potential users hooked perhaps even more likelier:
Took me a while to hit the "Walk" feature, given that this is perhaps for many the major showcase highlight here. First, having to scroll down on home page to "Try it out", then it defaults to "Fly" mode and the Fly/Walk toggle is kinda down&right, would suggest top&left (unless RTL default in user agent =). And when this default Fly mode first opens up, the apartment box is pretty small (1/5 of horizontal space here), might consider defaulting to the biggest zoom level that will still show it completely in all rotations (that's what the user'll do next).
Nice that the glassy surfaces reflect in Walk mode! Was surprised about lack of specular-highlighting, isn't that fairly cheap compared to reflections? Or maybe planned. Not a dealbreaker on the user side tho I'd guess. But given all the gfx goodies from reflections to ambient occlusion in there, I was a bit curious.
Another thing is that sometimes the canvas goes white with Chromium (Version 126.0.6478.182 (Official Build, ungoogled-chromium) Arch Linux (64-bit)). But the slightest redraw provocation (click-that-actually-moves, or drag-that-rotates) resolves that. (But if you know under what conditions your canvas would fill white, you can look around for what might mistakenly cause such conditions..)
The UI on the Landing page are slightly different from the original one(when you open the viewer in a new tab) it was done that way to fit the LP content.
The Chrome based bug came with the recent Chrome updates and I didn't have time to fix it and migrate all the HTML elements to the Canvas, but I will have to do it I guess
I also think that the walk feature should have wasd and arrow movement. Maybe it should transition more slowly if you use those keys but when I read "walk" then my expectation in a 3d environment is that I can navigate more freely with keys.
I would think that WASD movement would be very unintuitive for most people who don't play first person 3D games (although it would be nice for the minority who do expect it, and it shouldn't be too hard to implement either).
Oh my god same. Until I found this comment I couldn't even figure out how to trigger 3D mode at all, I was just staring at the 2D plan in utmost frustration clicking "3D plan image" over and over. This is a great product hampered by a not great UI right now.
I'm pretty sure that "3d plan image" are the orthographic top-down renders. To me they look very 2d, but I guess technically they are 3d renders (that do everything in their power to not be 3d)
I didn't understand that the icon (logo?) on the Interactive Space section was actually a button and thought it was just screenshots. Maybe a labeled button would be better.
I decided to give it a try, but got discouraged because there's no guidance during the process on what will or won't work. I don't know if the floorplan I have will "work," and I don't understand why it's asking for things like the total area of the floorplan or how that will affect the AI's ability to parse the image and create the rooms. I imagine it's necessary to give a point of reference for the scale of the drawing, but does it also infer things from text on the floorplan or not? Or is the visual language of floorplans standardized across states/provinces/codes to the point where I don't have to worry, as long as it's a "real" floorplan?
It's hard to commit to a purchase when I have no idea what the end result will look like. It would be really helpful to have a gallery of examples showing floorplans and their corresponding 3D results. That way, I could better understand what kinds of input work well and which don’t.
I agree. I have some floor plans of my apartment and I'd love to try but for $50AUD for both floors, I don't know what I'm going to get - or if it's editable after the fact ie adding furniture - or if I'm stuck with what I get on the first go.
The demo is fluid and works well, but I'm not sure I understand what problem this is solving when things like Matterport are already doing this with real images of spaces. It doesn't require floor plans. In many cases floor plans aren't easily available. Polycam and others are helping with automating the creation of 2D and 3D models from real spaces using LIDAR scanning. How is Roometron different from any of the other tools that are on the market?
Matterport are quite expensive, it can be as $500 / year to store few of your 3d walkthroughs. And Im not trying to replace it or something, I just knew that many online floor plan aggregators, construction builders and etc looking for a way to turn their floor plans into 3D or walkthrough videos.
Admittedly I have no idea what an "online floor plan aggregator" is, but I don't see why any of Matterport's commercial property management customers would want to put in the far greater amount of work this requires of them rather than just throwing some money at Matterport to come out and scan it?
We're building our first home with a local construction company down here in Australia. I remembered browsing house designs using a virtual 3d plan that enabled me to walk around the house. Your idea looked very similar. I found one example online and turns out they're using Matterport.
awesome work. The demo is incredibly smooth and I love the disappearing walls
As a renter, I would love to be able to measure distances in the 3d render. 'Will my couch fit here?'; 'How High are the ceilings?'; 'can I fit my bike above the tv?'; 'how far down the hall is my roommate?'; 'can I fit my desk and dresser next to each other?'
Every apartment I've rented started with me taking a Laser Distance Measure with me and making my own floor plan with height measurement as well.
I see the room square footage, but that's honestly less useful to me than the dimensions (again from a renter's perspective)
I was double billed and using your support chat widget resulted in "Opps.. Seems somethig went wrong". Nick, how do I get in touch with you? My user ID is "VvAShB".
I had a similar issue. I was quoted $33 because I put in sq.ft for the area instead of sq.m. When I corrected the area it quoted $17. I paid and it charged me $33.
I just paid for a project, but got charged $33 on Paypal instead of $17, and then it lost my project and no history of the transaction in my transactions on the dashboard.
I tried sending a message to support, and it threw an error, so trying here. I understand shit happens, I am not irate, but I would like to remedy this.
There is https://floorplanner.com
They are used in a variety of situations and have been doing this for over 15 years. Floorplanner is used in high end shops to show arrangements of furniture, on website to showcase apartments etc. that are for sale and a lot of different use cases.
What is the use case of Roometron that you see it does best?
I have not seen any of nice looking 3D apartment viewer yet,
most of the companies offer users to draw apartments yourself and furnish it, thats mostly not considered as a user friendly experience and then they ask to render your work into an image, there are no real 3D to play with.
I want to preface this by making it clear that this is supposed to be constructive feedback. You may want to have GPT-4 give some alternate wording for some lines on the homepage.
"Child bedroom" would likely be referred to as "Children's Bedroom" or "Child's Bedroom" and "Allow to go through the 3d space for an immersive VR experience" might be "Walk-through generated spaces in immersive 3D" for some examples.
I've been working in 3D CAD for about half a decade now, and I would expect smoother copy. This is the only critique I have, because the actual product looks great.
So I’ve used something similar to tour apartments virtually from a 3D camera. One thing I’m noticing when walking through the space is it feels like the perspective is really distorted compared to the actual 3D 360 photos in current apartment tours.
The Birds Eye view is neat though. Curious, have you gained traction with apartment management companies to adopt this?
Funny, I worked on something similar. My first job we needed to build up some fancy 3D alerting system for a client. I found a threejs based project on GitHub (I think it was called BluePrint 3D) and pitched it to my boss, it saved me screaming for help at figuring out how to build the same things in three JS with zero experience, but also saved us hundreds of hours to rebuild the same thing. It looked somewhat like this tool, though I'm sure this ones way more polished.
It too had a 2D editor for 3D, it was cool, but we were just building floorplans and displaying live data on those floor plans, so all the useful design stuff was scrapped for the most part. This looks nicely polished, good job.
It was a painful project due to the client asking for things that were just... well they were insane.
> A neural radiance field (NeRF) is a method based on deep learning for reconstructing a three-dimensional representation of a scene from two-dimensional images. The NeRF model enables downstream applications of novel view synthesis, scene geometry reconstruction, and obtaining the reflectance properties of the scene. Additional scene properties such as camera poses may also be jointly learned. First introduced in 2020,[1] it has since gained significant attention for its potential applications in computer graphics and content creation.[2]
- Business Concept: A b2b business idea from awhile ago expanded: "Adjustable Wire Shelving Emporium"; [business, office supply] client product mix 3D configurator with product placement upsells; from photos of a space to products that would work there. Presumably there's already dimensional calibration with a known-good dimension or two;
- ENH: the tabletop in that photo is n x m, so the rest of the objects are probably about
- ENH: the building was constructed in YYYY in Locality, so the building code there then said that the commercial framing studs should be here and the cables and wiring should be there.
My issue was moreso I had 0 experience with 3D and they wanted me to slap something together with threejs. I didn't even know JavaScript that well at the time. That is a pretty cool project though.
Do you want to see your future? Check out Homestyler [1]
I've used this several years ago to mock up some designs (eventually hired an architect). The tool was still helpful when I transcribed the architect-made plans to the online app, and twiddled around with some details, furniture layouts, etc etc.
It's gotten pretty bloated since, they've got libraries upon libraries of fugly elements (furniture, decorations, architectural details etc) that you can add, but some of the useful core functionality (3d renders, 2d plans with dimensions) still remain.
Love it. Congrats on shipping! Most of the 3D floor plan stuff out there looks pretty bad I found, but yours looks really great. I love a good orthographic camera and disappearing walls like that! Really scratches that itch for me.
What I'd personally love is to have this but with the ability to see inside walls and have layers for, electrical circuits, network cables or even where my studs are. Basically a full model of my house (or as some people call it, a digital twin). This is probably too niche, but heh, just putting it out there.
It seems to me that the greatest risk in a 5 year build out is "obsolete" technology. No technology is really obsolete (you could totally build a startup on classic ASP or jQuery), but I know as standards change, polyfills are no longer needed, libraries become abandoned, etc, it can be tricky.
On a side note, I wish this was around in 2022. While I see it's designed for apartments, I spent a ridiculous amount of time converting our new house dimensions into a 3d model as we were planning out furniture, etc.
If you click -particularly on a room's label, but also happens when clicking somewhere else-, everything disappears except for the labels. If you click again, everything comes back. Sometimes when click-dragging to rotate, this produces an annoying flash -white for an instant then everything's redrawn.
Also some textures seem to not be loading properly. All furniture remains black.
Hm, I wish this could use a smartphone with lidar (Like iPhones/iPads) to scan the environment and create the 3D models. The viewer itself is very impressive and I see it will soon support VR. But I've never seen the viewer as the problem - capturing of the environment in an easy and affordable way - that's what's needed. Right now you have to resort to expensive 3D camera rigs.
I'm CTO of a company called FORMATION. We build an app that uses indoor maps to allow workers in offices, factories, and other workplaces to interact with their workplace via search, QR codes, and just clicking around in the app. We are pretty early stage but a big bottleneck in onboarding new customers is getting decent quality indoor maps. There are a lot of apps and tools for this.
I'd love to see a 2D version of this. 3D is nice but a bit fiddly to interact with. 2D works much better. Especially on mobile. A second point is that the map is a the backdrop for our app and not the main focus. I think this is true for a lot of apps that use maps: the map is not the main feature but merely to context in which you present information.
So far, we've just been winging this. We work with external designers to clean up whatever images we receive to make them a bit nicer and then just georeference the bitmap on top of openstreetmap (via maptiler and maplibre).
If you want to discuss further, feel free tor reach out privately, my handle is globally unique ;-).
The rendering looks blurry on my display. It looks like the devicePixelRatio is not taken into account. If you want to improve the experience for the everyday user (who could have some ultrabook with a 125% to 250% scale factor, depending on model) I'd highly recommend adjusting the canvas resolution according to the devicePixelRatio
It's taken, but there are some restrictions for the high ratios (better to make it blurry than laggy) and there are also a settings button allows to force high resolution
“Time spent” isn’t the best thing to call out first when talking about a project.
Yes, it matters a lot to you but is it the most important thing for anyone else to know?
I’d much rather know what you built + key reason it exists.
I get that the 5 years thing is “just” marketing and a way to farm engagement — which is totally fine. This is just an alternative perspective on how to do that.
The landing page is refreshingly well done. It looks good, loads fast, feels smooth and stable. I also like the price point. It sounds like a lot for a saas, but doing it yourself takes too much time, and hiring someone is an order of magnitude more.
I have the floorplan & measurements of an apartment. I'd like to see how it looks with furniture. Does this allow me to generate multiple furnishing styles until I find one I like? Does it give me links to the product pages of the furniture?
I understand that this is a different use case from what you had in mind. I'm also open to recommendations of other websites.
So would the user be able to specify what furniture to be placed in the different rooms in the floor plan? And how do you source the 3d models? e.g. if I want to have specific IKEA furniture?
I’m wondering because I once spent some time writing a 3d design app that could be used for room planning, and the main challenge was striking the balance between user control (e.g. the user wants the couch to be changed or moved) and automation. Because the user doesn’t want to fiddle too much with the 3d software, as it is the entire premise of building such an app that it alleviates the need for 3d skills, but there is still the need for expressiveness and making changes.
And props for launching! Going from 80% launch-able to 100% is not an easy task, especially when balancing other work. There’s so many small things that add up, even things like privacy policies or ToS’es.
I'm most interested in the AI interior design. Is it using products that are available? Can we set budgets for the designs? Seems like a product in and of itself in that one feature.
This feature is marked as "Coming soon" and not ready yet, but the idea is to feed a room screenshot to AI and to generate a realistic design based on this and on a text prompt.
No constructive feedback, just wanted to say this is really well done. Congrats on the project, I'd be incredibly proud to build something of this quality.
For recent-construction and modern gut-renovation apartments, these simple geometric abstraction renderings are nice and helpful.
For older apartments, such as dominate many older cities or older parts of town, that existing competitor's tool, which includes photograph textures and more "organic" 3D modeling, seems more useful in getting a sense of what the place is actually like.
I just use magicplan (a freemium app) to create floor plans just by scanning with my phone. Of course it won't be as accurate as a proper 3D scanner but prospective renters aren't going to notice. You can of course edit the floor plan afterwards.
Yes, I did think to render the whole site with multiple buildings and even started to work on it, but decided to finish it later and launch the current version faster.
Do you mean that it works with ha-floorplan? If so, can you please explain to me how I'd proceed to make that work because it'd be awesome.
I always wanted to do something like this but the time it takes to get the proper render and plan is just too big of an investment for me.
Is there any way to preview the 3d model created from the floorplan? Some floorplans I have are a bit blurry and I have no idea how this will translate into floorplans so to have to pay $100+ without any guarantee of success is not ideal
The output will be the same quality as you can find on the landing page, it doesn't matter how blurry the original image, but if the original image is not working, you will be asked to upload a better one or simply refunded.
Look cool, I used HomeByMe for similar stuff when planning an extension. It had the benefit of optional offline render for photorealistic output, nowadays you could perhaps export the scene as GLTF and send it to blender.
Almost as large as the early access button or more eye catching, I would've skipped the Try It Out button had I not read someone else's comment. Fantastic site design, amazing demo app and clean 3D for a solo project, much luck.
The messaging around potential use case seen on the landing page is so concise and clear, further, product itself felt interesting and very interactive to me. Wish you good luck with your journey!
A lot of real estate agents pay photographers to do virtual walk-throughs of their properties. You might be able to partner with the photographers and get them to up-sell their customers with your product.
No, no, oh God please nooooo.
People will use this tool for their listings on real-estate classifieds portals like Zillow.
The real estate listed there is never a bespoke design for you and your family. In some locations, there are plenty of affordable homes so you actually can choose one with the layout nearest to your goals.
In the most cases though, affordable homes are rare so people don't really care about the current layout: they will buy any home and remodel it in according to their tastes.
To estimate the remodeling costs, it is better to work with the bare, empty layout plans, not cluttered with furniture and 3D effects, and having all measures specified.
They're already doing it (not with this tool, of course).
At least around here (Spain), when selling shitty (but expensive) apartments, they'll publish "artist renditions of possible remodels" instead of the actual current state of the home.
Absolutely stunning.
From quick play, in first person mode, moving from room to room isn’t as obvious, maybe consider adding a translucent clickable item at the rooms entrance.
This is pretty neat! I’m on mobile right now, but you mentioned that it’s VR ready - does the landing page work with WebXR? I’d love to try it out on my meta quest 3
I have used ArchiCAD to do the same thing before. My question is, what’s the advantage of having it as a web app? Can I, for example, embed it into my site?
Our attempt was providing tools making vrchat-like worlds in Godot Engine. We have many of the mesh construction tooling but we never made to the product.
@streakolay - What's the reason for the checkout price changing at the last minute? I see several people mentioning that they were quoted $17 and then charged $33 (x2) or $50 (x3), but I don't see any replies explaining the reason or confirming that it's a bug.
Surely it's not an intentional bait-and-switch, right? That would be an extremely scummy anti-pattern; the cart price should always match the final checkout price aside from reasonable surcharges like tax and shipping that should be displayed as soon as the needed information is collected.
The fact that the mentioned higher amounts are almost perfectly double or triple the quoted amount makes it sound like some kind of error, like either a server-side bug or a user accidentally adding several copies to their cart without realizing it.
Can you please clarify the reason for the discrepancy?
There are actually nothing hidden or not transparent about the payments,
it says $17 for the areas under 70 sq.m or 750 sq.ft, for the bigger areas the price is up. I'm not quite sure is it a bug there, but I turned off the payment form for now. And I already did refund the guys who messaged me about it, so please contact me if you have a payment problem.
I admire people who are able to really believe in an idea they come up with and spent many years building it. I'm sure OP when you started off, you saw competitor offering, what made the turning point that made you realize and committed, "you know out of a million other ideas, this is what I want to work with".
I have a million ideas on my diary and have yet to build any of them. They aren't truly unique but I see them as a "practice" that when one Mythical day I don't have worried and have ample time, I can take the first step. My current free time is occupied by job hunt and LC..
Also, in term of technologies, what is the core idea/tool that converts 2d drawing into 3d. Is it truly automatic or some core principles behind it, like OCR but for 3d models? I have zero exposure in OpenCV
I'm curious if you've seen Matterport yet? IIRC they had a 3d virtual tour embed that works by capturing apartments using real camera images (vs from floor plan). They seemed pretty common in SF Bay Area.
I love that it offers features like 3D floor plan images, video tours, and AI-powered interior design at an affordable price. This could really enhance the way properties are showcased online. Have you tried their service yet, and how was your experience?
Fell off the flow when I had to rasterize our plans, when I tried to add multiple rooms, and again during checkout. PDF support should be a must, ideally multiple room support. Making sure checkout works is also a must.
Path was Navigate to Roometron.com=>create account=>create new project=>go to upload my plans=>couldn't select my plans=>saw site didn't support pdf but did webp=>google "pdf to webp" converter=>found some suspicious sites to upload pdf's to=>went to chatgpt to write a script to convert pdf to webp=>saw the script was converting to png anyways=>went back to the roometron site and saw png was supported=>brew install poppler webp =>`pdftocairo -png plans.pdf page`=>upload page 1=>can't find where to upload the rest of the pages for multiple rooms (also my plans have multiple rooms per page)=>exit out=>seeing as this is a "shown hn" decide I'll try again and push through with just the first page of plans only=>see I need my square footage for the first page=>see its only listed for the floor=>manually measure my square footage for the uploaded page=>convert to square meters=>go to checkout=> enter in my credit card info=>can't scroll down to click "complete purchase" because of a rendering error presumably because the info was autopopulated=>quit the credit card flow=>attempt the paypal flow despite absolutely loathing paypal=>sign into paypal=>receive a payment request for $50 not $17=>think this doesn't look right=>quit the flow=>try to find a different payment option=>exiting the payment flow deletes my project=>completely give up here.