At SketchUp, we are interested in knowing how people typically move from 2D floorplans into 3D models. To better understand the different workflows and needs, we’d love to hear from you.
A few questions to guide the discussion:
What types of projects do you typically work on when converting floorplans to 3D?
Residential (single/multi-family homes)
Commercial (office, retail, hospitality)
Renovation/remodeling
New construction
Other (please specify)
What kind of floorplan formats do you initially start with?
Vectorized (exported from CAD software, with linework that can be snapped/measured)
Non-vector PDF or image (scanned sketch, photo, raster image)
Other (please specify)
What file formats do you most often work with?
JPEG/PNG
PDF (vector or raster)
DWG/DXF
SketchUp (.skp)
Other (please specify)
Which details do you expect to see included in a floorplan before it’s useful for 3D?
Dimensions / scale
Wall thickness
Windows and doors
Room labels
Furniture layout
Other (please specify)
What are the biggest challenges and time-consuming parts of your current 2D-to-3D workflow?
Where do you wish the process was faster or more automated?
What tools do you currently use for the conversion?
What level of detail do you need in the original floorplan to make it useful?
How do you handle missing information (dimensions, wall thickness, etc.)?
Your feedback will help us understand real-world workflows and guide us in shaping better tools. Excited to hear your thoughts and thanks in advance!
Just a remark: Your questions have several references to “Vector PDF”. SketchUp doesn’t support them. SketchUp for Mac can import PDF files as raster images, SketchUp for Windows doesn’t support PDF in any form.
Thanks Anssi, I updated that question. Wanted to check what is your initial floorplan data that you begin to work with when you have to convert floorplans to 3D, are they vectorized formats or raster images?
Natively no, but there are plugins that can import vector pdf into sketchup, I’ve used the simlab extension.
Most of the things I model start on sketchup, sometimes I require a topographic survey but nothing else, I’m the designer architect and I don’t use 2D cad then export to sketchup.
For redesign I usually get a dwg file from the contractor or the survey we make, sometime we just get the printed construction documents, in that case we have a desktop with a background light and mount cameras using tripods or use a scanner app if the size is small, then we import those files to sketchup and it doesn’t matter if they’re vector or raster, I always use them as a reference. Using the dwg lines to create the 3D geometry is not the best practice imo, unless you have developed a workflow that doesn’t create a mess with tags.
I avoid importing 2D files for anything other than basic reference, but the information in them can be helpful if you need to take vector measurement checks or identify something by its description e.g. room names. The data in an outsourced 2D DWG file can be unreliable and I would rather model from scratch to known dimensions so that I don’t get caught out by sloppy drawing and overridden dimensions.
New construction.
DWG or PDF.
DWG or PDF.
Building Geometry, Dimensions / scale, Labels.
Jumping to the original drawing for a clearer view of the proposals. (Can’t be fixed by SU)
Transparent background in PDFs. Import DWG annotation.
Open in SU or convert PDFs to DWG vector in an old copy of Archicad.
Same as the original drawing.
Request it from the lead designer or whoever created the source drawing.
Wow, that’s a lot of questions! And answers are not always simple. But here goes in my case:
Residential mostly.
Non vector PDF or (preferably) in PNG format. But if I have access to a DWG/DXF, I would always import that as it should be to scale and much of the conversion work has been done.
DWG/DXF/PNG/PDF in order of preference.
The first three in your list as a minimum.
Drawing over an image file.
It would be nice to have a built-in ability to convert line drawn images to a SU drawing.
Vectorized preferably. DWG into SketchUp, because that’s all there is for vectored art. I’m used to being able to import vectorized PDF into PowerCADD and converting it. In fact, that’s one of my workflows for going from SketchUp or Layout back to PowerCADD: vector based PDF, and yes, it’s accurate enough for dimensioning, but generally DWG is preferred over PDF because it at least has layers. I think PDF’s can have layers, but I’ve never seen one. Raster images in a pinch, and particularly Matchphoto of course with an uncropped photo.
One format that’s probably not on you’re list is the Clipboard. Can you imagine selecting art work in Adobe Illustrator, copying it to the clipboard, switching to SU and then doing a paste and having vectored art imported? At one point, Alfred Scott had that functionality between PowerCADD and Form-Z. No need to even write a file to disk for import.
DWG mostly. I have once turned scanned (raster) sketches into a SU model and applied the scans as a texture to the faces - That can be a pretty nifty effect.
Link:
BBCode
Mostly walls, doors, windows and fixed construction like stairs, balconies, cabinets, plumbing, and ceiling above. For a vectored import, I don’t use any dimensions, or hardly any text (though contour labels are helpful), but do need centerlines. Furniture isn’t much needed for snapping. I do need a PDF reference for every DWG for text and because stuff is often lost in translation. If SU did import text, there would be less need to look over at the PDF version, but imported text is always a mess (fonts, formatting, etc.).
In the very first step, renaming the imported DWG layers to what I need for names in SU is drudgery. It would help me greatly if I had the ability to append a text string to the start of every imported layer name. Adobe Lightroom has a great file naming tool on export that lets you put together smart text and simple ASCII text. It looks like this:
I append a tilde (~) to the name of every imported DWG layer. That character puts all reference drawings into a “neighborhood of tags” that sort to the bottom of the list.
nothing regular, but from what I can judge, recently…
yes. I do projects. the last 3 were a house, a flat and a showroom in a small castle.
ideally, vectorised. but I mostly work on existing buildings, therefore I often have either old scans, or my own dimensioned sketches.
png or tiff if scanned, pdf if provided by a third person. dxf if they have CAD ressources or if I converted a vector pdf.
dimensions / scale. for sure. that includes the walls and the holes in it. labels or furnitures, it depends what I’m supposed to do with it. if vectorised, I don’t mind it if they’re missing, too much info can be a pain.
missing dimensions. it’s always a vital one and you have to extrapolate. also, fully exploded autocad 2000 files, where every symbol, dashed line or hatch is exploded.
and if it’s a dxf, you loose stuff, colour, text… that can be really annoying
the ability to import vector pdf the same way as a dxf. no need to convert in a 3rd party software. also, a better dxf import, less exploded.
scale, then (if image) standard drawing tools, then faceup to go from lines to grouped walls. I’m about to learn / teach a plugin from adebeo, similar to 1001bits, with walls / doors / windows tools. it reminds me of archicad.
walls and holes. because again all I do is spaces. I don’t mind extra details (inside the walls, precise windows and doors, vector dimensions…) so long as they are simple to hide / delete (at least on their tags). at least when you start drawing, too much info can be worse than too little.
well if it’s vector, I might not have the dimension, but I’ll suppose it’s drawn correctly. on a scan / sketch, I’ll either do math and use logic, or measure on the image and round it up or down using logic again. I also use photomatch to infirm or confirm my deductions.
ie. if a door looks 80cm from the side of the wall, and on the scaled sketch I get about 81, and using photomatch I get about 79,5, I’ll go with 80.
.dwg or .png (if I get a .pdf I’ll just screenshot it to a .png)
.las: I’ve been doing a lot of work with point clouds, including those derived from photogrammetry.
it depends on the agreed upon LOD, for some its just accurate dimensions including wall thickness, other times it goes all the way to a furnished/landscaped space
if I get a .dwg, I will usually not import it for use but rather create an image to import. Vectorized data is great, but the importing often breaks things up too much and I’d rather manually model it than spend time cleaning up drawing entities and layers.
I haven’t really thought about it, TBH. When I first think about automation it makes me think about less time in a CAD environment and that’s the fun part about my work.
Eneroth’s “Flatten to Plane” extension is great for placing all imported entities onto the same plane. A lot of time’s I’ll run that really quick just as a “check” after importing.
If I can confirm the scale is uniform and the lines are straight, I don’t need a whole level of detail.
If there is missing information I’ll either state my assumptions and ask the client to confirm, or outright ask them what they would use.
I work mostly on new residential construction and building reuse. The reuse may not be residential.
I prefer to model my floor plans directly in SketchUp and in new residential work I build 3D spatial block models in SketchUp as a framework for the emerging floor plan.
I import 2D DXF/DWG/jpeg or whatever format files for reference when I am working on the reuse of an exiting building. I can’t rely on those files and I rework them in SketchUp.
I often import DXF/DWG survey plans and place them on a discrete tag or tags. Sometimes survey plans are provided in PDF.
As my operating system is Windows, I convert supplied PDFs to jpegs or take a screen shot, and I’ll work with DWG/DXF if available.
The more info the better but I don’t expect anything. The information is unpredictable. Dimensions are good as I usually rework the supplied building information in SketchUp. Site survey information includes dimensions and is reliable but if provided as a PDF I rebuild it as skp.
For building information - rebuilding and verifying the supplied information.
For site survey information - converting 2D spot levels and contours to SketchUp 3D, and joining, cleaning, simplifying and smoothing contours.
PDF to skp automation.
Automation or part automation of the process of converting a 2D site survey into 3D skp.
Tools to use in the reworking process, such as:
a way of identifying an imported building grid and nominating grid lines at x mm spacing,
a way of identifying imported walls and specifying a wall thickness in skp,
site survey tools as mentioned above.
Screen shots, brute force and SketchUp.
Information helps.
I rework the model until it is complete for my purpose, and has integrity. I’ll endeavour to source missing information.
What types of projects do you typically work on when converting floorplans to 3D?
Residential, new construction, commercial - mainly custom timber frame structures / homes
What kind of floorplan formats do you initially start with?
Vector CAD that I convert to DWG, raster CAD I import (Mac), PNG, JPG scanned hand sketches.
What file formats do you most often work with?
JPEG/PNG, PDF (vector and raster), DWG/DXF
Which details do you expect to see included in a floorplan before it’s useful for 3D?
Primarily looking for overall intent and out to out dimensions - I often develop the structural concept as well as wall / roof details so thicknesses are generally developed by me over the structure.
What are the biggest challenges and time-consuming parts of your current 2D-to-3D workflow?
2d conversion and clean up takes time to before I can get started.
Where do you wish the process was faster or more automated?
I would only want speed if accuracy is guaranteed. Many sources of information often have errors in them - sloppy CAD work, fat fingered dimensions that don’t add up, etc.
What tools do you currently use for the conversion?
QCAD Pro (Mac), CST CadNavigator, and Affinity Designer all help me convert files to SKP use.
What level of detail do you need in the original floorplan to make it useful?
Basic overall dimensions. I also like to have several section cuts for overall floor / roof heights / roof pitches.
How do you handle missing information (dimensions, wall thickness, etc.)?
I tend to drive this part of the design (at least for the exterior envelope) - so I can often disregard the drawings and specify what these will be.
