Best roof creation practices from plan view (blueprint)

Hello! I was provided with PDF blueprints of several house models of which I need to create architectural 3D renders. The walls doors and windows were simple enough to create, the roof no so much… especially because it is composed of 3 different height levels. I had high hopes for ROOF by TGI but it kept crashing on anything other than a rectangular-based roof.
Then I tried the follow-me tool method and that got me close but after cleaning the extra geometry I ended up with a weird complex section as shown below:

Finally, I was able to get it done by moving up the ridgelines by a calculated distance based on the slope angle and the distance to the edge of the roof. This method required some abstraction but worked fine. Shortly after, I realized that some of the roof faces were not closing up in a single plane so I ended up adding extra lines to close the face. This extra geometry is currently causing me extra trouble when using the offset tool on the roof faces with the intent of creating the shingle ridge caps.

Attaching roof model:
ROOF-TEST.skp (174.9 KB)

Is there a way to force these slightly offset lines to be on the same plane?

Is there a better way of doing this type of roof from a blueprint? I would like to streamline this process as much as possible.


With over a decade of roof design I can say that there is no software or tool out there that will automatically generate all roof geometry perfectly. In my experience, a human will have to get involved sooner or later to assure that roof will work. It looks like you are close to a working solution, now, but need to get that section at the back cleaned up!


Roof model looks fine to me. As you can see below, all edges appear to be on axis. Where is the problem?

You have a lot of hidden edges on the roof panes that should be coplanar but that seems not to be coplanar.

You should not have to do any calculations. From your model, I deduct that your roof has a 6:12 slope. You can enter this when rotating a guide with the measuring protractor.

Can you post the original roof plan, including its slope, although I can figure that from your model?

Attaching original roof plan:

RP.pdf (169.6 KB)

Sorry, I forgot to unhide the extra liens before uploading.

I’m not necessarily looking for a magic tool but would appreciate any tips on roof modeling from exiting roof plans. Right now I’m struggling to add the shingle ridge caps to add more detail and get better-looking renders. I thought a simple follow-me tool along all the ridges would do but the angle of the shape changes from ridge to ridge, I might have to do it individually or at least one shape per angle if that makes sense…

Also, can you elaborate on the issues detected on the back of the roof? I don’t have a picture of the finished house but thought I had the roof figured out. See a picture from a different angle:

That might be a good task for Joint Push Pull plugin. If you can draw the outline the hips and caps on the surface, Joint PP will let you just thicken them a little.


You might want to evaluate the Medeek Truss Extension.

Here is a video that shows some of the features.

1 Like

RT has described my basic method on the hips (JPP)… But I also use ValiArchitect InstantRoofNui for roof features. It can also make the roof shape but I don’t always use it for that part. I notice on your plan that you have different plate heights, and while it may not be necessary for this exercise, the modeler would want to know what the actual differences in the plate heights.

Your last picture is too dark. If you are demonstrating basic shapes, post a picture in monotone. What problems are you talking about? for the intersections of roof forms, particularly of differing plate heights, you have to follow the plan. In the right rear corner the low back roof plane continues up to meet the plane of the high front roof plane to form a secondary ridge. So you draw the edge of the right rear hip up until the end of your edge is coplanar with the center front roof plane. From that point the ridge edge goes back to meet the the front right hip of the central roof form. Simon shows this in his picture,

1 Like

Attaching beam schedule:

Attaching lighter picture

I’m parsing your description of the issue over and over and I still don’t see it. Sounds like it is right… but I also might have designer fatigue…
Here is a top view with the roof model [left], roof plan [right], and plan over model [center]:

Yes. It looks good here. Maybe the description is hard to follow but I don’t know how else to say it. You got the lower ridge (on the right) running as I’d expect. Is there something else bothering you about the design?

When you make any complex 3d form like a roof you have to consider some basic parameters…
Are all of the eaves [gutters] at the same level ?
Are all of the roof slopes the same angle ?
Changing one of these affects the result…

The recent examples assume the eaves-level is not critical.
If so you need to model the roof in discrete parts and intersect them together…


I don’t see any issues myself with the actual shape of the roof, but I couldn’t get some roof faces or slopes to be in a single plane (coplanar conforming lines). See dotted line subdivisions below:

If there are ‘hidden’ line subdivisions then the roof is not constructable in the real world !

It should not be beyond the wit of man to fix this…

Post the nearly-fixed SKP and we can look at it…

My first post has the latest version of the roof: “ROOF-TEST.skp

Side parallel projection views show all lines of a single roof face to be in “the same plane” however some of the faces are not closing up.

Also, look at the picture in my first post (the one with the issue circled in red). That was done with the follow-me tool and there are no hidden lines. The roof is constructible in the real world, at least the section outside the red circle…

That’s due to modeling method. Logically the roof planes should work as shown, You just got some inaccuracy into the model somehow, and You either live with is or start over. If you wish to do a lot more with the model, I’d start over.
You could use groups to keep parts separate until you ready to combine the different elements without messing up the planes.

Different plate heights do require finding a special approach. Maybe there are some other ideas out there. I admit I don’t have a specific method for this situation. I would probably just “model at it" using inferences once I established the pitch.
For regular roofs, your hips are all the same angle so you can use one hip to copy from or use as a inference for another.

Your roof has a bunch of infinitesimal errors that prevent the parts from being coplanar. Here is my version, made first with follow me and then drawing directly in space with the help of inferencing.
ROOF-TEST-2.skp (133.8 KB)