Thank you so much, I got it working, but now How would I unrap something like these railings?
Why do you want to unwrap that?
Got it from the warehouse, and doesnât seem to work with Flowify
Just explode until you have loose geometry then group everything. Unwrapping is for texturing, what flowify makes is bend geometry not unwrap.
I understand thanks
No, you didnât and neither did he.
This is what I said:
And this is what i meant.
You donât want to unwrap the deforming model.
You want to unwrap the target surface in order to achieve the flat grid properly.
In this way:
- it will always work
- it will always have the right size
- it will always have proper polygon density/distribution
I agree but what its showing me now is that the quad pattern is stretching the end of the grid so I need to find a way to show railings in those constraints,
unless you know of another way?
thanks again for any input
If you are using Unwrap UV Grid to create the grid, then it should not distort at all (see my example where I intentionally showed a non uniform grid to demonstrate exactly that).
Are you using it? Or are you still trying to impose an arbitrary grid?
Please share the file otherwise is difficult to tell what you are doing wrong.
Or at least run the âunsmooth quadsâ command on the curved surface, so that I can see the wireframe of the target mesh.
As I suspected, you didnât do what I said.
If you keep using that auto-grid feature, youâll keep projecting from an even grid (the auto-grid, where all the cells are the same size) to an uneven grid (your curved surface, which has quads of multiple different size).
This will always cause distortion/stretching.
In this situation you desperately need to use the Unwrap UV grid feature.
This (I already said it, but Iâll say it again) will flatten the grid preserving the same âunevenâ polygon distribution of the curved surface.
Next step: you need to fix those horrible zig-zagging curves.
I recommend that you take a look into Beziere Splines (or Fredo Splines which is even better) to achieve smooth splines and then use Curviloft or Vertext Tools (select the spline and shift + drag up to extrude) in order to create smooth nice looking surfaces which cause less visual artifacts.
By curiosity, would it work with an approach where:
- you create the railing along a curve in the horizontal plane (say via Shape Bender)
- you then shear the railing vertically along the stair (say, via FredoScale)
2 Likes
I am very thankful that you took the time to explain this and I will continue to lean although my time is limited and on top of that as I have reported and waiting to hear back -
my Sketchup 2026 is constantly crashing while Iâm trying to work. see pic
To begin with I know the stairs are not done right, Iâm a visualization guy, and although the stairs are not critical in the overall project, Iâve not had a challenge like this before. When I got the blueprints I could see that there wasnât enough room to properly make the curve with 7.5 resizers, so that eliminated using some stair plugins as I did try Shape Bender, 1001 Bit tools, Artisan Stairs and a few others. So then I used SU native tools with Fredo Scale to get the different stairs to fit, each side is a little different. I faked it by putting the steps together by hand, one at a time, so they are not equidistant from each other making it hard to make the curve properly. So then tried to use the fredo spline , (which I do have the Fredo 6 tools), but om trying all these new plugins and watching all the videos its difficult because if you miss one setting or accidently click on something you have to start over. Perhaps thatâs why Iâve been struggling with Flowify and Quadface tools. From what youâve been telling me, Iâve been unwrapping the UV Grid and then somehow keep using the autogrid feature. Being new with these tools, I seem to get lost in remembering the different settings, and exactly what they can do or donât do while trying to finish other projects. All I can do is keep trying and keep learning. I am a senior citizen but my love for working with all the technology keeps me going.
Peace
I took a time to check the model it wasn´t hard to make the flowify work.
Floify test 5.skp (5.5 MB)
This can be clever for simple stuff and the result will be pretty much the same for this kind of geometry.
If you only needs to deform in vertical direction of course Planar Shear hands down wins against Flowify.
In this specific case, in order to prepare the model, you need more or less the same number of steps.
For absolute beginners Shape Bender will be a lot easier to learn, which could be an advantage for simple stuff.
On the other hand, this approach comes with a few limitations:
- It only works on X axis. If you got geometry laying on Y axis, like in this case, you need to rotate everything (or work on roatated copies), whereas Flowify can work with the geometry âin placeâ
- It only allows for deformation happening in a single direction (whereas Flowify allows to work with geometry going in all crazy different directions)
- I doesnât support quads/subdivision/UV-unwrapping workflow
- It has significantly worse performances on hi-poly geometry (see the video)
This is not easier than what I proposed.
It requires quite a few more manual steps, which can exponentially grow in case of more complex surfaces, and require quite a lot of guessing to judge how much subdivisions you will need to add and where (and this will of course leave the door open for errors) as opposed to the unwrap UV grid approach which is 2-click and almost universal.
Your method also takes a substantial performance hit and generate heavier meshes as you can see the video.
Hereâs a detailed comparison of the 3 proposed methods, with a few bonus tricks and examples.
I hope it can be useful to better understand a few things.
3 Likes
Thank you so much for your time and effort in explaining this in several different ways. Especially Fredo6, who put together a 35-minute video demonstrating 3 different ways to handle this. I was especially fascinated by the video, seeing some of the more advanced work he has done.
For me, this has been a learning experience that I can build on, and thatâs all I could ever ask for. In most cases, learning comes in small bits at a time, and along the way, rarely all at once, and only after repeated trial and error, the learning accumulates. Like learning to play an instrument.
In the end, I used Flowify for the hand railing and Shape Bender for the main railing parts. In addition to SU native tools, when possible. Because of my time constraints, this may end up good enough to show quickly in a video, from a distance, while viewing areas of the very large house. I will work with the results shown below for now. Itâs far from perfect, but I do hope to move on from here.
Thank you Thank you
picture below.
It was easy for me, your method is easier for you, thatâs the beauty of sketchup.
Incidentally, I am not the author of the video, but Panixia. And not the author of the plugin, but Caul.
Thanks to Panixia for showing the 2 approaches in detail. I was surprised by the calculation time of Shape Bender vs. Flowify however. There may be a faster method since Shape Bender is based on a curve deformation onlyâŚ.
1 Like
I have to admit I was surprised too.
I expected it to be a tad slower (because itâs quite a lot older than Flowify and also because of the 2-step geometry preview/generation process), but definitely not THAT MUCH slower.
I guess itâs based on a more recursive algorithm or something like that, but being that they are not opens source, I doubt they can be analyzed.
I was 100% sure that the Flowify simple grid method would have been way faster than the complex grid method because, well⌠number of grid cells/number of mesh vertices to calculate of course.
The visual result would always be the same (if not more accurate) with less grid cells as long as they are spaced with the exact same density/proportions (because it just interpolates the vertices positions based on the distance from the corners of each cell).
Wait, wait.. I was wrong. They are not encripted, so the complexity can be evaluated (see below).
Gemini 3 Pro code analysis
Performance Analysis: Flowify vs. Shape Bender
A deep dive into the Ruby source code of both plugins reveals why Flowify is significantly more efficient than Shape Bender when handling complex meshes. The difference lies in algorithmic complexity and the way they interact with the SketchUp API.
1. Spatial Hashing vs. Linear Search
- Flowify: Uses a custom Spatial Hashing system (
VertexHash.rb,FaceHash.rb). It divides the 3D space into a grid (voxels). When looking for a vertex or intersection, it checks only the local âboxâ ($O(1)$ or $O(K)$ complexity). - Shape Bender: Relies on Linear Iteration. To find where a vertex belongs along the curve, it often loops through segment data sequentially ($O(N)$ complexity). As the vertex count increases, the calculation time grows exponentially.
2. Geometry Generation: Bulk vs. Incremental
- Flowify: Utilizes the
Geom::PolygonMeshobject and thefill_from_meshmethod. This sends the entire geometry to SketchUp in a single C++ call, bypassing the overhead of Ruby for every face creation. - Shape Bender: Frequently manipulates existing entities or adds faces one by one. This triggers SketchUpâs internal âcollision/mergingâ logic repeatedly, which is a major bottleneck for the CPU.
3. Topology Management: Projections vs. Physical Slicing
- Shape Bender: To ensure the mesh bends correctly, it physically âslicesâ the group using
intersect_withoperations for every segment of the curve. These Boolean operations are notoriously slow in SketchUp. - Flowify: Uses a Parametric UV Projection (
QuadGrid.rb). Instead of slicing the mesh, it maps the original vertices onto a virtual grid. This is pure linear algebraâmathematically intensive but computationally much faster than geometric intersections.
4. Language Overhead
- Flowify minimizes Rubyâs workload by calculating transformation matrices in memory and applying them in batches.
- Shape Bender performs many âon-the-flyâ calculations (vectors, weights, and transformations) inside Ruby loops for every single vertex, which is significantly slower than Flowifyâs âprepare then executeâ approach.
Summary: Shape Bender is a âbrute-forceâ procedural tool that forces SketchUp to do heavy lifting via Booleans. Flowify is an âengineâ that uses advanced data structures (Hash Maps) to minimize API calls and maximize throughput.
2 Likes
Interesting analysis. I think Gemini is a little confused regarding point 3. If you want to bend or warp something you need to cut/slice it. In this regard there is no difference between Flowify and ShapeBender. I suppose the confusion stems from the fact that the intersection code in ShapeBender is intermingled with the topological computations - Flowify has a distinct pipeline where intersection happens at the start.
In the example on youtube the time difference is the result of the very dense grid. The time complexity of Flowify is dominated by intersect and if you have a dense grid it will take a lot longer.
The good news is that the dense grid is not necessary. You only need a grid with vertical lines (like the first example in the video) if the source geometry is prepared properly. Flowify does not like inner face loops in the source geometry, if they are removed the simpler grid can be used. Here is a short description from another thread.
1 Like