Regression: Linked Toposolids produce inaccurate geometry for large Civil 3D surfaces (>50,000 points)

[Sean.Lawrence]

In Revit 2024 through 2026, the new Linked Toposolids functionality produces only an approximate representation of published Civil 3D topography surfaces when datasets exceed 50,000 points. This is a significant regression from previous Revit versions.

Using the same Civil 3D dataset, we have compared Revit 2023 Linked Topography (Toposurface) and Revit 2026 Linked Toposolids.

Attached imagery clearly demonstrates the geometric discrepancies between the two. The Revit 2023 workflow produced an accurate representation of the source surface, while the current Linked Toposolid implementation does not.

Why this is an issue:

An approximate terrain surface is not acceptable for infrastructure projects such as:

• Bridges
• Marine structures
• Transport infrastructure

The current behaviour introduces design and construction risk, particularly because:

• Users may not realise the surface is approximate
• Levels, set‑out, and clearances may be taken from incorrect geometry

On Current projects, our teams are being forced to:

• Abandon Revit topography workflows entirely, reverting to IFC or DWG‑based workflows. This is a regression in coordination, documentation, and productivity.
• Maintain legacy software versions (Revit 2023) solely to preserve surface accuracy. This is not sustainable long‑term and risks losing AutoDesk support.

Impacts:

• Increased coordination effort between Civil and Revit models
• Slower documentation workflows (loss of reliable contours, labels, view templates, etc.)
• Increased risk to design accuracy and downstream construction outcomes

  Idea to Implement:

• Removing the 50,000 point limitation on Linked Toposolids
• Reinstating the Revit 2023 Linked Topography (Toposurface) functionality in later Revit versions.

[Jon.Brooker]

100% agree with this @Sean.Lawrence. When linked topographies from civil 3D came in, it was a massive workflow improvement, and now we are having to hold onto Revit 2023 and upgrade to keep it working.

Going backwards would be a huge loss for infrastructure projects!

[Susana_Duarte_LMSI]

On top of toposolid behaving as a solid in section. 

We to had to use the create in Revit 2023 and upgrade to have the software behave how we need - and the problem is we then do not have the project references to do the pads... so we have to use gimmics to get that references in revit 2023.

Such a masive waste of time - terrible.

Please revise this - we really need to reach a compromize that works for everyone. And it's not impossible.

[mpowellBKMST]

Things were a *lot* worse before 2026 when the breaklines from C3D surfaces wouldn't be honored regardless of the number of points - but I agree that this limitation definitely needs to be resolved. 

They should at least be clearer with people about this limitation so we can no when to partition up linked surfaces to avoid it (this is something we've been doing to avoid it). 

Personally I'd prefer to see them remove the limit rather than reintroducing linked topographies - but something needs to be done.

[Sean.Lawrence]

Thanks for the comments, it's interesting to hear other people's ideas on the matter.

When I raised this issue recently, the product team explained that the current point limit balances performance with accuracy. They also mentioned that achieving higher accuracy is technically possible within their internal workflows. It would be highly beneficial if Revit allowed unlimited point counts, perhaps via the .ini setting (not capping it at 50k) or an advanced option, to precisely replicate the imported geometry in the source data and enhance reliability. This change would be particularly beneficial for projects with dense point datasets.

The source data in my original post included over 84,000 points. This is a relatively small project, and it resulted in many points being ignored, causing noticeable differences in the final geometry, whereas the 2023 TopoLink produced an exact match. Workarounds like dividing DTMs/TINs into smaller sections are not viable, as they would involve manipulating the received data and adding unnecessary complexity to the workflow not to mention further impacting other processes/dynamo functions down the line.

[pieter_haeck]

This post is spot on! Well-reasoned arguments! 

The Revit 2023 workaround is still the best solution. All other solutions are just an 'approximation' of the desired result.

My idea would be that Revit should be doing one of the 2 following:

• Support both TopoSolids and TopoSurfaces. This way, users can use the type of Topo-object that suits their needs in their particular workflow. Also being able to convert a TopoSurface to a TopoSolid, as it is the case now.
  
  or
  
  
• While linking Topography, Revit could link it as a TopoSurface in the background and then convert it in the same process to a TopoSolid

I personally would prefer the first suggestion, giving the user all flexibility for their specific workflow.

So basicly bringing together the best of both worlds.

[jessy.shao]

Thank you for raising this, @Sean.Lawrence  We would like to take a closer look at your specific case.

Could you please share the following files with us?

1. The source Civil 3D file containing the published surface
2. The Revit 2023 host file (accurate result)
3. The Revit 2026 host file (inaccurate result)

If you prefer not to share publicly, please contact Autodesk Support and reference this thread (ID: 14059408) for direct escalation to the engineering team. And my email: [email protected]. 

We are currently conduct some research related to partial selection, and if it interests you we also would love to invite you to participate in our next sprint review to have a closer look. +

Thank you.

-Jessy