Are meshes making point-cloud animation pointless?


After a very busy first quarter, I finally had time to make some much-needed new marketing materials. This had me spending too much of the last month looking for the latest on rendering point clouds for images and animations. I found a lot of articles and postings circa 2012-2014, but not much in the last twelve months. It made curious about a few things: Am I the only one still making animations of point clouds? Am I the only one that doesn’t have this figured out already?! Are we using meshes for visualization instead, due to the sudden ubiquity of UAVs?

Spending 50 hours on rendering over the last 30 days, I’ve had a bit too much time to consider things. I offer the following so that others may not have to spend their time repeating my mistakes—in hopes that they may pay it back by offering some other insights in the comments!

point cloud of campus

  • Autodesk 3DS changes – I can safely say that 90% of my time in 3DS has been spent trying to render point clouds and CAD/Revit models. Its learning curve is legendary, and I only dive in when I have days to devote to it. This time around, I hit an additional hurdle as Mental Ray, the preferred renderer for point cloud data, is no longer available. NVidia quit developing it and Autodesk purchased an alternate rendering engine called “Arnold” that is now considered the de facto renderer for point clouds. Since Arnold is newly acquired, there’s precious little documentation (even less of it related to point clouds) and a grand total of one tutorial on I was able to get a few pics rendered, but large data sets locked up or crashed 3DS every time. The few stills looked good but I simply don’t have the time to keep going without additional documentation or spending time with someone that has it down. V-ray is still an option but it does require a separate license as a plug-in.

    Even if I had the time, I’m not sure it’s a worthy investment. The wholesale changes in the ReCap team at Autodesk, along with the rendering engine changes, and the advancement of UAV collected meshes, have me wondering if I’m chasing a dying solution.

  • Stagnant solutions – There are a few standbys that I pulled out to get the job done. Leica Geosystems Cyclone, Autodesk Navisworks, and Cloud Compare all have rendering engines that are very simple to use. However, they are remarkably unchanged from their initial releases years ago. Cyclone still renders frames before giving the cloud time to fully load, and Navisworks offers very little in the way of good looking clouds unless they are RGB colored (texture mapped) from images. Navisworks is the most advanced due to the ability to nest animated objects within an animation but it still lacks some of the basic functionality needed to render good animations. Namely, all of these fail to have an editable timeline to correct less than desirable movements that are often auto-generated between keyframes. This makes for a LOT of trial and error to maintain a constant speed of movement and focal point with one’s camera.

Cyclone 360

  • Pointools – You know what did have a fully editable timeline along with some great animation wizard presets? Pointools. I still consider this the gold standard of point cloud animation engines. Its integration into the world of Bentley seems to have killed it as an option for a lot of people. However, it still lives. After a few weeks I punted to Pointools and was able to render point cloud animations of 180-240 seconds in the time it took me to animate a single frame in 3DS.
  • Sketchfab as a replacement? – I’ve developed several clients that are more than happy to get a point cloud via ReCap. However, I’ve got several others that do not have any sort of CAD and live in a world of PDFs and web apps. Sketchfab’s concept of being the YouTube for 3D models is proving a reliable replacement for animations when it comes to some of these clients. However, there are a few caveats. First of all, there are limitations to the size of a file that can be uploaded and getting the largest available subscription plan can get expensive – quickly. This size limitation results in users degrading the quality of their point clouds and/or models which, on some level, defeats the purpose when your intent is to show off your work. Secondly, you are no longer in control of the focal point. More than once I’ve watched people spin around happy with the 3D novelty but totally miss the asset or area of concern that drove you to send them the link in the first place.

A ContextCapture mesh

  • Are meshes taking over? – And now we come to what may be the biggest point, is all of this time wasted shoehorning scanning technology towards a use case that has been surpassed by other technologies? Assuming that we are only showing existing conditions as opposed to design intent, Unmanned Aerial Vehicles (UAVs) have now become an interesting option. When you start adding up the time to export your cloud to an animation suite, create an animation, and compiling in a video authoring suite, your budgeted time can quickly equal what it would cost you to create the same video with a UAV! This is especially true if you operate the UAV in-house. Additionally, the advancements in processing meshes from imagery seem to be outpacing point cloud processing of late and every animation suite can handle OBJ files. Since they are more common, you will also find far more educational resources when trying to learn better ways of rendering with meshes.

I’ve always enjoyed rendering animations to show off my work and to communicate the abilities of reality capture technologies to others. However, the extent to which the available processes of doing so have remained static over the past decade is astounding. I’d argue that nothing else in the reality capture space has remained as unchanged!

So, I bring it to you dear readers. Are you rendering these sorts of animations? If so, what are you using as your tools? If not, what has replaced the point cloud animation in your arsenal?


About Author

Sam Billingsley

"Confessions of a Hired Gun" is just that, the true tales spun by a guy who's been in the field, scanning for a living, as long as anyone in the industry. What works and what doesn't work? What does the client want and what should the client want? This is a place where you'll find advice and commiseration if you're in the scanning business, and a place where you can learn about best practices and what to expect from your scanning provider if you're an engineering firm or asset owner/operator.


Subscribe to our weekly enewsletter, delivering news and market information for professionals involved in 3D imaging technology.

You may unsubscribe from our mailing list at any time. Diversified Communications | 121 Free Street, Portland, ME 04101 | +1 207-842-5500


  1. Avatar
    Toby Terpstra on

    I appreciated your article and hearing about your efforts to continue using point clouds. My company still uses them in accident reconstruction for analysis and I do not see that going away. While we have methods for surfacing and do use these on occasion, I personally find that often a dense point cloud works well to not only take necessary measurements, but also convey what is needed visually. I am not sure where this will leave us for tools with the apparent lack of development in this area, but currently we are using mental-ray in 3DS 2017 for creating animations that incorporate point clouds. We have a number of 2018 licenses, but due to the lack of an obvious rendering solution, we simply do not use 2018. I guess 2019 will be out shortly and we will see if Autodesk offers any new rendering options for point clouds. Obviously we cannot delay upgrading to newer versions indefinitely. I am glad that Arnold supports point clouds, but I am not convinced that it will be the best solution for us moving forward. As with many areas in the fields effected by constantly changing technologies and applications, we are not concerned with what the software tools were designed to do, only what they can do, and how we can use them. Thanks for the straight-forward article, and in my mind the on target/ relevant topic choice.

  2. Avatar

    Hello Sam, You are exactly right and this is why Cintoo will be introducing Cintoo Cloud at SPAR 3D in 3 weeks from now. The core technology behind Cintoo Cloud allows to make your terrestrial laser scans cloud-compatible (so BIM-compatible): this is a point cloud-to-surface (mesh) technology that not only reduces the size of your projects by 20 to 30 times before upload, but keeps the project structure and greatly enhances the way you access the reality data in browser after upload. As a matter of fact, surfaces can be streamed with LODs, viewed with back face culling,… on any device. The good news is that this transformation can be inversed (i.e. surfaces-to-structured point clouds) so when you download back the reality data from Cintoo Cloud, you’ll find the original point cloud with its structure (scan locations and depth maps) because this is the data you need in Revit, Edgewise, Kubit and many other Scan-to-BIM desktop tools. More to come at SPAR 3D. More information on: Cheers.

  3. Avatar

    Sam you need to check out Arena4D Data Studio from Veesus. When it comes animating point clouds, meshes and CAD you won’t find anything faster or easier to setup. The software has a large user base with police forces and forensics teams from around the world. These guys need quick deliverables without the hassle factor you describe in your article. We are now seeing other sectors from defence, energy and manufacturing starting to adopt it as a means of controlling and disseminating large point cloud datasets within their organisation and supply chain.

  4. Avatar

    I should point out here that Pointscene is used similarly as Sketchfab to showcase point clouds of drone mapping and laser scanning projects. No need to render point cloud videos, as you can provide your customers with the actual data as interactive 3D web scenes, linked or embedded (publicly/privately) and with no actual limitations.

  5. Avatar
    jollyJamDodger on

    3ds can’t handle the large data set’s – I don’t expect Arnold’s is the issue. Modo or Clarisse are ok with large datasets.

    Coming from the VFX industry we always end up with geo (environments are manually modelled from points), but that’s to do with A) noise in the point clouds B) surfacing/texturing(although megaScans is an answer to that now) C) needing to rig and parent group in instances to animation etc..

    IMO Point clouds are the future, just not todays future.

  6. Avatar
    Sam Billingsley on

    First of all, thanks to everyone for the suggested solution. Second of all, thanks for actually reading the article! The comments here are spot on versus the comments I’ve received on social media which have been much more a debate about the “pointless point cloud” title of the article (thanks, Sean…).
    After I get a chance to try some of the other solutions offered I’ll post a follow up here in the comments section.

    • Avatar
      Sean Higgins, SPAR 3D Editor on

      Mea culpa. I was trying to stir the pot on this one and I might have been a little TOO successful. Please forward all hatemail to the usual address.

      • Avatar

        Please show all hatemail – – this post is very good, I don’t agree with all of it but the message is there.

        The point cloud is dead – long live the point cloud (just in a different interface?)

  7. Avatar

    I, like yourself, am trying to figure out a new pipeline for rendering point clouds in 3dsMax now that Mental Ray is no more. In working with our software reseller they did come across this article from SolidAngle (makers of Arnold) on how to render point clouds with the Arnold renderer…

    I’ve only done a few tests to date but I was able to get it to render a point cloud from within 3dsMax. Might be worth a look see.

  8. Avatar

    In it’s day pointools was an excellent tool, sadly development largely stopped on the product after being acquired by Bentley. Thankfully the guys from Veesus with Arena4D have stepped into the market filling the gap that pointools left.

  9. Avatar

    Great article, and I van only underline your conclusions. My only successful attempts have been using Pointools and with Geoverse. An example of what comes out of Geoverse:
    However, for me, both are to expensive if used for point cloud analysis/viewing/browsing. For that we use Pointscene. Really easy uploading, streaming, visualization and tracing/vectorizing options!

  10. Avatar

    Hi Sam,
    Here are my comments submitted originally via LinkedIn. See you at SPAR!

    We started over 100 years ago with 2d hand drawn plans. This progressed to CAD drawings in the 1960’s, then meshes in the 1970’s with the advent of high speed, high volume inspection based on CMM and fringe projection inspection systems (among other techniques). Laser scanning in the late 1990’s to now has provided us with highly accurate reality measurements, complete with real and/or false colour representation.

    Meshes have their place, pointclouds have their place, animation fly-throughs have their place. Most important is to choose the correct tools for the correct application. Making blanket statements like “pointless” (having little or no sense, use, or purpose) without clear context does not help our understanding of the tools to properly represent the project to the appropriate audience.

    On the other hand, a pointless pointcloud is truly pointless, which means that you would not even be able to render a mesh since there are no measured points to connect!

Leave A Reply

© Diversified Communications. All rights reserved.