Modeling of existing buildings and their damage

Hi,

I will have to build 3D models of several existing and operational buildings. Based on the point cloud from photogrammetric survey, a building model will be created. Then, it will be necessary to include information about damages to the building's facade, such as cracks and fractures, in the model. Therefore I have two questions:

1. Does OpenBuildings Designer have any special features that facilitate building modeling from point clouds? The facade of the building is the most important, and the interior does not need to be modeled.

2. How can I store information about damages in the model in such a way that the exact location of each damage is recorded? Additionally, the information about damages should be exported to IFC along with the entire model.

Parents
  • it would be good if you described the purpose of your work. Who will use it for what?

    My guess is it is for renovation of facades.

    If so, I suggest you create a idealised model of the buildings, 2d or 3d. By idealised I  mean that everything is ortogonal, same sizes, equal  distances etc  - as you would for a new building. It is many times faster and easier to use.

    A crack to the left of the third window will be just that.

    I don't see a scenario where you get real world coordinates of a crack to the mm from a pointcloud into accurate points in a dgn that is later used by some instrument to find excat that point.

    Your goal is to produce a pdf/paper drawing with annotations of categories of damages so someone can specify what to do about them and ad that text to the drawing. Right?

    If you use OBD you can make the mark up of the damages so you can get quantities if that is desired. If not, consider just using a photo and make annotations on it.

    There are instruments that produces a pointcloud with colour photo overlayed. Fantastic results.

    A tip might be to buy a pen tablet (wacom.com) which appears as a new screen to your PC. You can then place a second MicroStation app window on the tablet and use the PLACE FREEHAND SKETCH tool to trace your damaged areas from the pointcloud.

    regards /Thomas Voghera

Reply
  • it would be good if you described the purpose of your work. Who will use it for what?

    My guess is it is for renovation of facades.

    If so, I suggest you create a idealised model of the buildings, 2d or 3d. By idealised I  mean that everything is ortogonal, same sizes, equal  distances etc  - as you would for a new building. It is many times faster and easier to use.

    A crack to the left of the third window will be just that.

    I don't see a scenario where you get real world coordinates of a crack to the mm from a pointcloud into accurate points in a dgn that is later used by some instrument to find excat that point.

    Your goal is to produce a pdf/paper drawing with annotations of categories of damages so someone can specify what to do about them and ad that text to the drawing. Right?

    If you use OBD you can make the mark up of the damages so you can get quantities if that is desired. If not, consider just using a photo and make annotations on it.

    There are instruments that produces a pointcloud with colour photo overlayed. Fantastic results.

    A tip might be to buy a pen tablet (wacom.com) which appears as a new screen to your PC. You can then place a second MicroStation app window on the tablet and use the PLACE FREEHAND SKETCH tool to trace your damaged areas from the pointcloud.

    regards /Thomas Voghera

Children
  • There are instruments that produces a pointcloud with colour photo overlayed. Fantastic results.

    Point clouds with colour photos overlaid have been improving. Leica Truview overlays panoramic photos with pointclouds to allow measuring. This is better than the displaying the RGB info on the points themselves which makes understanding the objects very hard unlless you really up the resolution which sloows Mstn etc down to a crawl.

    Hybrid photo / laser scanners are coming onto the market. Point clouds are easy to convert to meshes and textured with the panoramic photos. And once you do that you can convert to something like FBX that Mstn should be able to import with the applied textures. No need to mess with cumbersome PCs.

    OTOH, photo and video-based 3d capture is advancing very quickly, driven by AI and Nvidia etc. Combined with global positioning, SLAM, Google Maps + Cesium etc etc, this is really making digital twins more and more mainstream... even die-hard Mstn CAD, BIM drivers will be affected at some point.

    Fact: the larger the client or project the more context info needs to be processed. But even at the smaller end of the scale, firms like Matterport are starting to make an impact in the CAD design space. CAD users would say sure but that Matterport file has to be exported out as dxf (small fee) and CAD'd up before you can really do any 'design' (expensive). With AI, you can now auto-segment the PC or mesh model. CC has its Detectors which I suspect will grow.

    This Matterport Genesis demo is instructive. Auto-segmentation enables subtractions (-ve) from the model. Additions (+ve) to a model is a given with CAD. Taken together you have the basic prerequisites for designing within a dynamic hybrid context. Clean up and CAD-ifying and even adding 'BIM' asset classification info to the scan will get a lot easier. All that AI-assisted object / product subsitution is icing on the cake.

    The contextual 'digital twin' dimension also opens up a lot of analytical modeling benefits: lighting (technical not just the CGI stuff), infra-red (heat loss, moisture levels), acoustics, people movement etc etc. Even just the visual aspects only will be immensely useful for asset tracking, condition / performance analytics. The analytics (and surveying) will increasing be where the fee is as more and more 'CAD' fee hours will be automated using AI.

    If photo-overlaid point clouds are 'fantastic'... we ain't seen nothing yet! Stuck out tongue winking eye