Hello Klas from a fellow Swede (I presume)! First of all, thank you for your excellent tutorials! I'm about to kick off a small start-up providing drone services. Therefore I've just configured a few backend server blades for client frontend, live video streaming, photogrammetry and hopefully a GIS workflow. I'm an absolute newbie with both photogrammetry and GIS-work so I have the work cut out for me... 😛 I've been working quite a lot with IT infrastructure, EE, scientific work and such though. As a learning exercise I'm now attempting to make a georeferenced 3D map for something called 'ATAK' (Android Tactical Assault Kit) or 'Android Team Awareness Kit). It's basically a Moving Map application. My first trouble is to find the correct tool chain for my goal. Today I'm sitting here with ngnix, MySQL, AliceVision Meshroom, Blender, MeshLab, qGIS and WebODM configured and running. No add-ons so far. The servers are a bit old-ish but should be adequate up to a certain point. I'm handling +300 48 Mpix photos as input to AliceVision Meshroom. But there's the pain threshold. My first experiment is a huge (1.5 x 1 km) quarry which I have successfully built a few 3D meshes from, in four different resolutions, to see how much difference it makes and if the overhead in processing time is really worth it (not). I've worked with datasets containing 350 images in 5, 12, 20 and 48 Mpix. 48 Mpix took a full 12 hours to process into a 3D mesh. Mostly due to too little RAM in that blade. I'll try to double it eventually. Then I cleaned up the .obj mesh somewhat in Blender, exported it to .ply and imported it as a new layer in qGIS. -Then I had to search for some tutorials - and found your videos! I couldn't even get a 3D viewport to show my mesh - but its onscreen anyway and so far I'm proud after only two days doing this! 😄 Now I'm wondering if my toolchain is adequate or even correct for my intended purpose? 🤔 My first goal is to produce a map, useable in an application called 'ATAK' ('Android Team Awareness Kit' or 'Android Tactical Assault Kit' depending on who's using it). The map format can be OSM (OpenStreetMap) or a number of other formats. I'm guessing I can convert the result from my tools to any format with other tools... I'm basically looking to get a georeferenced 3D output from drone photography. Both visual and infrared (thermal). Later point clouds from LiDAR scans will be used too, but one thing at a time I guess... My 'input parameters' for my work pipeline and toolchain are FOSS and Linux for the time being. FOSS might be negotiable at a later time but not Linux. I've got two questions for anyone knowing what I'm talking about: 1: Will above mentioned toolchain be adequate for what I'm looking to do; create a useable, georeferenced 3D map for common map applications? 2: Is there a good opensource tool (Linux) with which I can make area and volumetric measurements on my map or a 3D model created with above mentioned tools? This need is for another use, not for 'ATAK'. I'm looking for a simple viewer with ability to do these measurements. Since this will be client side I'm looking for Linux, Windows and iOS applications to view and measure these maps / models. Not necessarily FOSS but preferably. Commercial pay-ware as an absolute last resort. If I'm missing something essential I'm extremely thankful for suggestions! That is; suggestions about what tools I need to configure and learn. The steps necessary I will try to learn myself at this point. Therefore your videos looks like a pot of gold right now! Best regards from Northern Norway!
@@KlasKarlsson Thank you. I'm trying hard to do that right now - but I'm trying to get it running with CUDA support (using Ubuntu 24.04 LTS and Docker-Desktop). 😛 It ran for a bit (without CUDA support) but then stopped. So I'm more focused at getting the --gpu variety running. Might install a non-virtuslised Docker machine to try it with CUDA support since this seems to be the way to go process timewise... Looks like this is a troublesome venture... 🙄
I've tried all of those and I keep getting "not recognised file type". Ply uploads but doesn't show on the map, and when I click zoom to layer all I get is a blue screen... I'm using reality capture to create and export the maps/models
Happy New Year! and thanks for the New Years Gift!! its reignited my interest in the whole GIS/BIM area! Thanks again!!
You are awsome !!! Is very useful see you work with 3D in QGIS. Keep going ! (You have a very human way to explain things)
Amazing content as always
Hello Klas from a fellow Swede (I presume)!
First of all, thank you for your excellent tutorials!
I'm about to kick off a small start-up providing drone services. Therefore I've just configured a few backend server blades for client frontend, live video streaming, photogrammetry and hopefully a GIS workflow.
I'm an absolute newbie with both photogrammetry and GIS-work so I have the work cut out for me... 😛
I've been working quite a lot with IT infrastructure, EE, scientific work and such though.
As a learning exercise I'm now attempting to make a georeferenced 3D map for something called 'ATAK' (Android Tactical Assault Kit) or 'Android Team Awareness Kit). It's basically a Moving Map application.
My first trouble is to find the correct tool chain for my goal. Today I'm sitting here with ngnix, MySQL, AliceVision Meshroom, Blender, MeshLab, qGIS and WebODM configured and running. No add-ons so far.
The servers are a bit old-ish but should be adequate up to a certain point. I'm handling +300 48 Mpix photos as input to AliceVision Meshroom. But there's the pain threshold.
My first experiment is a huge (1.5 x 1 km) quarry which I have successfully built a few 3D meshes from, in four different resolutions, to see how much difference it makes and if the overhead in processing time is really worth it (not). I've worked with datasets containing 350 images in 5, 12, 20 and 48 Mpix. 48 Mpix took a full 12 hours to process into a 3D mesh. Mostly due to too little RAM in that blade. I'll try to double it eventually.
Then I cleaned up the .obj mesh somewhat in Blender, exported it to .ply and imported it as a new layer in qGIS. -Then I had to search for some tutorials - and found your videos! I couldn't even get a 3D viewport to show my mesh - but its onscreen anyway and so far I'm proud after only two days doing this! 😄
Now I'm wondering if my toolchain is adequate or even correct for my intended purpose? 🤔
My first goal is to produce a map, useable in an application called 'ATAK' ('Android Team Awareness Kit' or 'Android Tactical Assault Kit' depending on who's using it). The map format can be OSM (OpenStreetMap) or a number of other formats. I'm guessing I can convert the result from my tools to any format with other tools...
I'm basically looking to get a georeferenced 3D output from drone photography. Both visual and infrared (thermal). Later point clouds from LiDAR scans will be used too, but one thing at a time I guess...
My 'input parameters' for my work pipeline and toolchain are FOSS and Linux for the time being. FOSS might be negotiable at a later time but not Linux.
I've got two questions for anyone knowing what I'm talking about:
1: Will above mentioned toolchain be adequate for what I'm looking to do; create a useable, georeferenced 3D map for common map applications?
2: Is there a good opensource tool (Linux) with which I can make area and volumetric measurements on my map or a 3D model created with above mentioned tools? This need is for another use, not for 'ATAK'. I'm looking for a simple viewer with ability to do these measurements. Since this will be client side I'm looking for Linux, Windows and iOS applications to view and measure these maps / models. Not necessarily FOSS but preferably. Commercial pay-ware as an absolute last resort.
If I'm missing something essential I'm extremely thankful for suggestions! That is; suggestions about what tools I need to configure and learn.
The steps necessary I will try to learn myself at this point. Therefore your videos looks like a pot of gold right now!
Best regards from Northern Norway!
It sounds like you should take a look at WebODM (Open Drone Map) - ruclips.net/video/LK5bN38VqIU/видео.html
@@KlasKarlsson Thank you. I'm trying hard to do that right now - but I'm trying to get it running with CUDA support (using Ubuntu 24.04 LTS and Docker-Desktop). 😛
It ran for a bit (without CUDA support) but then stopped. So I'm more focused at getting the --gpu variety running. Might install a non-virtuslised Docker machine to try it with CUDA support since this seems to be the way to go process timewise...
Looks like this is a troublesome venture... 🙄
Amazing work, I will try to do this also
Hi!
I would like to ask where is this data from?
thank you
can i import a dae into qgis?
docs.qgis.org: "supported formats include wavefront .obj, .glTF and .fbx."
I've tried all of those and I keep getting "not recognised file type". Ply uploads but doesn't show on the map, and when I click zoom to layer all I get is a blue screen... I'm using reality capture to create and export the maps/models