I was thinking about buying or building a quad copter to get accustomed flying and experimenting with 3D modeling. I see you can buy very basic programs that will create 3D models from "geo-tagged" or "geo-referenced" digital photos for less than $200, and there may be some open source / freeware as well. These are not the expensive orthophoto/DTM building programs, they just appear to make a 3D model.
It seems that the key here is that if you can "tag" your photos with the GPS coordinates and the camera orientation, then importing them into the 3D software goes much smoother. I see there are some cameras that have GPS built in for "geo-tagging" with coordinates, but I'm not sure if any camera is made that can detect the "tilt" or "compass orientation" of the camera (and I would think that pointing straight down might confuse the compass orientation). Does anyone have a suggestion on a decent camera for a "hobbyist" that will work well with modeling software?
I fly these things, building a 61" wing now that will be an automated topo ship. I don't know much about the cheaper software packages.The most important thing for the camera is the sensor and lens, not positional info. You will orient your pics and eventually camera position using ground control targets and features. You need good focused pixels, because the software will be assigning them NEZ values as it computes the model. Look up Agisoft ,it's the best software for this out there.
As far as what cameras are good, it depends on how much weight you want to lift. At first I'd start smallish with a 14mp p&s with decent optics. You want a wider focal length, maybe 18-24mm.
The sensefly ebee uses a Canon Elph for instance :
https://www.sensefly.com/fileadmin/user_upload/images/1Camera-comparison.pdf
That being said, I think the future will be RTN in the sky and automatic model building in real time in the ground.
I think as the cost of sensors comes down and the rest of the electronics around the sensor get smaller, these things are going to get into accuracy levels that will outperform ground crews.
> I fly these things, building a 61" wing now that will be an automated topo ship. I don't know much about the cheaper software packages.The most important thing for the camera is the sensor and lens, not positional info. You will orient your pics and eventually camera position using ground control targets and features. You need good focused pixels, because the software will be assigning them NEZ values as it computes the model. Look up Agisoft ,it's the best software for this out there.
First, let me say I have no experience in photogrammetry, outside of a couple of decent classes in college, so I'm pretty much starting from scratch here, with nothing more than the theoretical stuff I can remember from 20 years ago ...
Agisoft was one of the programs I was looking at. They off a $179 "Lite" version that says it includes "Aerial and close-range triangulation, Point cloud generation, 3D model generation and texture mapping".
http://www.agisoft.ru/products/photoscan
I assume this is a "getting to know you" package with very limited functions, but since this is just a training/research/education type "hobby" at this point, I think getting familiar with a "lite" package, just to get a taste of how it works, is a good stating place.
But just so I understand here, I have to orient the pics manually in the software? I was under the impression that using a camera that can geotag the pics was essential, so that the software can roughly orient the photos before it starts trying to match up the pixels in the overlapping pics?
> As far as what cameras are good, it depends on how much weight you want to lift. At first I'd start smallish with a 14mp p&s with decent optics. You want a wider focal length, maybe 18-24mm.
>
> The sensefly ebee uses a Canon Elph for instance :
>> https://www.sensefly.com/fileadmin/user_upload/images/1Camera-comparison.pdf
>
> That being said, I think the future will be RTN in the sky and automatic model building in real time in the ground.
> I think as the cost of sensors comes down and the rest of the electronics around the sensor get smaller, these things are going to get into accuracy levels that will outperform ground crews.
You don't happen to post at DIYdrones.com, do you? I'd love to see some pics of what you are building.
I have an account but don't post much. I hang out on FPVLab mostly.
Here is my wing :
Total cost is going to be less than 1000 including my ground hardware and a camera. When I'm not on my phone I can break it down for you.
It will be my first topo platform, so all I know so far is what I've read.
Sorry brushing up on the Agisoft process, you are right, having a rough camera position will help with your work flow, which makes sense. It alievates the alignment process which makes picking ground control points much easier.
In my wing APM will be triggering the shutter and logging positions, so I'll be doing that offboard of the camera.
Yes. I just watched the AGISoft demo video and it does appear that they pulled in a log file from the autopilot rather than using the coordinates from a "geotagged" camera, so I guess that makes a GPS enabled camera somewhere between redundant and useless, but it does open the camera options up.
I am also trying to accomplish my first airborne topo platform. I was thinking along the same direction as you, to use the Raw GPS observations from the UBLOX. With the APM I can set way points with aligned paths. I chose to build a stable Quad to do this, but I am interested in your wing set up. I am also lacking the expensive software to process the data though.
What frame did you choose and where will you mount the camera? All I need is a wing and a RC boat with sonar and I will have the complete topo package, hopefully for under $4000. Below is a pic of my new quad build, still under construction. APM control board of course.
You should go to HxGN in June and check out what Leica has to offer....
[flash width=560 height=315]//www.youtube.com/v/5Xsx9n4XSuE?version=3&hl=en_US[/flash]
If the on board GPS is capable of storing a positional log, you could easily populate the geographic position of each image later based upon a time sync.
