xy reference points without a z-value

[garynobles]

Hi, I am trying to align chunks using markers, but my reference markers are a 10m grid without z information.
I try to put in only xy values as a reference point but photoscan wont let me do it without a z value.
I put 0 in as the z value but obviously photoscan flattened my landscape.
Is there a dummy value I can put in for an unknown z value or another way around this?
Thanks
Gary

[mwillis]

Gary, is this UAV data or something else?

[garynobles]

Hi, its a low level survey (camera on tripod) about 1500 photos in total of a Bronze Age cemetery in Greece.
We had a site 10m grid so we have xy coordinates for each grid square corner, but no height information.
I'm now aligning the various chunks. Is there anyway to get Photoscan to null the z coordinate? any NA value?

[stihl]

I'd suggest not to use the markers but instead use a scale bar.
Simply put a marker at the position where it should (a corner of this 10m grid)  and then put another marker that's also in the grid but 10m further.
Then select both markers and create a scale bar out of it and make it 10m long and then optimize using several of these scale bars. I'd create two that are perpendicular to each other so X and Y is scaled correctly. In the reference settings menu use local coordinates.

[mwillis]

Gary, good to see another archaeologist using Photoscan.  I would use Google Earth to zoom into your project area and then find the elevations as best you can for the grid corners.  It won't be precise but it may be good enough to get the model to resolve and to get the chunks to align.

[dcm39]

I don't think this is possible in Photoscan at the moment, but I would really like to see it implemented!

This is also very frequently a problem if one has only handheld GPS to reference a model - in this case the vertical accuracy of the GPS is very much lower than the horizontal accuracy. Mathematically, the problem is solvable with the elevation of one point and only xy locations for all the rest, so it would be great to see it implemented.

Using google earth can work, but it is a very rough and ready solution - this uses the underlying SRTM dataset for elevations, typically accurate to ~15m, so you can end up with significant tilts in your model. Photoscan will trade off the fit in x-y to fit this elevation better unfortunately.

[sakaic]

Hi Gary,

I do something similar and my process is similar to what stihl recommended. I only use scale bars to help with the precision of alignment and the entire model is free floating in the 3d space. To align the model to earth reference I only insert the longitude, latitude and altitude of 3 points in the model. I also then relax the error of the GCP. At that point, you can play with the altitude values of the points to ensure the "flatness" of the area.

[garynobles]

Thank you for your responses on this, I've nearly finished aligning the chuncks.

 stihl your idea seems the best, is there any online info on how to do this, I looked in the photoscan manual but information is vague.

mwillis, yes there are a fair few archaeologist working with photogrammetry, some using it better than others. As for using google earth for elevation points, I would be strongly against it, Google Earth, or any similr app, is extremely inaccurate, the georefereced images can have great spatial errors, in some cases things have been 100m+ off.

dcm39, we seem to be in the same boat! The GPS guy took z coordinates of my gcps but discarded the ones for the site grid, so I have a few known height points (10) and 30 known xy points without a z, surely this could be something photoscan could resolve? e.g. what is the elevation a point xy if z is not specified?

sakaic, I'm wanting a full 3D model, I will be creating an orthophoto for 2D GIS but I will also explore the 3D GIS possibilities.

Thank you all, if anyone else has a solution I am eager to hear

Gary



 

[stihl]

Hi Gary,

Unfortunately the documentation of Photoscan is limited as you've found out yourself.

If you have absolute measurements of identifiable objects on your dig site you can use these for scaling your project.

For instance if you know how large something is on the dig site by having measured it with a measuring tape you can use this to scale it.

You can do this by simply adding two markers on the object you know the size of. You need to mark these GCP's (ground control points, aka markers) on all the images on which they are found.
After that you can select the two markers in the marker panel on the left side of the photoscan menu. Select both and right click and choose "Create scale bar"
A new scale bar will appear underneath the marker panel which will say by default point1_point2. You double click the row for Distance and enter the measured distance. For instance 1 meter.

For a good result I'd suggest having _at least_ two scalebars. One for the X axis and one for the Y axis, meaning one scalebar being perpendicular to the other one. This ensures that both axis are properly scaled and nothing gets deformed.

After that check the scale bar accuracy in the reference settings menu (the menu where you set the coordinate sytem (use local coordinates)). By default this value is set at 1mm which is good.
Then use the "Optimize Cameras" button to scale the entire project based on your given scalebars.

Let me know how this works out for you.