Hey everyone,
I’ve got a project where I’ve added a known distance using control points added to 7-9 images, I’ve inserted the distance as a constraint, but each time I try to use “update” in the alignment pane I get the message:
“Operation failed: No component with valid ground control constraints found”. (see screenshot)
However the points and the distance can be clearly seen in the 3D model, so I have no idea why it keeps telling me this. I’ve also tried realigning all images with the constraint / control points already inserted, but again it doesn’t scale as it should (distance and actual distance differ significantly).
Any suggestions?
Hi Thomas Van Damme
However the points and the distance can be clearly seen in the 3D model, so I have no idea why it keeps telling me this. I’ve also tried realigning all images with the constraint / control points already inserted, but again it doesn’t scale as it should (distance and actual distance differ significantly).
Can you be more specific on what sort of results you get ? some screenshots after the alignment - results would be great.
As for the CPs, are images used in that particular COMPONENT ? click on the CPs and look at the images whether they have values regarding reprojection error…
I added the control points by selecting relevant images (images that contain the scale bar) from the +Images dropdown menu. The alignment was then done using all images, incl the ones with control points. The screenshot includes the px deviations for each point, which are all within bounds (no triangle with exclamation marks next to it).
The one thing I can think of that might be causing this issue is that the points are located within areas that are masked using alpha-masks (images are PNG), but since the location of the CPs is still accurately shown in the 3D model I would think that this shouldn’t create a problem.
Thomas Van Damme wrote:
…The screenshot includes the px deviations for each point, which are all within bounds (no triangle with exclamation marks next to it)…
Just a side note - the yellow triangle signifies something different: Yellow triangle with an exclamation mark.
Hi Thomas Van Damme
The one thing I can think of that might be causing this issue is that the points are located within areas that are masked using alpha-masks (images are PNG), but since the location of the CPs is still accurately shown in the 3D model I would think that this shouldn’t create a problem.
YES, this can be a problem… try to put them outside the mask. 2-3 images per CP are enough…
Ugh. That’s a little bit annoying. The reason I masked it away in the first place is because it contained the scale bar which shouldn’t show in the final model.
Edit: placed the control points somewhere that was not masked, pressed update, and now it works.
Followup question: now that the model is scaled, is it possible to export an orthophoto containing a visual scale bar?
Hi Thomas Van Damme
No, it is not implemented but it should be easy to do, as you already know the pixel dimension, so you can easily set the scale bar say in Photoshop (pshop) etc, if one pixel is 1cm, then selecting 100 pixels means 1 meter, create a selection with precise size (pshop) and then use the guidelines to make it easier to scale the scale bar in the image correctly…
Aha as always thank you Milos, I had never considered scaling using pixel dimensions before, will give that a shot! Not sure where to find the ortho pixel dimensions info but I’m sure if I have a peek around I’ll see it (not at work now so can’t check RC).
On a side-note: if this is relatively easy to implement (based on pixel dimensions) it would be really cool if we could automatically export an ortho at a specific scale (for instance choose scale 1/10, 1/20, 1/50, whatever, and then set dpi at whatever density you need in order to be able to see all details at that scale) with a little scale bar at the bottom or something like that. A lot of clients can’t work with 3D software and just want scaled orthos as outputs, it’s always an extra step having to scale them in software such as Photoshop or Illustrator. The way of generating orthos in RC is really awesome in comparison to other photogrammetry software, so extra ortho flexibility might be a nice area where RC could excell.
Thanks again Milos, using pixel dimensions for creating scale bars worked flawlessly 
For anyone who might find this useful the workflow was as follows:
- build orthophoto’s of your scaled model (max dimensions for JPG export is 40 000 pixels, so set it at max that if you want to use JPG)
- click on the orthophoto in the 1D pane on the left in order to get information on how many units / pixel are in the ortho (in my case this was 0.000366 m / pixel)
- export the orthophoto and open it in Photoshop
- calculate how many pixels long the scale bar in Photoshop should be (I wanted a 1 meter scale bar, so I divide 1 by 0.000366 in order to know the number of pixels that go in 1 meter => 1/0.000366 = 2732.24)
- use the line tool in Photoshop, draw a random straight line, in the top pane of Photoshop set the desired width to the number of pixels you need (so in my case 2732.24), set the height to something that looks good.
- export the resulting image and DONE.