Using only Z in GCP

[Doszpot]

Hi,

I'm using PS to construct underwater scenes.  I have a series of scale features included in each scene and have been using these to create scale bars.  As the images are collected underwater, it is easy to collect the Z data (through depth) however X and Y are near impossible to collect between scale features.  Is it possible to input only Z data for each feature without having X and Y auto-populate to 0?

Thanks

[Alexey Pasumansky]

Hello Doszpot,

I think that you can follow this approach:
- using Rotate Object tool properly orient the reconstructed model in the coordinate system according to the axis shown in the bottom-right corner of the Model view,
- create markers in the Model view in the points of interest (later you can adjust their projections on the individual images in Photo view mode), if markers are already created, then create the new markers next to them and copy the values from Source values tab of the Reference pane for these newly created markers,
- modify Z values for markers according to your measurements,
- use Set Accuracy option for the markers in the Reference pane to define very low accuracy of XY and adequate accuracy for Z, for example: 1000/0.5 would mean that XY accuracy will be 1 km and Z accuracy - 50 cm. Then press Update button on the Reference pane to apply the referencing information (you may need to check on the markers in the Reference pane first).

[Doszpot]

Hi Alexey,

Thanks for your help.  I should probably clarify a few things.

When modelling the underwater seascape, the scene is prone to bend or twist especially in the Z axis, that's why inserting the Z data would be useful.  This means that the Z data would need to be assigned to points prior to creating the model.  Would I be able to give PS the Z data of detected markers prior to building the model in order to prevent this twisting?

Thanks,
Neil

[Alexey Pasumansky]

Hello Neil,

You can input the coordinate information for the markers prior to the alignment operation, then just use some approximate XY values and very low accuracy for them (big values).

[willfig]

Hi Alexey

I'm working on this project with Neil.  We map transects that are about 50 meters long by about 15 meters wide.  But they are not always straight. We have five features spread along the transect that are composed of pairs (or triples) of coded targets.  So we've done as you've suggested.  That is, we run an initial marker detection and then put in the known z-values (depth to the nearest 0.1m, as a negative value) for each target.  As we don't know X and Y we leave them as 0.  I then added "60.0/0.5" for the error tolerance for all of these.  This seemed to work.  Though I note our model error is huge, due to the massive x,y error on the targets.  Based on independent testing I've found that error value scales very nicely with the accuracy and precision of models like we're building.  So I'm wondering if there is any way to have these targets included in the scaling of the model but for the error to not be affected by this process.  Thanks.

Will

[Alexey Pasumansky]

Hello Will,

You can include the known distances to the referencing/scaling procedure - create the scale bar instances between the pairs of markers or cameras, then input the known distance to the Reference pane for the scale bars, check on the scale bars and press Update button (you may also need to adjust the measurement accuracy for the scale bars).

As for the XY coordinates of the markers, after the initial referencing you can input the estimated XY values for the markers instead of all-zeros to reduce the error value. But still keep the similar XY accuracy for them, as the planar coordinates are not properly measured.

[willfig]

Hi Alexey,

Thanks for the response.  I thought inputting the estimated positions would be a solution but I don't know where I get that.  How do I find those for each marker?

Will

[Alexey Pasumansky]

Hello Will,

You can see them in the Reference pane, just switch to the Estimated values tab.

Also you can export them to the text file together with the source values (Export button on the pane's toolbar) and them import back XY from estimated columns and Z from source.

[willfig]

Hi Alexey,

Thanks.

Will

[willfig]

I'm importing a file with all the coordinate values for my markers so I can set the z-values as discussed above.  For the accuracy field I'm using "60/0.25" so that the x&y accuracy will be 60 meters and the z accuracy (depth) will be 0.25 m.  x,y,z fields import without issue however the accuracy gets converted to "240.00" on import.  Presumably its treating that input as an equation.  Am I doing something wrong or is this a bug?  Thanks.

Will

[Alexey Pasumansky]

Hello Will,

If you are importing the accuracy values from the file as well, I can suggest to use separate columns for that. Either 60,60,0.25 or 60,0.25.

In the Import CSV dialog you need to enable Accuracy option in the Columns section and specify the order of the related accuracy columns. If you have single column for X and Y accuracy, you can define the same column number of both.

[willfig]

Hi Alexey,

That worked great!  Thanks.

Will