load AAA-exported traces

[jonorthwash]

Related to #81, AAA-exported splines (traces) should be loadable.

One of the problems with this is that AAA can export splines with arbitrary amounts of metadata. The current example AAA data (Qumuq) includes what is probably close to a bare minimum of metadata to match the splines to frames in the ULT files.

[keggsmurph21]

What is AAA?

[jonorthwash]

http://www.articulateinstruments.com/aaa/

[keggsmurph21]

The standard base edition of AAA includes facilities for analysing EPG data. More to follow.

lmao, hopefully it will follow :^)

Thanks, @jonorthwash !

[jonorthwash]

If you click on the AAA tab downloads page, there's a lot more info.

[mr-martian]

I have now, in some sense, gotten data from SPLINES.dat into metadata.json. However, it will take some further experimentation to get the image dimensions right. It's also possible I'm associating them with the wrong frames and I'm wondering if there's a chance it might actually be in polar coordinates.

[jonorthwash]

Two things:

• The frames can be be exported in polar coordinates, but that's not how I chose to export them in this case.
• The image dimensions should be based on the edges of the frame.

[jonorthwash]

This is how tracing in AAA looks (with Qumuq data).

[jonorthwash]

The are spaced at every n° intersect, probably every 1°.

[mr-martian]

@jonorthwash I can't read the screen, could you tell me what file and frame that is?

[jonorthwash]

I'm not sure. You can probably tell the former from the length of the target words in the list and the percent through the list it is, and the latter from the numbers under the image. But it looks like the trace probably isn't right anyway. The point was to show you what it looks like broadly.

I can look at traces and potentially assess what's wrong with them..

[mr-martian]

I think I figured it out.

[mr-martian]

It currently imports all traces as red and ignores anything with a set confidence interval below 50.

It also assumes that all columns in SPLINES.dat past the first 3 are either rectangular coordinates or confidence intervals, which might not be entirely true.

[jonorthwash]

It currently imports all traces as red and ignores anything with a set confidence interval below 50.

It seems it is importing them into the different layers though ("Tongue", "Roof"), so recolouring is straightforward.

A few things I see:

• A black background would be good for the images, instead of white
• The frames in the frames tier aren't lining up with the annotations in the annotation_0 tier.
• The "Read scan line data" should be selected by default if it's the only one, so that ultrasound images are loaded when a file is loaded.
  • In general, I'm thinking this part needs a redesign. I'm not sure what the right approach is, but I'm starting to think we want a menu on top. Then the Data menu would have options for which format of data to read (with unavailable options greyed out), with the current one checked and whichever makes most sense chosen automatically.

[jonorthwash]

It also assumes that all columns in SPLINES.dat past the first 3 are either rectangular coordinates or confidence intervals, which might not be entirely true.

Correct. It should probably be based on substrings "X,Y values" and "Confidence values" in the headers, the former matching 84 columns and the latter matching 42.

[jonorthwash]

• A black background would be good for the images, instead of white

→ #129

• The frames in the frames tier aren't lining up with the annotations in the annotation_0 tier.

→ #127

• The "Read scan line data" should be selected by default if it's the only one, so that ultrasound images are loaded when a file is loaded.

→ #128

It also assumes that all columns in SPLINES.dat past the first 3 are either rectangular coordinates or confidence intervals, which might not be entirely true.

Correct. It should probably be based on substrings "X,Y values" and "Confidence values" in the headers, the former matching 84 columns and the latter matching 42.

If this is what's being done now, then this should be fine. I'm going to close this issue since the basic functionality is working.