GeodeticGraticule and grid line resolution when zooming in

[surrealwaffle]

I've been attempting to make use of the osgEarth::Util::GeodeticGraticule and I'm having a bit of a hard time understanding how to configure the layer's resolutions. As far as I understand it from the source, the number of grid lines and the camera's extents of the map are used to determine the resolution of the grid lines.

On tests/graticule.earth (after fixing the last resolutions entry), this works well until I have zoomed in away from the poles, where I see fewer and fewer grid lines prior to hitting the limits of resolution. For instance, I can see lines for longitudes are separated by 1 degree by moving the camera over to the next longitude, but I cannot see both longitudes on screen at the same time (unless the camera is near a pole).

Given the configuration in the file, I would expect to see a smaller resolution value chosen and the response in this discussion leans in that direction. Am I misunderstanding how the resolution is chosen, or is this an issue (and, if so, is there some reference material available for the implementation of GeodeticGraticule::getViewExtent())?

Edit:

> .\osgearth_version.exe --caps
[osgearth info] Hello, world.
[osgearth info] [Registry] Note: GDAL_DATA environment variable is not set
[osgearth info] [Capabilities] osgEarth Version:  3.7.3 build 180
[osgearth info] [Capabilities] GDAL Version:      3.12.0
[osgearth info] [Capabilities] OSG Version:       3.6.5
[osgearth info] [Capabilities] OSG GL3 Features:  yes
[osgearth info] [Capabilities] OSG FFP Available: no
[osgearth info] [Capabilities] CPU Cores:         20
[osgearth info] [Capabilities] GL_VENDOR:         NVIDIA Corporation
[osgearth info] [Capabilities] GL_RENDERER:       NVIDIA GeForce RTX 3050 Ti Laptop GPU/PCIe/SSE2
[osgearth info] [Capabilities] GL_VERSION:        3.3.0 NVIDIA 581.32
[osgearth info] [Capabilities] GL CORE Profile:   yes
[osgearth info] Goodbye.

[surrealwaffle]

I think this line may be causing my grief.
zf = horizon.valid() ? horizon->getDistanceToVisibleHorizon() : 1e6;

In my case, the horizon distance will be used to construct the frustum even if the camera is at an altitude of 1000 meters and pointing directly at the surface.

[jasonbeverage]

I think what you are seeing is the height above ellipsoid fading happening where the graticule fades out and disappears once you get close to the ground. Try commenting out this line and replacing it with:

float alpha = antialias;

The fragment shader does break down once you get so close to the ground so you'll see some rendering artifacts, but I'm curious to see if this is more like what you are expecting to see.

[surrealwaffle]

Attached is an image of what I see with the recommended alpha change in place:

The image was taken from osgearth_viewer on the following .earth file:

<map name="osgEarth Graticules">

    <xi:include href="readymap_imagery.xml"/>
            
    <GeodeticGraticule name="Geodetic" open="true">
        <color>#ffff00ff</color>
        <label_color>#ffffffff</label_color>
        <grid_lines>10</grid_lines>
        <resolutions>10 5 2.5 1.0 0.5 0.25 0.125 0.0625 0.03125</resolutions>
        <grid_lines_visible>true</grid_lines_visible>
        <grid_labels_visible>true</grid_labels_visible>
        <edge_labels_visible>true</edge_labels_visible>
    </GeodeticGraticule>
    
    <xi:include href="viewpoints.xml"/>
</map>

The grid resolution chosen here is 1.0, although I would have expected a value smaller (0.25). Zooming in further exaggerates the effect.