Grid material and Ellipsoid primitive #601

emackey · 2013-03-27T22:04:12Z

This is my first attempt at creating a Cesium material.

This branch now includes another merged branch with changes to the EllipsoidPrimative to make it appear double-sided (to show the far wall of the ellipsoid even when outside).

There are some known depth issues where ellipsoids intersect, but these existed prior to this branch and may eventually be addressed by #607.

This branch used to include logic for adding additional grid lines. This code has been reverted, as it made overlapping ellipsoids look too crowded. It is still available for inspection on other branches.

pjcozzi · 2013-03-28T15:24:43Z

Cool. I won't be able to check this out until next week, but @bagnell is welcome to review and merge this sooner if you want it to make b15. At a quick glance, the code looks OK.

Also, update CHANGES.md and the Fabric Wiki.

emackey · 2013-03-28T15:38:59Z

No rush. Before this goes in I could use some advice on how to get the Volume demo to show both the front and back sides of the ellipsoids. Currently when the camera is outside an EllipsoidPrimitive, it only sees the outside wall (facing the camera), but when the camera flies inside, it sees the inside wall. It's as if the backface culling changes when the camera goes inside, or maybe the depth buffer is blocking the farther wall. Is there any way to fix this?

pjcozzi · 2013-03-28T16:04:21Z

In some cases, this will actually be useful. We can also imagine different materials for the front and back. For example, the front has thick lines and the back has thin lines. I coded all of this in Insight3D, but with the fixed function. It is much easier with GPU ray casting.

Anyway, the issue is we only take the closest ray intersection, not both. So if the closest one has an alpha of zero, we never check the back. We could do both and even refraction if we want. This will require a change to the ellipsoid itself, which we should do (at least initially) in a separate pull request.

Also fixed an issue with the final alpha value.

…bagnell

…bagnell's desk. Also commented out picking in the Volume demo because its bugs are driving me completely insane.

…soid Make Ellipsoids double-sided.

Conflicts: Apps/Sandcastle/gallery/Volumes.html Source/Shaders/EllipsoidFS.glsl

mramato · 2013-03-29T13:17:13Z

Source/Shaders/Materials/GridMaterial.glsl

+uniform vec4 gridColor;
+uniform float holeAlpha;
+uniform vec2 lineCount;
+uniform vec2 lineThickness;


What unit is lineThickness in? I want them to be pixels and for the thickness to be uniform in screenspace, just like polylines are. That doesn't look like what's happening now though. Also, why call it lineThickness when we use width everywhere else?

The units are different depending whether your GPU supports GL_OES_standard_derivatives. Without derivatives, the unit is a percentage, where 0.0 is no line and 1.0 is all line and no hole. With derivatives, the units are somewhere in the vicinity of tens-of-pixels, a unit scale chosen to make available some reasonable default values that work pretty well in both environments. I'm open to change of course, but I'm going to wait for @pjcozzi to come back from GDC and have a chance to review this.

As I said in person, these units have changed to be more pixel-like, and the default is now 1.0.

…ipsoids.

…ial.

emackey · 2013-04-01T21:59:07Z

This is almost ready for merge. Didn't do anything with @mramato's comments on material uniform names yet, @pjcozzi or @bagnell any comments on that? I'll take a look at some of our other materials and see if we have any kind of standard names for things.

emackey · 2013-04-02T19:40:52Z

OK, I think this is ready for final review.

mramato · 2013-04-02T19:41:56Z

Specs?

pjcozzi · 2013-04-02T19:49:29Z

@emackey will look at this tomorrow.

emackey · 2013-04-02T19:54:17Z

Spec added for Grid material. The EllipsoidPrimitive gets no new parameters or tests now that the multi-res grid is reverted.

mramato · 2013-04-02T20:41:39Z

Apps/Sandcastle/gallery/Materials.html

@@ -310,6 +310,11 @@
        primitive.material = new Cesium.Material.fromType(scene.getContext(), 'Grass');
    }

+    function applyGridMaterial(primitive, scene) {
+        Sandcastle.declare(applyGridMaterial); // For highlighting in Sandcastle.
+        primitive.material = new Cesium.Material.fromType(scene.getContext(), 'Grid');


It would be nice to actually set all the uniform values here, so that people can see what they can manipulate and play with it in place.

True, but in a previous note we discussed upgrading this demo for auto-discovery of all uniforms on all materials. This demo in its current form contains many materials with only their default uniforms.

Well, my reply to that would be to add uniform settings for all of the materials then, this demo overall is kind of useless without them (from a purely coding standpoint). That being said, if you don't have the bandwidth for it, we can just punt to a later date; no big deal.

mramato · 2013-04-02T20:46:45Z

Might a user want to have the cell color be different than the grid color? Why not let them set cellColor instead of cellAlpha?
It looks like if I set cellAlpha to 1.0 but set gridColor to have an alpha of 0.5, the result is not what I expect. The alpha of the gridColor seems to affect cellAlpha.

mramato · 2013-04-02T20:47:23Z

I realize my comments are all design/usability related. I'm just trying to determine what actual requirements for customers may be.

emackey · 2013-04-02T21:02:33Z

Perhaps gridColor is poorly named, it is the overall color and alpha for the entire GridMaterial, not just the lines themselves. I deliberately don't want to allow the cells to be a separate color. Reducing the alpha of the overall material makes the whole thing translucent (so the cells are affected by this as well as cellAlpha).

mramato · 2013-04-02T21:08:08Z

Then I would rename gridColor to just color, since that's what we use for every other material. cellAlpha is probably fine for a name, but we need to be clear in the doc that it's in addition to the overall color alpha.

mramato · 2013-04-02T22:33:09Z

All tests pass and Sandcastle demo works fine. This is fine with me assuming @pjcozzi is satisfied with the glsl, I'll let him merge when he's happy.

emackey · 2013-04-03T02:27:14Z

Updated Fabric wiki page.

pjcozzi · 2013-04-03T19:57:24Z

Apps/Sandcastle/gallery/Volumes.html

@@ -46,7 +46,8 @@
        e.center = ellipsoid.cartographicToCartesian(
            Cesium.Cartographic.fromDegrees(-75.0, 40.0, 500000.0));
        e.radii = new Cesium.Cartesian3(500000.0, 500000.0, 500000.0);
-        e.material = Cesium.Material.fromType(undefined, Cesium.Material.RimLightingType);
+        e.material = Cesium.Material.fromType(undefined, Cesium.Material.GridType);


Can we just add a new volume to show the grid material? I want to keep these examples to show the rim lighting and stripe material.

pjcozzi · 2013-04-03T20:18:58Z

@emackey did you test this on mobile?

pjcozzi · 2013-04-03T20:21:02Z

On AMD, there are some artifacts due to the partial derivatives (note the black where the ellipsoid overlaps the globe). Not sure how hard they will be to fix. I'm OK merging without them as this is a standard problem with partial derivatives in my experience.

pjcozzi · 2013-04-03T20:21:31Z

@emackey just those few comments. This is good otherwise. Thanks! Materials are always welcome.

emackey · 2013-04-03T21:30:33Z

Just tested Grid material on Motorola Xoom. Works great, derivatives are enabled and all.

emackey · 2013-04-03T21:38:10Z

I'll live with the derivative artifacts for now. One option might be to detect the tangent points on the ellipsoid (or places on the material where the normal is 90 degrees from the camera) and color them separately. But that's for another day. This is ready for merge.

pjcozzi · 2013-04-03T21:44:14Z

Merging.

Grid material and Ellipsoid primitive

emackey added 3 commits March 27, 2013 17:51

Add a new grid material, my first attempt at a Cesium Material.

e565e00

Add new grid material to Materials demo in Sandcastle.

73333ca

Update some wording.

83ec82a

emackey and others added 9 commits March 28, 2013 13:32

Make Ellipsoids double-sided.

4bb8a39

Messing with ellipsoid distances.

c06cf3c

Pull out the main computation into a function that can be called twice.

7c57228

Also fixed an issue with the final alpha value.

Stole some code from this Virtual Globe book I found laying around on @…

88359dd

…bagnell's desk. Also commented out picking in the Volume demo because its bugs are driving me completely insane.

Merge pull request #603 from AnalyticalGraphicsInc/double-sided-ellip…

da980b2

…soid Make Ellipsoids double-sided.

Merge remote-tracking branch 'origin/master' into grid-material

71a1e74

Update default thickness.

79df12c

Merge branch 'grid-material' into ed-test-ellipsoid-dist

9de519b

Conflicts: Apps/Sandcastle/gallery/Volumes.html Source/Shaders/EllipsoidFS.glsl

Work in progress

3486f42

mramato reviewed Mar 29, 2013
View reviewed changes

emackey added 9 commits March 30, 2013 16:20

Make grid lines and other 2D textures not look "bendy" on Ellipsoids.

24a6280

Rename variable.

08ab9d8

Merge branch 'grid-material' into ed-test-ellipsoid-dist

29b7f75

Set up a uniform and JS variable for maximum texture repeating on Ell…

ebf80f1

…ipsoids.

Merge branch 'ed-test-ellipsoid-dist' into grid-material

529a495

Trying to make "lineThickness" values behave like pixel units.

5a674c6

Adjust default values.

6c05660

Merge remote-tracking branch 'origin/master' into grid-material

eb792e5

Revert Volumes demo, but keep new ellipsoid size, and keep grid mater…

dbfa74f

…ial.

emackey added 2 commits April 2, 2013 15:22

Rename holeAlpha to cellAlpha.

57dd229

Merge remote-tracking branch 'origin/master' into grid-material

82f667e

Added unit test for Grid material.

0d6c4d4

Merge remote-tracking branch 'origin/master' into grid-material

1ba4519

mramato reviewed Apr 2, 2013
View reviewed changes

Rename gridColor to color.

ef000b0

Update CHANGES.md

6e7937b

pjcozzi reviewed Apr 3, 2013
View reviewed changes

emackey added 2 commits April 3, 2013 16:33

Merge remote-tracking branch 'origin/master' into grid-material

cf11d97

Completely revert Volumes demo.

8f804fa

pjcozzi added a commit that referenced this pull request Apr 3, 2013

Merge pull request #601 from AnalyticalGraphicsInc/grid-material

1c3c8aa

Grid material and Ellipsoid primitive

pjcozzi merged commit 1c3c8aa into master Apr 3, 2013

pjcozzi deleted the grid-material branch April 3, 2013 21:44

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Grid material and Ellipsoid primitive #601

Grid material and Ellipsoid primitive #601

emackey commented Mar 27, 2013

pjcozzi commented Mar 28, 2013

emackey commented Mar 28, 2013

pjcozzi commented Mar 28, 2013

mramato Mar 29, 2013

emackey Mar 29, 2013

emackey Apr 1, 2013

emackey commented Apr 1, 2013

emackey commented Apr 2, 2013

mramato commented Apr 2, 2013

pjcozzi commented Apr 2, 2013

emackey commented Apr 2, 2013

mramato Apr 2, 2013

emackey Apr 2, 2013

mramato Apr 2, 2013

mramato commented Apr 2, 2013

mramato commented Apr 2, 2013

emackey commented Apr 2, 2013

mramato commented Apr 2, 2013

mramato commented Apr 2, 2013

emackey commented Apr 3, 2013

pjcozzi Apr 3, 2013

pjcozzi commented Apr 3, 2013

pjcozzi commented Apr 3, 2013

pjcozzi commented Apr 3, 2013

emackey commented Apr 3, 2013

emackey commented Apr 3, 2013

pjcozzi commented Apr 3, 2013

Grid material and Ellipsoid primitive #601

Grid material and Ellipsoid primitive #601

Conversation

emackey commented Mar 27, 2013

pjcozzi commented Mar 28, 2013

emackey commented Mar 28, 2013

pjcozzi commented Mar 28, 2013

mramato Mar 29, 2013

Choose a reason for hiding this comment

emackey Mar 29, 2013

Choose a reason for hiding this comment

emackey Apr 1, 2013

Choose a reason for hiding this comment

emackey commented Apr 1, 2013

emackey commented Apr 2, 2013

mramato commented Apr 2, 2013

pjcozzi commented Apr 2, 2013

emackey commented Apr 2, 2013

mramato Apr 2, 2013

Choose a reason for hiding this comment

emackey Apr 2, 2013

Choose a reason for hiding this comment

mramato Apr 2, 2013

Choose a reason for hiding this comment

mramato commented Apr 2, 2013

mramato commented Apr 2, 2013

emackey commented Apr 2, 2013

mramato commented Apr 2, 2013

mramato commented Apr 2, 2013

emackey commented Apr 3, 2013

pjcozzi Apr 3, 2013

Choose a reason for hiding this comment

pjcozzi commented Apr 3, 2013

pjcozzi commented Apr 3, 2013

pjcozzi commented Apr 3, 2013

emackey commented Apr 3, 2013

emackey commented Apr 3, 2013

pjcozzi commented Apr 3, 2013