Polyline Materials #578
Polyline Materials #578
Conversation
…rical to cartesian coordinates to built-in functions.
…h color components of 1.0, but not less than 1.0.
…, and blue components. Add doc and tests.
…lor in the Sandcastle. Tweak vertex shader.
…e polyline Sandcastle example.
|
Performance numbers aren't ideal, but they are tolerable. We'll continue to work on new abstractions for users to optimization specific use cases. |
|
@bagnell can you add the new materials to the Fabric wiki page? |
| * vec4 currentColor = mix(bottomColor, topColor, step(0.5, textureCoordinates.t)); | ||
| * vec4 color = czm_antialias(bottomColor, topColor, currentColor, dist, 0.1); | ||
| */ | ||
| vec4 czm_antialias(vec4 color1, vec4 color2, vec4 currentColor, float dist, float fuzzFactor) |
There was a problem hiding this comment.
Add a test for this, even if it is coarse-grained.
There was a problem hiding this comment.
Can the new grid material use this?
There was a problem hiding this comment.
I think the grid material can use this, but I believe @emackey did his own anti-aliasing by blurring the area at the edges of the grid line.
Conflicts: Source/DynamicScene/DynamicPathVisualizer.js Source/DynamicScene/DynamicPolylineVisualizer.js
|
@pjcozzi The precision problem is fixed in this branch by subdividing the lines and uses only 8 vertex attributes. It is also fixed in the |
|
As far as I know, every mobile GPU under the sun has 16. I know @kring is also convinced that we should treat 16 as the maximum. http://webglstats.com/ backs up this claim statis that 99.9% of devices support 16. |
|
@bagnell how many subdivisions does an access line from a satellite to a facility need? @mramato webglstats.com says 87.8% of mobile have 16 attributes. If we know what the 12.2% is that doesn't (it's mostly Android devices), I am happy to assume we always have 16, especially considering that I expect @bagnell's answer to the above question will be scary. Worse case, we will have two rendering paths. I also pointed @bagnell to these stats from unity, but it sounds like he couldn't determine which GPUs only have 8 attributes. |
|
I should have looked more closely at webglstats, so I apologize for the 99% partial truth. I still think 16 is a reasonable maximum to expect, what are the chances that a device with 8 would really be able to run Cesium anyway (for other reasons, not just graphics chipset)? Anyway, I know I'm not an expert, so I'll leave you guys alone now. |
|
@pjcozzi The problem seems to occur between 90k and 95k meters. so a new position will be added every 90k meters. In simple.czml, the first access line between the ISS and Geoeye 1 at 3/15/2012 10:32:06 is 50 points. |
Conflicts: Source/Scene/Polyline.js
|
@pjcozzi I wasn't able to reproduce the the issue with the |
|
New performance numbers (master vs. polylineMaterials): Static data with the same material:
Static data with the unique materials:
Dynamic data with the same material:
Dynamic data with the unique materials:
|
|
@pjcozzi I think this is ready again. I still have to update the wiki. |
|
Merging. Will write issues for remaining minor problems. Please update the materials wiki and email cesium-dev as there will be a lot of interest in this. |
I don't expect this pull request to be merged for a while, but I wanted to get feedback from @pjcozzi and anyone else who is interested. This branch contains wide polylines and a few new materials like outlines and arrows. Volumetric lines and a few other materials that are specific to polylines are still experimental and in different branches that are not a part of this pull request.
Any CZML polylines/paths will have the default red color material. @mramato is looking into what needs to be done there.