Updated aikido_rviz for the ShapeNode refactor

[mkoval]

No description provided.

[mklingen]

This pull request broke color in aikido_rviz btw. All of my meshes are white again :(

[mkoval]

Which type of color are you talking about?

• embedded textures
• embedded vertex colors
• color set on the ShapeNode's VisualAddon
• color set on the aikido_rviz marker

[mklingen]

The color is an embedded material (not a texture or vertex color)

[mkoval]

I'm not sure what that means. Can you post an example?

I can't imagine how these changes could break embedded materials. The only thing that has changed in this pull request is how I handle VisualAddon and ShapeFrameMarker color.

[mklingen]

its worth noting I updated dart as well, though I don't know how that could have caused this either.

[psigen]

Unrelated to this PR: what is the commented ResourceServer doing here?

[psigen]

@mkoval: This might take me a while to test 😓

[mklingen]

Hmm, still not entirely sure why my color is failing. I think I may have an old version of rviz with this bug present:

ros-visualization/rviz#752

Basically, if a marker color is published at all, it completely overrides the embedded material color. I think this PR possibly publishes a mesh color where none was published before. I will try deleting my rviz debs and updating to the latest version to see if it fixes my problem.

EDIT: Bah, nvm upgrading rviz didn't fix the problem.

[mklingen]

Woah, check out the documentation on the flag use_embedded_materials here:

http://wiki.ros.org/rviz/DisplayTypes/Marker

Apparently the mesh marker must have the flag use_embedded_materials set to true, and the color of the marker must be all 0s for it to work.

But apparently people haven't even been able to get that to work either!?

ros-visualization/rviz#933

Rviz is all kinds of messed up. Maybe I'll just switch the ADA model over to using textures instead.

[mkoval]

The interplay between use_embedded_materials and color has changed a half dozen times over the past year. I saw it change three times over the course of the summer while using DART on Robonaut.

I believe the intended behavior is what's listed under the [1.8] block:

Since version [1.8], even when mesh_use_embedded_materials is true, if the marker color is set to anything other than r=0,g=0,b=0,a=0 the marker color and alpha will be used to tint the mesh with the embedded material.

That is the behavior I see on my desktop with the Fuze bottle mesh. Are you sure you're using the latest version of RViz?

[mklingen]

Yes. I will put together a test with some embedded materials (not textures)

On Wed, Apr 6, 2016 at 6:14 PM, Michael Koval [email protected]
wrote:

The interplay between use_embedded_materials and color has changed a half
dozen times over the past year. I saw it change three times over the course
of the summer while using DART on Robonaut.

I believe the intended behavior is what's listed under the [1.8] block:

Since version [1.8], even when mesh_use_embedded_materials is true, if
the marker color is set to anything other than r=0,g=0,b=0,a=0 the
marker color and alpha will be used to tint the mesh with the embedded
material.

That is the behavior I see on my desktop with the Fuze bottle mesh. Are
you sure you're using the latest version of RViz?

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#33 (comment)

[ClintLiddick]

I tested this as-is with load_urdf from personalrobotics/dart_examples, and herb loads without issue.

[mkoval]

I think I addressed all of @psigen's immediate comments. I am merging this because it is causing all of our builds to fail. We can do any minor cleanup in another PR.