robotology/icub-models

Fix iCubGazeboV2_5_visuomanip colors

traversaro opened this issue · 7 comments

traversaro commented

The model iCubGazeboV2_5_visuomanip added in #42 has strange "random" colors. For automatically generated models, the issue will be fixed by robotology/icub-models-generator#217, but for iCubGazeboV2_5_visuomanip if we want to fix the issue we need to it specifically for this model.

fyi @xEnVrE @pattacini

pattacini commented

"Random" in which sense? Isn't the "standard" anonymous whitish color?

if we want to fix the issue we need to it specifically for this model.

Fine with me!
Since @xEnVrE authored it, I'll leave to him the final choice.

  • 👍1
xEnVrE commented

Hi @traversaro @pattacini,

fine with me. I imagine that I could check how the automatic pipeline changes the materials in the automatically generated models and apply the changes in the manually generated model.

@traversaro do you think this solution is doable? (I can take care of it.)

  • 🚀1
traversaro commented

"Random" in which sense? Isn't the "standard" anonymous whitish color?

Actually, not. The whitish color we were mostly used to is due to a bug in URDF parsing that was both in iDynTree and in Classic Gazebo, but if you displayed the model in a ROS tool like RViz you were always getting it as an "Arlecchino". @GiulioRomualdi is fixing the issue in iDynTree (see robotology/idyntree#961), so we were going to see an arlecchino robot in iDynTree, but in Classic Gazebo we will continue to see a whitish color.

  • 👍3
traversaro commented

Hi @traversaro @pattacini,

fine with me. I imagine that I could check how the automatic pipeline changes the materials in the automatically generated models and apply the changes in the manually generated model.

@traversaro do you think this solution is doable? (I can take care of it.)

Ok!

  • 👍1
xEnVrE commented

I made the changes.

Some notes:

  • I took the colors from the latest icub-models (not sure why the red color was picked up for the entire robot, however I will leave it in order to have coherent colors w.r.t. the automatically generated colors)
  • I had to change the color of the part indicated with the green arrow. It is a metallic part, hence it should not be red, I guess.
  • The top and palm cover of the hand are black as in real robots
  • The fingertip should be black, however the mesh is shared with the last part of the finger, hence I could not make it better than that.
  • The following were generated using the meshcat-based visualizer provided here

image

image

Eyes bulbs

I will use the idyntree-model-viewer to show you the eyes as the meshcat-based one does not support spheres and cylinders, that are used for the visual description of the eyes.

image

If this configuration looks good I can open a PR with the changes. @traversaro @pattacini

Btw, I noticed that using the idyntree visualizer the colors of hand and fingers are not taken into account. Is this because they are daes instead of stl?

  • 👍1
pattacini commented

Fine with me 👍🏻

  • 👍1
traversaro commented

(not sure why the red color was picked up for the entire robot, however I will leave it in order to have coherent colors w.r.t. the automatically generated colors)

Basically the one that do the modification get to choose the color, see robotology/icub-models-generator#217 (comment) . :D

Btw, I noticed that using the idyntree visualizer the colors of hand and fingers are not taken into account. Is this because they are daes instead of stl?

Yes, in the iDynTree visualizer .dae files can specify their own color inside the file itself, so the URDF color file is not taken into account. This is also the behaviour of Gazebo if I recall correctly.

  • 👍1
  • 😄2