DRC Terrain texture warps when gazebo camera viewpoint changes

[osrf-migration]

Original report (archived issue) by John Hsu (Bitbucket: hsu, GitHub: hsu).

The original report had attachments: drc1.1_heightmap_stereo_bug.jpg, drc1.1_heightmap_stereo_working_custom_env.jpg

---

from drcsim #23

[osrf-migration]

Original comment by Stefan Kohlbrecher (Bitbucket: Stefan_Kohlbrecher).

---

• set attachment to "drc1.1_heightmap_stereo_bug.jpg"

Here's the result of stereo_image_proc with DRCSim 1.1.0. As visible, the points for the Turtlebot and Pioneer are at a reasonable distance, while the heightmap terrain points are completely wrong. I strongly suspect that this is related to the strange shader applied to the terrain.

[osrf-migration]

Original comment by Thomas Koletschka (Bitbucket: thomasko).

---

Well that's what you would expect, right? The weird shift in the heightmap shader increases the parallax and hence it looks like it's farther away - not just to the human eye. ;)
The same thing will affect you when computing visual odometry because the surface will appear to move faster than it actually does I think.

Just use a simple plane or other mesh with texture attached to it until it's fixed.

[osrf-migration]

Original comment by Stefan Kohlbrecher (Bitbucket: Stefan_Kohlbrecher).

---

• set attachment to "drc1.1_heightmap_stereo_working_custom_env.jpg"

Yes, it's expected, just wanted to provide confirmation. Attached is a screenshot of a custom environment. As expected, stereo works there (ignoring the replacement texture bug (issue #239) for now).

[osrf-migration]

Original comment by Nate Koenig (Bitbucket: Nathan Koenig).

---

• changed state from "new" to "resolved"

Fixed in pull request #156

[osrf-migration]

Original comment by Thomas Koletschka (Bitbucket: thomasko).

---

This does not seem to be fixed actually. Using the current gazebo 1.3.0 and drcsim 1.1.0 release you still get very bad distortions: http://youtu.be/n7sWSQ_eyD0

Using the corresponding pull request for drcsim which changes the normal map for the terrain to a flat one also doesn't fix the problem but rather avoids (most of) the symptoms. However even with the flat normal map the terrain still has a more or less small offset in z just that it's constant.

[osrf-migration]

Original comment by Nate Koenig (Bitbucket: Nathan Koenig).

---

• set version to "all"

[osrf-migration]

Original comment by Nate Koenig (Bitbucket: Nathan Koenig).

---

• changed state from "resolved" to "closed"