Improve Ogre2GpuRays performance

[iche033]

🦟 Bug fix

Summary

Tuned Ogre2GpusRays' (lidar) ogre2 compositor to bypass computationally expensive phases involving lights. Since we are just generating depth data, we do not need to care about the lights in the scene so we can skip these computations.

The compositor parameter description can be found in this ogre doc.

• enable_forwardplus - affects performance when there are many lights in the scene. However we only care about depth data for lidars (and not the rgb color images that are affected by lights) so set this to be disabled. Specifically, a big chunk of time was found to be spent in ogre scene manager's _cullPhase01 function. Disabling this flag gave the most perf boost.
  • Note the ogre doc lists enable_plus which is a typo.
• light_visibility_mask - tells ogre that we don't want to see any lights

I used this lidar_profile.sdf world for profiling. It consists of a single lidar with 160 degrees FOV and 640 samples.

• Note: To simulate this lidar, internally Ogre2GpuRays uses 3 environment cameras for collecting depth data - the number of env cameras is important as it directly affects to total render time. If a lidar has a smaller FOV (<90), only 1 env camera is needed and the update time would be close to ~1/3 of the time shown below.

Here're before and after plots of time taken to do a Render operation

The time taken is now roughly 2-4ms for each render update. The performance is still behind ogre 1.x in which the render operation for the same lidar takes ~1ms

Checklist

• Signed all commits for DCO
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (eeaf6e8) 75.54% compared to head (3e65518) 75.54%.

Additional details and impacted files

@@                Coverage Diff                @@
##           gz-rendering8     #955      +/-   ##
=================================================
- Coverage          75.54%   75.54%   -0.01%     
=================================================
  Files                177      177              
  Lines              16888    16889       +1     
=================================================
  Hits               12758    12758              
- Misses              4130     4131       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ahcorde]

@iche033 can we backport this to fortress ?

[iche033]

@iche033 can we backport this to fortress ?

yep I'll do that