I'm working on a maze environment, and studying the scenario of first-person visual navigation. The attached are the xml file and two camera images. I have two questions: 1) From the first image (named "world.png"), the colors of some walls are dark, which is problematic in the scenario of visual navigation. Is there any way to let the entire environment bright? 2) From the second image (named "first-person.png"), which is the first-person view of the agent, and the box geom indicates the direction of first-person view. However, the image is quite odd. It seems that the images, which the agent collected during its exploration, sometimes contain information w.r.t the places inside walls (boxes) and outside the maze. The reason I guess might be the agent could see through walls. Can you give me some advices? You can simulate the xml to get the above images directly by pressing "]" to switch cameras. I'd appreciate if you could help me, and this will help me a lot with my research!