View Single Post
  #6   Spotlight this post!  
Unread 20-02-2005, 20:08
Unsung FIRST Hero
Greg Marra Greg Marra is offline
[automate(a) for a in tasks_to_do]
FRC #5507 (Robotic Eagles)
Team Role: Mentor
 
Join Date: Oct 2004
Rookie Year: 2005
Location: San Francisco, CA
Posts: 2,030
Greg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond reputeGreg Marra has a reputation beyond repute
Re: Do you need the exposure values?

Quote:
Originally Posted by Greg Marra
Why do we need those numbers. Can anyone explain to me what CR 19 (the camera register that I believe those numbers are inserted into) does?

It seems to me that "track green" in the GUI almost always works under any lighting conditions, regardless of whether I have calibrated yet or not.

And if anyone knows the commands that "track green" in the GUI sends to the camera, I'd be extremely interested in knowing.
And lo and behold, I have answered my own questions.

"What does 'Track Green' do?"
For the answer to this question, see "CMUcam2GUI_fe\src\config.txt"

"What is CR 19?"

CR 19 sets the camera's internal auto-exposure control level. 0 would create a totally black image. 255 should create a totally white image. The key is keep the image dark enough that bright areas (like what we're trying to track) don't get drowned our by background noise. Basically you want this number to be as low as possible without totally blacking out your image. What the specialized FIRST GUI does is calculate this number for you, saving you a lot of work. I speculate that the progress bar indicates how far it is (0-255) and that the reason why it often doesn't finish is the calibration value is below 255.


Now here's the really big question:

How much variation is there in color detection from unit to unit, and can I trust the numbers first gives me?