Tom Penny said:
How can you walk in a game with just motion controls and no controller without an auto walk / stop commands?
The scenario I was painting was similar to the classic point 'n click single screen or panning multi-screen rooms or areas that were designed to show everything to the player from one perspective, similar to how a stage set is meant to face one fixed direction for maximum accessibility.
Anyway, you could initiate the space being tracked in the game software by generating it with the player at the center or generate it based on the trackable area in front of the lens, stretching out to the edges of its field of view and then allow the player to move inside of the actual tracked space in front of the camera as that translates itself to the corresponding space in the virtual on-screen room. The player moving will move the corresponding player character. Obviously, since the player always has to be looking toward the camera and display to play, their player character can face in the appropriate direction during interactions and special animations that match up with item, character, or terrain in the screen irrespective of which direction the actual player is facing at that moment, though their facing could be used in specific moments distinct and made known to the player just prior to tracking actual player facing.
Only the player's position in front of the camera will translate to relative position on-screen where objects and characters that exist in the on-screen virtual space are auto-navigated around if the player moves through or stands inside of them...the way point 'n click titles move characters around impassibles by relying a bit of pure pathfinding and/or navigation nodes. If the current game screen is larger than the display, you can make the extreme edges of the trackable space in front of the camera similar to dragging you cursor to the edges of the screen in a top-view-ish interface seen in most RTS titles where you'll scroll based on cursor position.
Or you could make the player's position in front of the camera always at the center of the current game screen and simply move them based on their distance and direction from the 'home' position initiated at the start of the session. Just imagine drawing a grid from the player's position in front of the camera and then visualizing it overlaid or integrated with the game world visuals to allow the player to know where they are and how much they need to move in front of the camera to start walking in a direction (one cell away from 'home' position), running (two cells away from 'home'), or stopping in place (back to 'home'). If the grid incorporated smaller cells to occupy with just one foot or leg inside of the 'home' position, they could be the ways to turn or rotate the view. Not that different to imagining a virtual floor pad, like a DDR grid or old NES Power Pad. Except that this could be any shape and sized cells generated on-screen for the game to use for various functions of world navigation and could, of course, be changed on the fly. It only needs to be visualized consistently on-screen for the player will be necessary alongside a realtime representation of the player themselves (in avatar or other game character form) to keep them informed as to where they are and how they are moving.
Really, it wouldn't be all that different to what is possible with a standard camera in an augmented reality setup except that this isn't utilizing video feed to drive the on-screen visuals. There are also possibilities with voice commands and joint use of the game controller. Honestly, it sounds more complicated to use than it probably would be in practice, but if the basics are laid out plainly and consistently, the rest should fall into place within its limits.