93ed0f096d
* Refactor/deduplicate camera eye code Camera Eyes previously had duplicated logic across several files. This change uncooks the spaghetti. Additionally, half-baked support for TG's multicam feature has been removed, as it was not functional or in use. * lets ff now * Camera Eye refactor fixes and finishing touches This change completes a refactor of AI eyes, which were previously used by xenobio consoles, syndicate and abductor camera consoles, shuttle docking computers, holograms, and, of course, the AI. Duplicated logic has been extracted to an abstract base mob, /mob/camera/eye, from which new types for each of the above now derive. Functionality is largely the same, with only a few minor cosmetic differences (i.e. camera eyes are now appropriately named given their type and user), as well as a quality-of-life enhancement for holograms, slowing their movement speed to base run speed to prevent users from accidentally zooming out of calls. * Camera eye refactor: Fix AI acceleration toggle The acceleration toggle was broken in the camera eye refactor, as previously the boolean was stored on the AI rather than its eye. This change fixes that. * Camera eye refactor: Fix syndicate cam visibility With the camera eye refactor, the syndicate advanced camera consoles lost the ability to view maintenance tunnels and other areas without active cameras, seeing static in their place instead (as all other cameras do). This change reinstates the original behavior. * Camera eye refactor: Convert spaces to tabs * Camera eye refactor: Fix CRLF * Apply suggestions from code review General minor code quality improvements suggested by GDNgit Co-authored-by: GDN <96800819+GDNgit@users.noreply.github.com> * Apply suggestions from code review Rename parameter names to avoid src accesses, remove an ambiguous and unused mob_define and holopad range variable from a previous WIP, change the for loop in /mob/camera/eye/relaymove to a for-to loop, and change the chat message warning, sent when an AI Eye is created on an AI that already has one, to a stack trace * Adds toggle to AI commands for fast holograms * Refactor ripped Hologram Eye relaymove Previously, the relaymove proc for hologram eyes was redundant and nearly impossible to read. It has been separated out into a few different named procs, and has had its use of `spawn` removed. * Remove unnecessary src access * Fix bug involving shuttle placement outlines The camera eye refactor that this commit is a part of introduced a bug that prevented shuttle placement outlines from showing up on first use of the shuttle console. This change fixes that bug. * Unrevert some changes from #26306 lost in merge * Remove erroneous free xray vision on advanced cams * Autodoc camera acceleration vars * Remove redundant null var initialization per code review Co-authored-by: Drsmail <60036448+Drsmail@users.noreply.github.com> Signed-off-by: asciodev <81930475+asciodev@users.noreply.github.com> * Changed variables to camel_case, autodocs, cleanup Changed a number of camera eye-related variables to camel_case style, added appropriate autodoc comments, as per code review. Also removed an unused cameranet function, modified the call signature of a cameranet function to be more semantic, and changed a qdel-on-initialize in camera eyes to return INITIALIZE_HINT_QDEL instead. Co-authored-by: Luc <89928798+lewcc@users.noreply.github.com> * Remove stray qdel(src) per code review Co-authored-by: Luc <89928798+lewcc@users.noreply.github.com> Signed-off-by: asciodev <81930475+asciodev@users.noreply.github.com> --------- Signed-off-by: asciodev <81930475+asciodev@users.noreply.github.com> Co-authored-by: GDN <96800819+GDNgit@users.noreply.github.com> Co-authored-by: Drsmail <60036448+Drsmail@users.noreply.github.com> Co-authored-by: Luc <89928798+lewcc@users.noreply.github.com>
AI Freelook
Credits
Initial code credit for this goes to Uristqwerty. Debugging, functionality, all comments and porting by Giacom.
What is this?
This is a replacement for the current camera movement system, of the AI. Before this, the AI had to move between cameras and could only see what the cameras could see. Not only this but the cameras could see through walls, which created problems. With this, the AI controls an "AI Eye" mob, which moves just like a ghost; such as moving through walls and being invisible to players. The AI's eye is set to this mob and then we use a system (explained below) to determine what the cameras around the AI Eye can and cannot see. If the camera cannot see a turf, it will black it out, otherwise it won't and the AI will be able to see it. This creates several features, such as.. no more see-through-wall cameras, easier to control camera movement, easier tracking, the AI only being able to track mobs which are visible to a camera, only trackable mobs appearing on the mob list and many more.
How it Works
It works by first creating a camera network datum. Inside of this camera network are "chunks" (which will be explained later) and "cameras". The cameras list is kept up to date by obj/machinery/camera/New() and Destroy().
Next the camera network has chunks. These chunks are a 16x16 tile block of turfs and cameras contained inside the chunk. These turfs are then sorted out based on what the cameras can and cannot see. If none of the cameras can see the turf, inside the 16x16 block, it is listed as an "obscured" turf. Meaning the AI won't be able to see it.
How it Updates
The camera network uses a streaming method in order to effeciently update chunks. Since the server will have doors opening, doors closing, turf being destroyed and other lag inducing stuff, we want to update it under certain conditions and not every tick.
The chunks are not created straight away, only when an AI eye moves into it's area is when it gets created. One a chunk is created, when a non glass door opens/closes or an opacity turf is destroyed, we check to see if an AI Eye is looking in the area. We do this with the "seenby" list, which updates everytime an AI is near a chunk. If there is an AI eye inside the area, we update the chunk that the changed atom is inside and all surrounding chunks, since a camera's vision could leak onto another chunk. If there is no AI Eye, we instead flag the chunk to update whenever it is loaded by an AI Eye. This is basically how the chunks update and keep it in sync. We then add some lag reducing measures, such as an UPDATE_BUFFER which stops a chunk from updating too many times in a certain time-frame, only updating if the changed atom was blocking sight; for example, we don't update glass airlocks or floors.
Where is Everything?
cameranet.dm: Everything about the cameranet datum.chunk.dm: Everything about the chunk datum.eye.dm: Everything about the AI and the AIEye.updating.dm: Everything about triggers that will update chunks.