We're definitely looking to resolve as many issues as possible, both for our own sake and to help with modder-specific bugs. For example, the skintone and navmesh bugs are both being addressed in an upcoming patch. Keep checking patch notes on bethBlog for that. We are also looking into other issues, such as large refs disappearing - but those fixes will be a few patches further away. We also just started a test of a potential fix for the grass-doesn't-update bug which many landscape modders are running into.
I'm sure there will be a number of things we just can't change or fix in the long-run, and I know how unbelievably frustrating it can be when a bug that blocks your progress. Our hope is to minimize this as well as possible, and hearing from the community is the best way for us to focus our efforts.
.
Beware of trying to fix bugs that don't belong to you. While I am fairly certain that a range of skintone bugs have been following
TESCS and
GECK around for a while now, I've seen the YouTube video of that "navmesh bug" (where actors can't find their way once enough cells are loaded and the player returns). There is a similar issue where, if the computer is pressed to the limits of its performance, objects and odd bits of landscape randomly disappear. I have seen this disappearing model problem and I am convinced that it is a performance issue because once the computer is running the game without a heavy load from other programs, the problem seems to go away. This raises a couple of questions I think are worth asking.
.
Does this navmesh bug simply occur whenever enough cells are loaded (without having returned to a given cell)?
And is the problem diffused by quitting to desktop and restarting the game (or restarting the operating system)? If it does, then it sounds like a performance issue.
.
I saw something similar when working with the
GECK for
Fallout: New Vegas. Basically,
Fallout: New Vegas had been running for quite some time and, simultaneously, so had the
GECK. The navmesh was failing in places where it had been previously repaired, tested and verified as functioning correctly. The problem was intermittant rather than persistent, and was temporarily resolved by restarting the game. Again, this points to resource management, which is usually the purview of the operating system (although some high load software has its own resource management because what ships standard with the operating system may prove unreliable). If games like
Skyrim,
Oblivion,
Fallout 3,
Fallout: New Vegas use their own inbuilt resource management routines to free up or reallocate memory occupied by redundant data - then this might be the location of navmesh and other disappearing model/object problems as well as "memory hole" issues.
.
The modern .NET Framework has a system for recycling buffered resources (such as memory) - based on manipulation of the
Windows API. It is worth noting that dreaded
Windows memory hole is owned by
Microsoft (and is probably "protected by Copyright Law and International Treaties"). All irreverence aside, you could waste a lot of time chasing other people's bugs if you are outsourcing resource management to an external API like IE, DX, or .NET.
.
By the way, based on the typical development time frame and the size of your staff, I've always suspected that Bethesda's game builders have worked with the same tools - before they get debugged. May I suggest that by ironing out those bugs you could cut development costs and delivery timeframes considerably? Some suggestions which come to mind:
.
1. Making the external boundary of navmesh bold and in a colour which cannot be missed will highlight navmesh tears - which play havoc with actor navigation and mod debugging alike.
2. If all snap to grid settings are automatically suspended on entering navmesh editing mode, this would streamline workflow.
3. The message box concerning whether or not to view the linked object also needs the option to simply view the object reference properties - for when mouse and gizmos fail.
4. A popout sidebar with the current hotkey list on one tab, functions on another, etc
5. Models with standard dimensions and a high rate of repetition (e.g. walls) could have a paint on-option, like Command & Conquer.
6. The paint-on option can follow terrain if the models concerned are enclosed in the plane on which they join.
.
And a bit of advice: Whenever you are writing direct to an interrupt (and bypassing the API), always include a subroutine which consults the registry, on load, to find out which interrupt you are looking for. Both
Skyrim and
CK sometimes ignore hardware settings in Windows and interrogate the wrong interrupt as a consequence; mouse buttons being a case in point. If the user has left-handed settings,
Skyrim consistently confuses the left and right mouse buttons while the Rendering Window in the
CK requires both to be pressed when repositioning objects. This needs to be addressed or lefties will have to start resoldering the microswitches on their pointing devices to force the issue. Imagine the warranty implications.
.
And now for some more criticism.
.
1. I think that adding the "gizmos" to the
CK is a fantastic idea because most people with any exposure to graphic modelling are already familiar with the concept. I also like the existing hot keys for manipulating models in the render window because they are fast and efficient when things are correctly set up - it's just a matter of remembering which key to press.
2. Rendering objects one by one instead of hanging the window until every object is rendered allows better control of the rendering window. If you ever do a slide out help sidebar (on the RHS) for hot key lists and other quick help, then including the F5 key in the list might be helpful.
3. I think the OP has a great idea. There is probably some room for "layering" or grouping of render objects - especially those which just get in the way. Last I checked there was an option to turn vegetation rendering on or off from the toolbar. If this were extended to allow subspaces, effects, etc. to be toggled it would make life easier. I would also suggest a menu system for toggling on or off rendering of all models belonging to any given kit - which will allow progressive mods (e.g. pitch a tent, upgrade to a shack, then to a cottage, then to a house, then to a manor, then to a blockhouse, then to a keep, then to a castle, etc.) to be developed or edited with more facility.
.
One more thing. If you find your people are having to either "Mcguiver" or hard-code their way around a non-functioning feature in the course of developing the game, then there is a problem with the tool which needs to be addressed before it costs a fortune in lost time. Look at it this way; if your teams spend the first 20 minutes of each day discussing the procedures, pitfalls and tool defects associated with the days work (while the team leader documents the meeting and, separately, procedure modifications), they are better prepared, have less accidents and make less mistakes. This also gives team leaders the opportunity to meet, subsequently, for their 20 minutes to share findings, raise issues which need to be addressed and get the appropriate resources allocated before the horse leaves the barn. I've seen this kind of system more than halve delivery times - which represents more savings and the capacity to deliver on a much larger scale. Moreover, as the procedures become better optimised and require less modification, the time spent in daily start-up meetings dwindles considerably. This kind of approach is well worth the investment.
.
Anyway, I hope you find something useful here. Good luck with the bug-hunt.