this is the most necessary thread ever. here's my 10 cents
Quote: "Consoles and some PC game engines handle things differently, by allocating a large chunk of the heap and then manage this heap directly for all game data and resources. It is then easy to say this heap is for 'level one', use it to the max, then discard the whole heap ready for 'level two'. Unfortunately, to implement this would require taking over quite a large chunk of aforementioned systems with some kind of custom memory manager and I would not even know where to start on a mission of that magnitude. "
Perhaps before new stock media and graphical improvements like "PBR" and whatnot is included, this should have all attention devoted to it. As it is: it's one of the only things stopping a competent studio from creating a game that advertises GameGuru to other competent studios and creates a more financially "liquid" audience of actual gamedevelopers using GG. Take a look at the prices of assets on average in the unity store or UDK marketplace and think about what a lucrative market you could attract by fixing what is a core fundamental flaw that needed to be looked at since day 1.
Quote: " There is a hope that if there are some small memory leaks, and some items are not being deallocated when they could be, then this would go a long way to ensuring levels can load without significant fragmentation worries."
absolutely, but I think focusing on the core memory management, even if that means outsourcing some of the work you yourself cannot do would put gameguru into the "viable solution" category for a lot more knowledgable aspiring game devs. These are people who are going to bring all sorts of experience and assets with them to our community.
Quote: "the engine has never really had an overhaul that specifically focused on the memory management of the engine and chunks of the code derived from the even older DBP technology that pre-dates even FPS Creator, so there is plenty of scope for memory management improvements. A quick hack could be to restart the application for each level loaded, thus clearing the fragmentation issue, but this would add 5-10 seconds onto your level loading time, and is a bit crude, but it's another solution that might work for the short-term while plans are put in place to solve the larger issue of game engine memory management."
that sounds like a fantastic course of action. Load times in GG are very long already so this would be a great solution, especially if some time was spent to create a seamless experience, say via a dummy program that will keep the loadingscreen up while the app reboots! Once load times (another serious GG issue) are looked at (hopefully in the future once core functionality is stable), if this was still even necessary it would be a small load time price to pay for the ability to create more levels.
Quote: "From feedback so far, there is a camp which argues that GameGuru does not yet produce games of sufficient quality to warrant efforts to improve the standalone game side of things (including multi-level games), and much prefer energy spent on performance, AI and general core features."
One could argue performance should be looked at first, given that you wouldn't need to create a lot of levels if you could create longer ones, but I notice that most people's maps around here don't nearly span the full expanse GG gives you, and if it does, it is usually very underdetailed, or running at a very slow frame rate. Even The Big Escape's 5 minute enjoyable linear experience proves that a perfectly servicable FPS game experience can be made, which was (i had assumed) the plan all along.
In creating short, linear experiences, gameguru can handle decently detailed scenery on a pretty good scale at a pretty decent FPS (improving on FPSC in all the basic necessities, more enemies, bigger maps, terrain, shaders, etc) and as such I have noticed that's what most developers that get the best results tend to do; they make bigger-scale FPSC games.
in one SHORT level, gameguru can create some truly visually splendid awesome experiences, but one short level is not enough to make an impression these days, especially with how short gameguru's performance constraints require that level to be.
That being said, 20 short levels could make that profound impression that continues to sell gameguru (and thus help fund its development) for years to come, but as it is, that's made very difficult due to these specific constraints stopping larger projects, especially ones with a high level of graphical fidelity. (which coincidentally are the ones that will probably be the most successful) Therefore I think memory management/multi level capability should be priority number one, with frame rate coming right behind and load times for dessert
