Autosaves in TDM

[Oktokolo]

You can give this argument about everything, as Spring already mentioned.

I could. But i don't. And it would be totally wrong anyway.

Most stuff defining a mission is not easy to be changed by a player. Even ghosts don't add AI, patrol routes, traps, hidden mechanisms, noisy surfaces, lights... to a mission as they play it.

 

Personally I always play a game the easiest way possible, and I am not talking about the difficulty level, I choose the highest one normally. What I mean is that I see games as a sum of problems I have to solve, and I am always searching for the easiest and sometimes most elegant solution. I would never restrict myself in the ways you mentioned. Why should I?

You should not as you obviously do not like to. But if you would like ghosting for example - you would try to ghost missions regardless whether they feature a hardcoded ghost restriction or not.

 

 

Of course my arguments are subjective. We are talking about TDM mission restrictions. Especially, whether the ones, that are hard to ignore (if hardcoded) but easy to add by the player if not, should be hardcoded or not. It is a subjective topic, this thread is evidence enough

[peter_spy]

Or even on a treatment of save systems as wholly internal to game mechanics: if I have twenty minutes available and the mapper has rashly decided I ought to have had thirty... but as I said, if we're at the point of raising philosophical differences of opinion about the relationship between desires and rational activity then I don't think we're going to reach agreement about save mechanisms. My practical recommendation to the OP would be to keep Expert for the kinds of restrictions noted in this thread and make everything else available on lower difficulty as well.

 

This is both kind of nitpicking (difference between 20 and 30 minutes, reallY? even though it's not even designed this way, since checkpoints are usually location and task-based), and it implies that only your approach is rational, which isn't true either.

 

I'm not saying it's easy, but you could design a whole mission with save system managed by the game only. And it would be a fun mission, where players would have to learn the environment better, improvise in the face of AI alarm, or live with consequences of their actions (if there are such choices to be made). Exploring other choices, striving for level mastery would be possible by replaying the map, not mashing the save and load buttons. That of course requires very good mission design to begin with.

 

Some futher listening/reading:

http://www.roningamedeveloper.com/Materials/RandySmith_MiGS_2007_SaveLoadCompulsions.ppt

http://www.gdcvault.com/play/556/How-To-Help-Your-Players

[Springheel]

In my opinion, stealth games should be challenging. If they're too easy, they're not fun.

 

It follows that stealth games should be designed to be challenging. The challenge should be built into the game, not left to the player to figure out. If a game requires the player to create self-imposed restrictions in order to make it challenging, then something is wrong with the design of the game.

 

Obviously, different people will prefer different levels of challenge based on their skill level and tolerance for frustration. This is why providing a range of options is good. However, those options still need to provide the requisite challenge without asking the player to tie one hand behind their back.

[peter_spy]

To be more precise, probably all developers' presentations on stealth game design I've ever seen stress that the game should look like it's challenging. So it's more about subtle way of making players feel smart. To "Create the illusion of a securely guarded area that the player can sneak through. Make the player feel like they are special enough to do the impossible".

[VanishedOne]

 

This is both kind of nitpicking (difference between 20 and 30 minutes, reallY? even though it's not even designed this way, since checkpoints are usually location and task-based), and it implies that only your approach is rational, which isn't true either.

It implies that Obsttorte's stated views on motivation and reasons for acting strike me as philosophically implausible, which they do; of course in a scholarly discussion I'd be interested to see any broader supporting arguments, responses to counterarguments, etc. but here that would just take the thread wildly offtopic. So I merely noted that we did not seem likely to reach agreement. Yes, of course I continue to hold certain views; Obsttorte evidently holds different and incompatible ones; and yet life will go on. Happens all the time.

 

And yes, really; though the salient point is that one number is greater than the other. The designer works with imperfect knowledge, particularly when it comes to life outside the context of the game. That's a solved problem: instead of trying to apply game design principles to saving, one leaves it to the player. If you want to unsolve it then, as designer, you must choose what assumptions your design will incorporate.

[peter_spy]

Not really, the player psychology is much bigger problem, at least from the designer standpoint. That's why insead of letting player save all the time and expoit that feature, designers create checkpoint systems that try to handle this seamlessly, so the player focuses on gameplay. You may think that manual saving is the best thing since sliced bread, and of course you're entitled your opinion, but it's not superior to anything else, as you're trying to imply. It's just one of many systems.

[Obsttorte]

It implies that Obsttorte's stated views on motivation and reasons for acting strike me as philosophically implausible, which they do; of course in a scholarly discussion I'd be interested to see any broader supporting arguments, responses to counterarguments, etc. but here that would just take the thread wildly offtopic. So I merely noted that we did not seem likely to reach agreement. Yes, of course I continue to hold certain views; Obsttorte evidently holds different and incompatible ones; and yet life will go on. Happens all the time.

The thread is about whether to use a checkpoint system, and in the intention of the thread author, to use it for the purpose of increasing the difficulty on higher settings.

 

As the one who suggested and implemented the code to support this, which took me several hours btw. (10-20, not sure anymore), I think that I can state why I did that. And as noted above the reason was never to provide yet another feature for increased difficulty. So I am not happy when it gets reduced to that aspect.

 

And I don't see why this discussion is philosophically or scholarly just because whe don't get to a consens. There is no need for a consens. Tastes are different. But as this is a platform for persons who want to create Fan Missions (which I see as a form of art), I think that these discussions are extremely useful and important. If we don't question why we do like or dislike certain features or why we use some of them in our fm's, we will hardly be able to improve ourselves. But again, I don't have to agree with you just to learn something from you and vice versa.

 

If you don't consider such conversations beneficial, I suggest you don't participate on them

[teh_saccade]

I just leave a note in an obvious place that says, "SAVE ME!".

That's my checkpoint.

[teh_saccade]

Not really, the player psychology is much bigger problem, at least from the designer standpoint. That's why insead of letting player save all the time and expoit that feature, designers create checkpoint systems that try to handle this seamlessly, so the player focuses on gameplay. You may think that manual saving is the best thing since sliced bread, and of course you're entitled your opinion, but it's not superior to anything else, as you're trying to imply. It's just one of many systems.

TL;DR - checkpoints tend to be geared toward more linear play and would require more thought regarding their placement in games such as TDM (Look at difference between SC:CT and SC:Conviction, christ if I have to watch Ezio have that same conversation one more time... I'm gonna never play this game again). Limiting saves or disabling quicksave is one alternative. Manual save is perfectionist's / libertarian dream. Also sliced bread isn't that great - I prefer to cut it myself, because the loaf is usually fresh baked rather than some mass-produced stuff that doesn't fill my toaster and rips if you spread the butter against the grain.

 

(that's what we call "a metaphor" - it's a touch more complicated to understand than a hackneyed simile or maxim)

 

 

 

There are many ways to get players to feel the urge to manually save without having to use checkpoints that (in my experience) can be poorly placed... How many times do I have to listen to that conversation or watch that cutscene?

 

Good design leads players to believe it's time to save because it feels as if they're entering a "danger zone" or have just accomplished something.

 

In games such as TDM, water works well - esp if it is a long swim underwater that might cost a little health. Bit of tense music as the section approaches, change in lighting and perhaps a small jump-scare can promote the feeling that, "shit's gonna happen, I probably better save".

 

There's nothing worse than transitioning into a new area only to find that the AI is timed in such a way that you've arrived in the middle of all the patrols, or have to wait - AGAIN - for that scripted conversation before you get the objective to do blah blahblah blah.

 

That's why I always blackjack birdie and just take the instructions - I don't got time to listen to him when the guard is chasing me because he saw me climbing over the wall and up the balcony because of my bad timing.

 

Personally, even though I am not a great fan of poorly implemented checkpoints - they can be very useful if manual saves are not allowed.

It would certainly increase the difficulty of many missions that are "expert" level and mean no more "oops I got spotted, better quick-load" and force players to deal with consequences for their actions.

 

What's the point in all those flashbombs, moss arrows, noisemakers (that don't seem to work through visportals atm) if you don't ever need to use them because you can simply reload if the situation doesn't go your way and you get ... chased and have to hide and escape..?

 

When a checkpoint is placed BEFORE a danger event and AFTER some bullshit I don't want to have to go through again and again and again - they work well.

 

Because I'm stupid, I often make a save at the start of a mission and don't save the game again.

If I die - I have to restart. If I mess up - either I have to restart or deal with the situation.

 

It can be extremely annoying in "don't get seen or fail mission" or "don't kill or fail mission", because sometimes I like to pickpocket a broadhead and stick it into the wall to distract and get past that guard, instead of sitting here, scratching my ass, waiting for the "ghost" path to appear. Or run or jump so someone comes to investigate that area while I creep over there. Or chuck a candle and accidentally drop it and have to run away, get into a swordfight (if didn't fail because spotted) and accidentally land a killing blow instead of making the guard run away crying for help.

 

Much prefer optional objective "don't get spotted/kill" - at least then I don't have to restart because I accidentally pressed spacebar or didn't make that jump to mantle and got into a sword fight (where I can't fight back) and was shot to bits by arrows.

(besides, I like to be a psycho and headshot people with the one arrow someone forgot to remove from an archer, that I use to headshot everyone in a weaponless mission sometimes - sometimes I even like to kill the people I've KO'd, just because no-one will ever wake them up so they may as well be dead)

 

Judicious placement of checkpoints might be a good thing for players who are similar to me and don't like light-gems and use the moss arrows to land quietly behind that guy and creep over there and pass by that section using that rope arrow I found, collect the loot later on the way back, cos I've scoped it now and I spotted the guards and watched his routine - plus, I don't need the loot cos it's not gonna affect the next mission or change what I can buy in the shop because no-one really does that...

 

But repetition - it pisses me off.

If I had to do that over and over and over until I got it just right, from 5 minutes ago - esp. if it was "being spotted" failure - I'd uninstall that mission.

 

 

Meta gaming with no-saves / limited saves / "checkpoints" is fine.

So long as it is executed in a way that is not grinding my will to live as much as it makes me grind my teeth.

 

 

I still prefer "ambience / gameplay" prompts to get a player to save, but this can be abused.

Quicksave and quickload leads to routine perfectionism, which isn't real life - it doesn't increase player skill. It only increases their ability to negotiate a section.

 

A checkpoint is essentially an enforced quicksave.

However, it was replaced by another "quicksave" as the player went through that door - there's no quickload if they've backed into a room with no exit and there's someone looking for them and a guard on their way... Gotta deal with that situation.

 

Or restart the mission.

 

 

IMO - checkpoints and/or limited manual saves (disable quicksave, but allow quickload from "checkpoint" as retry) would definitely add to both the difficulty of some of the "expert" level TDM missions as well as encourage a more "I have to deal with the consequences of my fuck-ups" type gameplay.

 

Which I encourage.

 

 

However, it would require the mission to be tight as a scotsman's coin-purse. Because if I found a way outside the bounds of a level (which happens fairly often) in order to bypass a large chunk by walking along single face brushes, high above the city, from which I've collected every box and chair to make my stairway to no-clip heaven - I'm going to miss out on a whole lot of checkpoints and be denied the joy of taking the risk to explore the mission as fully as possible.

 

It'd have to be more corridor-like.

 

That's the whole idea of a checkpoint, right?

 

It's a point along a route, that you check-in at.

 

 

It'd work in a "free-roam" level, but no-one ever explained to me how to make the jobs board / pub hub with mission zones I wanted and I'm still shit at scripting.

Even need help with it in unity and it's in the manual and mostly built in.

 

 

 

I agree that a checkpoint system would promote repetition, which is boring.

But I also agree that it would counter the whole "oops I did it again" thing, which is also boring...

 

Though - reallly - who gives a fuck how anyone else choose to play their game..?

If they want to quick-save/quick-load, who cares..? It's their choice (even if I do think it's lame when I watch a video and the person is OCD quicksave) - same as it is mine to punish myself by making one save per mission at the start, so I can quickly reload when I die / fail.

 

Are we in competition with them, the other players..?

Did I miss the TDM leaderboards? Because I'd probably want to be on them for most murders committed with the highest stealth score, yet fastest time to complete exactly the same objectives as the rest of you. Smash n grab. Or maybe ghost if I felt like it - who knows. I LIKE TO MIX IT UP.

 

Is there even a wiki page on how to put in checkpoints and has anyone really put together a strung-together campaign yet, where the loot and items I collected last mission counts towards this one?

 

 

This whole conversation appears totally pointless, if no-one actually implements the checkpoint system and proves that it is something that is worth using, effective and an attractive feature for a mission in which players want to raise their game by making it more difficult for themselves, but don't have the self-control to not save every corner...

 

 

 

(I spoilered it so you don't have to, biker)

Edited November 22, 2017 by teh_saccade

[Oktokolo]

You may think that manual saving is the best thing since sliced bread, [...]

There is something even better: Manual saving + configurable multi-slot quicksave + configurable multi-slot autosave. Multi-slot to be safe from savegame corruption (does not happen often - but if it does, it is always near the end of a mission).

I do not think, that TDM absolutely needs the full set. It is fine with manual saves and quicksaves.

But timed multi-slotted autosaves that are only triggered when not hunted, poisoned, under water, midair, exposed to bright light or standing right next to an AI would actually improve the usability of the save system. No more need to tap the quicksave key, just set the autosave timeout and slot count once and never think about manual saving or quicksaving again.

Maybe the reason, it is not in the game already, is that saving interrupts gameplay for half a second and therefore is somewhat immersion breaking when it occurs without being directly triggered by the player...

[peter_spy]

Yup, 2 quicksave slots, 2 autosave slots (to avoid file corruption problem), and manual save slots is kind of all-around system. As for saving conditions, AFAIR Dishonored doesn't even allow manual saves while you're in combat, which makes sense too.

 

I don't think checkpoints are typically for linear games, although they're sure easier to implement there. MGSV is an open-world game, and it has checkpoints only. The game saves before you pass any outpost, town or other section of the map, and when you enter/exit mission zones. There's no manual save. And this was a conscious design decision. First off, you don't have to think about saving. You also can't do cheap "perfectionist" runs. You have to learn the location layout, see where key items are, watch for AI patrols etc. You have to improvise and use gadgets if you fail at stealth, which you typically do when you don't know the location yet. You can also flee, but you'll have to start that area from scratch. That seems a bit frustrating to "compulsive perfectionists", but it also makes any perfect stealth/ghost runs feel like they're truly earned. That is a victory not cheapened by mashing Save button every time you think you might fail. These missions rarely take more than 15-20 minutes, so there's not much progress lost. Story missions are longer, but they have checkpoints every now and then.

 

Dark Souls is a semi-open world game, and it autosaves basically all the time. Every time you hit/kill a monster, every time you get hit/killed, when you open/close the game menu, talk to an NPC etc. Bonfire autosaves are more of the "mental checkpoints" for the player, so you can divide your playing sessions into manageable chunks. You can Alt+F4 out of the game and you barely lose any progress.

 

These two games save it for you, but what is more important, they make you deal with consequences of your actions, with Dark Souls being obviously more brutal than MGSV. But in both cases the world and gameplay design had to support this system, so it's definitely not an easy thing to make.

 

Another thing is what Pete already mentioned: sometimes I'm so immersed in the game, that I actually forget to save. And after some 20 minutes of gameplay, I make a stupid mistake, fall off a ledge and get killed.

 

(Edit: too many typos)

Edited November 22, 2017 by Judith

[teh_saccade]

That's not a bad idea - to put a delay on quicksaves or add some cost element (loot, time to complete, addition to stealth score - "you wanted 0 score? well you saved 40 times, so that's +35 stealth score").

There are many games I enjoy that allow only a certain number of saves for the whole game (eg, Warband, perhaps 8 for whole game) or for the mission as difficulty increases (eg, Hitman, from unlimited saves to zero per mission). Farcry 2 prompts saves at certain points with on-screen messages, but allows free/quick saving also (many times played for hours and died and realised... oh... haven't saved all session). Same with STALKER or Death to Spies or old Sniper Elite.

Covert Action, the mother of all stealth games, allows save only when not "in action" and at the impasse screen - I guess that's a "manual" checkpoint save between actions.
Roguelikes are becoming increasingly popular, but idk how that would translate to TDM - maybe only in campaign or longer missions, but it's a good model to use for example for checkpoint saves in transition only.

Anyway..:


Additionally, there is the fact that "number of saves" will be reported on the debrief screen at the end of a TDM mission. Maybe I could run this number so high it would no longer fit in the GUI?

Sometimes I like to have a total clean sheet - 0 everything, including saves (failed this time, he was gonna see me - so I had to put an arrow in creep's mate's head... plus - you gotta deal with the uber-guard to get the vase without alerting him).



Any autosave would have to not count as a save, because I don't want it to count as a save... :`(

What happens if I perform an action that disallows me from completing an objective and the game autosaves?

Since all the exploits to get around a level "silently" are being revealed (like the "draw arrow, hit esc twice. walk past the difficult bit, even though they can see you - you're silent - and release the arrow to teleport there" one)... If I spam save thousands of times - I can't complete a mission until the on-screen text for "game saved" stops appearing, which is why the easy mission that usually takes 3-4 mins as a warm up took so long... Waiting at the exit for the "game saved" to stop displaying... It also starts to lag the game horribly after the first few hundred saves...

Even though the mission timer says it was only 15 minutes... it took, in fact just over 2 hours to complete this mission due to the "game saved" message denying the "objective / mission complete" message - thereby stalling the mission from completing.



But hey - at least the GUI can handle 4 digits on the save counter, right?




And I had fun, waiting... til I got bored, did the laundry, went for a shower, shave, made food, went to the pub, came back, made tea and it still wasn't completed. Now it has - finishing this post.

----

Also - there's still the issue of hiding under an elevator and it coming down on you and you not getting crushed to death by it, getting stuck there and if there's an autosave on that elevator for transition... you're screwed unless the level designer knows to make sure to add the damage just in case that particular situation happens. Which it does.

(I'm ALWAYS going to bring up movers not causing damage and getting stuck under lifts at every opportunity).

The amount of times I've failed missions with elevators, because I get stuck under them - grabbing the loot - and some AI decides to trap me in the space underneath and won't get off or doesn't cycle it back up and I'm not waiting for their random interest pathway to lead them back to the elevator route...

Edited November 22, 2017 by teh_saccade

[Oktokolo]

What happens if I perform an action that disallows me from completing an objective and the game autosaves?

Then you load an earlier autosave slot. That is the reason, why other games provide multi-slot autosave and quicksave systems.

 

No save system is perfect. But there are techniques that can make them extremely robust against crashes and bad luck. Multislotting and prevention of saving in immediate danger are low-hanging fruits that are rather easy to implement and work for all missions of the past and the future without needing intervention from mission authors.

 

But one also could store the N last game states in a ring buffer and provide the player with the power to literaly rewind himself out of a mistake. Shadwen implemented that - it worked fine and was fun to use as it did not involved any delays or load screens.

Would have to be combined with slotted autosave to also provide automatic protection against crashes though.

I don't think we will see this implemented in the id Tech 4 engine (it would certainly be a monstrous amount of work). The game state buffer approach costs memory (naive approach is most likely unbearable and differential snapshots with "keyframes" all N seconds need more developer time for implementation) and is harder to implement than traditional save systems. A compromise would be to implement interruptionless saves by copying the current game state into a buffer on demand and then streaming that to disk while the player continues. That still might give the player a micro stutter if the engine does not follow a strict copy-on-write philosophy already and therefore has to delay progress until the current state has been duplicated in RAM...

 

And of course: You could use hand placed checkpoints. That is the one system that has to be designed in and taken care of in each mission seperately. It does not just work for all missions of the past and future - like the other save systems, that are either actively triggered by the player on demand or by a timer guarded by rather simple mishap-avoidance code...

[Destined]

Then you load an earlier autosave slot. That is the reason, why other games provide multi-slot autosave and quicksave systems.

 

It is not only "other games" that provide a multi-slot system. At least for quicksaves TDM has had a multi-slot system since 2015 (see this thread for reference). I don't see, why this should be a problem to transfer to the autosaves. If I remember correctly, file corruption was the main argument for it. I don't think it would be much of a problem to transfer this system to autosaves.

Edited November 23, 2017 by Destined

[teh_saccade]

So how many autosaves must it make and retain? 2? 20? 200? All of them?

Such a system will dictate level design.

Think of it as a level designer or a programmer, not a player and it might make more sense.

Game saves, scripted event happens, game saves, player actions, game saves, player interrupts scripted event / bugs out, game saves, player walks through that door with the autosave, player accidentally passes into area of no return or fks up holy water by not having any water arrows or misses that rope arrow or - as with most TDM design - didn't know the ghost route and timing and is being chased around a nondescript mansion / city street maze with few hiding places and a dead end save - needs several to go back to original scripted event or in order to progress.

Thinking of where to place autosaves in a level - they are places I would normally save a quck-slot myself, because I didn't grow up with an xbox (it was a BBC Micro).

What's the point in talking about mexican-bung-hole saving systems if it's impossible to implement?

Pointless waste of words.


Until I see autosave system put into TDM mission and used effectively - I don't see the point in continuing this discussion.
It's not like anyone will listen to anything, anyway. They do what they want, when they want (or have time), prioritise according to whim.


Also, Shadwen kinda sucked.

[Oktokolo]

So how many autosaves must it make and retain? 2? 20? 200? All of them?

The minimum count to prevent bad luck mishaps is depending on the time between autosaves and how buggy or badly designed the game or mission is.

The autosave interval is upper bound to the maximum time, the player would try to replay in case of a crash or mission-breaking bug. Its lower bound is given by the annoyance of the interruptions because saving is not done in background in TDM.

A reasonable default could be ten minutes.

For a ten minutes interval, three slots would provide a coverage of the last thirty minutes. But storage is plenty today, so i go crazy on it and recommend an insanely large default of ten slots.

Players will probably configure it to anything from off and no slots to every minute and a thousand slots.

[teh_saccade]

Until I see autosave system put into TDM mission and used effectively - I don't see the point in continuing this discussion.