Saving in Games: Part 4 - Temporary Savings
Before I call it a day on this "Saving In Games" topic, I wanted to bring one last idea to the table. I've tried to establish why current save systems in games are horribly inconvenient: we should be able to stop playing a video game at any point in time, not only when we find or reach a Save Point. I do not want to live a life where my main hobby requires me to yell at friends and family, "Give me five more minutes! I need to save!"
So while I addressed one solution (giving all consoles a sleep mode) let's be realistic: it ain't happening any time soon. Because we know this to be true, current game designers can't really count on this system to make their games more convenient to play. So is there anything game designers can do now to help soften the problem with saving in games?
I think there is something they can do. Right now, the way games stand at this moment, what is the biggest fear that game designers have about Save Points? Easy, I've covered that already: letting people save whenever they want potentially makes a game too easy. But we need to dive into this issue just one more level deep: how does it make games too easy?
Well, if you can save at any point in time, you will always save your best performance. Then you can play the game from there and, if you die, struggle, or play poorly, you can load up your previous save and try again. If you do the next section well, save after that and continue this pattern. The problem is that these sections can be 10 second intervals. Anyone who has played emulated games on a PC, for example, knows how powerful Save States are. If you successfully navigate one tricky platforming section that involved three difficult jumps, you can save your state immediately afterwards and then attempt the next section. If you fail the next three tricky jumps, you can always start from that saved state and try again. Many action games are designed to force a player to pass a slew of challenges in a row. They want you to make all 6 tricky jumps in a row. That's where they want to have their challenge come from. Saved States ruins this and makes this type of game too easy.
So we are at an impasse then: allowing players to save at any point in time makes games too easy, but not letting players save any time they want is too inconvenient. What can be done?
Actually, once again the solution is already out there and has already been implemented. The first time I personally experienced it was with The Legend of Zelda: Majora's Mask. Since then, I've seen it in games like Castlevania: Dawn of Sorrow and Super Princess Peach. It's a fairly elegant solution: the non-persistent saved game.
The concept behind the non-persistent saved game is that, upon being loaded, the saved game erases itself from memory. Thus, you can't use it as a repeatable starting point. If you load the game and then perform the next section badly, you can't reload the last saved game because that saved game no longer exists. This completely eliminates the above problem of making games too easy. You're not providing players with a safety net, you're just giving them a way they can stop their game and continue without leaving their systems on.
A very obscure example of how this system can help is with Koei's historic war simulation games: the Romance of the Three Kingdoms series. I bring these games up, even though no one has played them, mainly because it is a particularly sour spot with me. These games are fun and my brother enjoys them quite a bit, but they are tough to play. See, they all comprise of two halves: domestic tasks and wartime tasks. You can save at any given time during the domestic tasks, but the wartime tasks? You're out of luck here. Once you send your troops into battle, you have to play out the entire battle. There are no saves during the wartime tasks. But that presents a problem: one of the fun aspects of these games is that you can play multiplayer, so you can try to fight for control of China against your friends. However, when two humans play the battles in these games, because of the pacing and level of strategy implemented in these battles, they can sometimes take up to 2-3 hours to complete!
I will tell you this: if you have to plan ahead and put aside 3 hours every time you want to go to battle in this game, it's a poorly designed game.
Koei doesn't want to let you save in battle because, during the course of battle, a lot of bad things can happen: your general gets captured, the enemy successfully lands a ruse on your commander, fire spreads onto your troops and burns a ton of your army away, etc. And the game would be pretty pointless if you can save anytime in battle and reload if something unfortunate happens to your army. If you could do that, you could play the battle entirely safe. So to this day, Koei has never allowed you to save mid-battle and, thus, are forced to play for three hours straight sometimes. But if you implement the non-persistent save, it'd be fantastic to be able to save in the middle of a battle and, because the saved games aren't persistent, Koei's fear of a safe battle is assuaged. This is a perfect example, albeit obscure, of how non-persistent saves can improve gaming in general.
This solution, actually, is better than consoles with sleep modes. It allows you to turn off games and play other games in the meantime. My friend Eric told me a story of how he put his DS into sleep mode and later found his girlfriend had used his DS and started playing a different game, losing the game he had currently sleeping. In these situations, sleep mode obviously doesn't help at all. But non-persistent save states do.
Of course, the first question that comes up is this: what if you die? What if the power goes out? What if your portable system's battery dies? With non-persistent saves, does that mean you've just completely lost your entire game? Could you potentially lose 60+ hours of your latest Final Fantasy game?
Yes, you would. Which is why the non-persistent saved game system should only be implemented on top of the save systems we know today. Keep games with the occasional hard save points. These are the ones that persist. Thus, you can save outside of a dungeon in a Final Fantasy game at a Save Point, creating your persistent game. Then, once inside the dungeon, you can save at any given time to your non-persistent saved game. If ever you have a power outage or your party dies or some other catastrophic event occurs, you will continue your game from outside the dungeon, where you saved your persistent saved game.
Again, this system has been used in other games, and I find it odd that it hasn't been used more often. The problem is that it was used in games like Dawn of Sorrow and Princess Peach to compensate for problems associated with portables. Once again, I think it is a gaming convention that should just be used in general with all games. It solves every problem and creates no new ones as far as I can tell.
Unless you count the programmatic difficulties involved in saving a game at any time you want. I'll fully admit that this is a difficult task for game programmers to accomplish, particularly with games that have a lot going on at once. But I do think it's possible and is worth the effort.
Games are entertainment. They are here to give us a fun way to enjoy our time. They shouldn't be something that causes us or those we know unnecessary frustrations. If the industry just worked at it, the entire problem of not being able to stop your game at any given time can be solved. Hopefully, people will realize that this is a pretty significant problem, and hopefully game designers will take the time to solve this problem.
Thanks for reading, and have a great weekend.
- jchensor
So while I addressed one solution (giving all consoles a sleep mode) let's be realistic: it ain't happening any time soon. Because we know this to be true, current game designers can't really count on this system to make their games more convenient to play. So is there anything game designers can do now to help soften the problem with saving in games?I think there is something they can do. Right now, the way games stand at this moment, what is the biggest fear that game designers have about Save Points? Easy, I've covered that already: letting people save whenever they want potentially makes a game too easy. But we need to dive into this issue just one more level deep: how does it make games too easy?
Well, if you can save at any point in time, you will always save your best performance. Then you can play the game from there and, if you die, struggle, or play poorly, you can load up your previous save and try again. If you do the next section well, save after that and continue this pattern. The problem is that these sections can be 10 second intervals. Anyone who has played emulated games on a PC, for example, knows how powerful Save States are. If you successfully navigate one tricky platforming section that involved three difficult jumps, you can save your state immediately afterwards and then attempt the next section. If you fail the next three tricky jumps, you can always start from that saved state and try again. Many action games are designed to force a player to pass a slew of challenges in a row. They want you to make all 6 tricky jumps in a row. That's where they want to have their challenge come from. Saved States ruins this and makes this type of game too easy.
So we are at an impasse then: allowing players to save at any point in time makes games too easy, but not letting players save any time they want is too inconvenient. What can be done?
Actually, once again the solution is already out there and has already been implemented. The first time I personally experienced it was with The Legend of Zelda: Majora's Mask. Since then, I've seen it in games like Castlevania: Dawn of Sorrow and Super Princess Peach. It's a fairly elegant solution: the non-persistent saved game.
The concept behind the non-persistent saved game is that, upon being loaded, the saved game erases itself from memory. Thus, you can't use it as a repeatable starting point. If you load the game and then perform the next section badly, you can't reload the last saved game because that saved game no longer exists. This completely eliminates the above problem of making games too easy. You're not providing players with a safety net, you're just giving them a way they can stop their game and continue without leaving their systems on.
A very obscure example of how this system can help is with Koei's historic war simulation games: the Romance of the Three Kingdoms series. I bring these games up, even though no one has played them, mainly because it is a particularly sour spot with me. These games are fun and my brother enjoys them quite a bit, but they are tough to play. See, they all comprise of two halves: domestic tasks and wartime tasks. You can save at any given time during the domestic tasks, but the wartime tasks? You're out of luck here. Once you send your troops into battle, you have to play out the entire battle. There are no saves during the wartime tasks. But that presents a problem: one of the fun aspects of these games is that you can play multiplayer, so you can try to fight for control of China against your friends. However, when two humans play the battles in these games, because of the pacing and level of strategy implemented in these battles, they can sometimes take up to 2-3 hours to complete!
I will tell you this: if you have to plan ahead and put aside 3 hours every time you want to go to battle in this game, it's a poorly designed game.
Koei doesn't want to let you save in battle because, during the course of battle, a lot of bad things can happen: your general gets captured, the enemy successfully lands a ruse on your commander, fire spreads onto your troops and burns a ton of your army away, etc. And the game would be pretty pointless if you can save anytime in battle and reload if something unfortunate happens to your army. If you could do that, you could play the battle entirely safe. So to this day, Koei has never allowed you to save mid-battle and, thus, are forced to play for three hours straight sometimes. But if you implement the non-persistent save, it'd be fantastic to be able to save in the middle of a battle and, because the saved games aren't persistent, Koei's fear of a safe battle is assuaged. This is a perfect example, albeit obscure, of how non-persistent saves can improve gaming in general.This solution, actually, is better than consoles with sleep modes. It allows you to turn off games and play other games in the meantime. My friend Eric told me a story of how he put his DS into sleep mode and later found his girlfriend had used his DS and started playing a different game, losing the game he had currently sleeping. In these situations, sleep mode obviously doesn't help at all. But non-persistent save states do.
Of course, the first question that comes up is this: what if you die? What if the power goes out? What if your portable system's battery dies? With non-persistent saves, does that mean you've just completely lost your entire game? Could you potentially lose 60+ hours of your latest Final Fantasy game?
Yes, you would. Which is why the non-persistent saved game system should only be implemented on top of the save systems we know today. Keep games with the occasional hard save points. These are the ones that persist. Thus, you can save outside of a dungeon in a Final Fantasy game at a Save Point, creating your persistent game. Then, once inside the dungeon, you can save at any given time to your non-persistent saved game. If ever you have a power outage or your party dies or some other catastrophic event occurs, you will continue your game from outside the dungeon, where you saved your persistent saved game.
Again, this system has been used in other games, and I find it odd that it hasn't been used more often. The problem is that it was used in games like Dawn of Sorrow and Princess Peach to compensate for problems associated with portables. Once again, I think it is a gaming convention that should just be used in general with all games. It solves every problem and creates no new ones as far as I can tell.
Unless you count the programmatic difficulties involved in saving a game at any time you want. I'll fully admit that this is a difficult task for game programmers to accomplish, particularly with games that have a lot going on at once. But I do think it's possible and is worth the effort.
Games are entertainment. They are here to give us a fun way to enjoy our time. They shouldn't be something that causes us or those we know unnecessary frustrations. If the industry just worked at it, the entire problem of not being able to stop your game at any given time can be solved. Hopefully, people will realize that this is a pretty significant problem, and hopefully game designers will take the time to solve this problem.
Thanks for reading, and have a great weekend.
- jchensor
5 Comments:
Two problems that arise from ephemeral saves:
1. You can't save before a major plot branching and try both paths. The advantages of being able to do this (rather than having to replay the entire game for each desired plot) are obvious.
2. You can't save close to the end of the game and revisit it later. With a permanent save point, you can save your level 99 character to show off, save at the coolest boss fight to play it over and over, or just save at the end of the game to show your final time and watch the ending again.
Possible solution: Standard save points, plus an ephemeral quicksave. Halo is a good example - you can save at major places and continue from fairly close to where you left off, but you can't just hit F5 every ten seconds.
By
Anonymous, at 5:45 AM
Yup, 100% agreed. I did mention that the ephemeral saves should always be implemented on top of the standard save system. A lot more problems arise with only ephemeral saves than just the two you've listed. So I completely concur: have the non-persisten saves coupled with your standard saves, and you'll definitely have the best system.
- James
By
jchensor, at 11:07 AM
Hi James,
The main problem for ephemeral game saving you have described has been / is memory size constraint.
When a save point is placed by a designer or the game is saved at the end of a level, the state of the game is predictable by the designers and the programmers. They know what to store in the memory so they know how big the memory should be allocated.
This applies to Halo-type checkpoints as well. Checkpoints usually make sure that there are no enemies on the screen. Checkpoint saves do not take much memory because it doesn't need to. It only saves the progress of the game and few game entity states.
Quicksaves are not predictable, so how quicksaves are implemented is the system tries to save the current state of all entities. If you have 100 guys on the screen and you have to save the AI state, the animation state, and misc stuff for each guys. How about particles on the screen?
That's action games. How about RPGs? RPG save games need to save states of not only the characters on the screen, but some NPC in a town you visited while back because it is possible for you to save your game before a boss fight, leave the dungeon, and talk to a NPC in distant town about a side quest. How much maximum memory should the programmers allocate to accomodate this?
Your solution of ephermal saves would only work for certain types of games. I could imagine it working for action, adventure, linear RPGs, etc. Any opened ended RPGs or RTSs would not work... X360 and PS3 got 512MBs each. But take away 10-20% for system reserved and executables. PS2 and XBox did not have enough memory to do ephemeral saves without a big sacrifce. If the developers are willing to sacrifice few megs of memory for this feature (and the time dedicated to implement this feature), I'm all up for it.
Keep up the good work on your blogs, James!
- Hyun
By
Anonymous, at 2:04 PM
Yeah, that's the main thing I was afraid of: how difficult it would be to actually implement ephemeral saves (is that their official term?) from the programmer's view point. I mean, having that type of save for a game like Dynasty Warriors would be nigh impossible. There are just too many things going on on the screen and behind the scenes to save. Saving off the state of so many threads is probably impossible without developing a smart system on how to do it. And even the smartest of systems won't be able to save that kind of information.
It's gonna have to be a game-by-game evaluation. I think a game like Prince of Persia, for example, should be really easy to do this for (though, with more frequent save checkpoints, probably unnecessary). And I think for Final Fantasy games, it shouldn't be that hard, particularly if they only allow it in dungeons, since they are usually pretty self-contained. Also, if RPGs use an ephemeral saved game mainly as a layer on top of the permanent saved game, that might help. So the meat of the information saved is still in the normal save. But the ephemeral saved game just adds more information on top of that save, almost acting like a "diff" between the saved state and the current state, and the game interprets from there.
Obviously, implementing these kinds of saves would require a lot of assumptions upon reload and, quite possibly, a lot of sacrifices (saving the states of minor things like particles and such). For many games, it may be wholly impractical to even attempt such a thing, but for those games where it's possible, I'd love to see it become a norm.
- James
By
jchensor, at 4:07 PM
oh nice, i didn't realise some games had already implemented it! great article!
joey
By
Anonymous, at 4:29 PM
Post a Comment
<< Home