EPTThe Games Industry And ATVI - Page 74
| Forum Index > General Forum |
|
Harris1st
Germany6796 Posts
| ||
|
Hat Trick of Today
79 Posts
On March 21 2025 07:27 JimmyJRaynor wrote: This is interesting... https://www.pcgamer.com/gaming-industry/pc-gamers-spend-92-percent-of-their-time-on-older-games-oh-and-there-are-apparently-908-million-of-us-now/ newer games used to offer better graphics and better AI enemies. Oddly enough, the enemy AI just ain't gettin' any better and neither are the graphics. Graphics are absolutely getting better every year, it’s just that we’ve reached the point of diminishing returns rather than them not actually getting better. This goes hand in hand with GPU performance increases plateauing despite continuous spikes in prices, especially in the low and mid end of the market. I’m still using my GTX1080ti because of the dual combination of dreadful GPU pricing and subpar raster performance increases. AI is in a different boat where players actually don’t see/notice the better AI and often prefer dumber AI. An example of the former is Warframe, where you don’t see what the AI is capable of because enemies always die too fast. Examples of the later are just about all RTS/4x/turn based strategy games because the ultimate goal isn’t to emulate a human being but to provide the player with a system they can exploit without noticing. So the player can feel good about doing the thing. There is also fine line between an AI feeling like a dumb human vs an AI feeling like an omniscient cheater. Elden Ring stuffed this up where they attempted to improve the responsiveness of the boss AI by having the boss input read you but in most cases just frustrated the heck out of the player. The player doesn’t actually want the boss to input cancel out of parry attempts, they want a gameplay loop where timing a successful parry results in a successful parry. The AI is also not infallible, humans are good ant pattern recognition and will still end up exploiting the boss AI once they understand the boss inputs reads. As the article you posted concludes for you, this is the reason why gaming companies so desperately look to creating a hit live service game because they’re clearly the games that hold vast majority of people’s attentions (and wallets). It isn’t any surprise that PC gamers spend 92% of their time on games older than 2 years old. The cash geyser that is Genshin Impact is something like 4 years old, Fortnite is almost 8 years old, Roblox, WoW and FFXIV are older than dirt. Even a game like Rainbow Six Siege, which can still get 6 digit active users, is 9-10 years old. What keeps people engaged in all of these games is content so that’s where the development resources go. Content is the one thing is that obvious to the player and casts the widest net, the one thing least limited to system specs or player preferences. | ||
|
DarkPlasmaBall
United States44019 Posts
On March 21 2025 07:27 JimmyJRaynor wrote: This is interesting... https://www.pcgamer.com/gaming-industry/pc-gamers-spend-92-percent-of-their-time-on-older-games-oh-and-there-are-apparently-908-million-of-us-now/ newer games used to offer better graphics and better AI enemies. Oddly enough, the enemy AI just ain't gettin' any better and neither are the graphics. I'm a sucker for retro games, pixel art, and early-3D polygons, but obviously the graphical capabilities have gotten way, way, way better for new consoles over the past 20-30 years. (I don't value realistic/HD graphics as much as, say, fun gameplay, but all you need to do is compare new-gen games to old-gen games to see that the graphics are like night and day.) | ||
|
KobraKay
Portugal4219 Posts
Which means that AI in those games is often ignored or left as barebones because of the abovementioned focus. | ||
|
JimmyJRaynor
Canada16596 Posts
On March 22 2025 07:56 DarkPlasmaBall wrote: I'm a sucker for retro games, pixel art, and early-3D polygons, but obviously the graphical capabilities have gotten way, way, way better for new consoles over the past 20-30 years. (I don't value realistic/HD graphics as much as, say, fun gameplay, but all you need to do is compare new-gen games to old-gen games to see that the graphics are like night and day.) Over the past 7 years graphics have not gotten better. Whereas, from 1977 to 1984, graphics on the SAME HARDWARE improved by orders of magnitude. 1977's Atari 2600 game line up and 1984 Pitfall2 do not look like they're the same console. Engineers started using stuff like Bresenham's line algorithm to greatly improve graphics on limited hardware. From an engineers perspective...what engineers did with the Atari 2600 was awe inspiring. I have little respect and zero awe for UE5 and what today's engineers bring to the video game industry. One exception to the "AI not getting better" was SC2's Alphastar. It was super cool. It is a crying shame Blizzard didn't keep a version of Alphastar playing in each league of the ladder. The thing was mothballed. Pretty sad. | ||
|
WombaT
Northern Ireland24264 Posts
Alphastar did what it set out to. And yeah was cool I’ve also never seen a confirmed cost for the project, although I’ve seen estimates of about 12 million dollars just for the computation to train it. Never mind the additional expenditure of the expertise to direct it Equally, it never really succeeded in emulating an actual human player. It played strange styles, it did things humans couldn’t do reliably to do this. In most single player games, as Hat Trick says, we don’t want the AI to be too good. And in multiplayer we want the AI to emulate human players we’d be facing so we can practice vs like a bronze human, or a plat human or whatever. Stealth games are a genre I’m a big fan of, to pick one. Those games only work because the AI is somewhat predictable and doesn’t learn much. There’s a facsimile of realism but it is very much a facsimile. If the AI becomes smart enough to discover the standard tricks of the trade in the genre such as ‘hide reasonably close by until the heat dies down and try to not get spotted next time’ and thus notices this and never drops alert status until you’re found, as would actually happen IRL, well man good luck with that game. If patrolling guards have fully independent and bespoke AI-trained patrol routes, sure more realistic for sure, more fancy tech! Except it may become borderline impossible to even get through a room unspotted. Plenty of games need relatively dumb AI to function. They won’t actually be improved at all by making it smart. It’s easier for designers, it’s easier for players. It’s more fun for players in many instances. A video game is really just a puzzle, you’ve got your goals, you’ve got to figure out what the rules of your game world are, and how to achieve them within that. Could be goombas move like this, I can jump on them. Could be x demon does y, shoot those demons like in Doom. Could be some more complicated interactions in other genres. Any kind of genuinely smart, or adaptive AI starts fucking with this tried-and-tested model. | ||
|
DarkPlasmaBall
United States44019 Posts
On March 23 2025 21:24 JimmyJRaynor wrote: Over the past 7 years graphics have not gotten better. Given your moving of the goalposts, I will happily accept your revision (although, technically, it's still really just diminishing returns in most cases, as WombaT described). | ||
|
Harris1st
Germany6796 Posts
On March 22 2025 06:05 Hat Trick of Today wrote: Graphics are absolutely getting better every year, it’s just that we’ve reached the point of diminishing returns rather than them not actually getting better. This goes hand in hand with GPU performance increases plateauing despite continuous spikes in prices, especially in the low and mid end of the market. I’m still using my GTX1080ti because of the dual combination of dreadful GPU pricing and subpar raster performance increases. AI is in a different boat where players actually don’t see/notice the better AI and often prefer dumber AI. An example of the former is Warframe, where you don’t see what the AI is capable of because enemies always die too fast. Examples of the later are just about all RTS/4x/turn based strategy games because the ultimate goal isn’t to emulate a human being but to provide the player with a system they can exploit without noticing. So the player can feel good about doing the thing. There is also fine line between an AI feeling like a dumb human vs an AI feeling like an omniscient cheater. Elden Ring stuffed this up where they attempted to improve the responsiveness of the boss AI by having the boss input read you but in most cases just frustrated the heck out of the player. The player doesn’t actually want the boss to input cancel out of parry attempts, they want a gameplay loop where timing a successful parry results in a successful parry. The AI is also not infallible, humans are good ant pattern recognition and will still end up exploiting the boss AI once they understand the boss inputs reads. As the article you posted concludes for you, this is the reason why gaming companies so desperately look to creating a hit live service game because they’re clearly the games that hold vast majority of people’s attentions (and wallets). It isn’t any surprise that PC gamers spend 92% of their time on games older than 2 years old. The cash geyser that is Genshin Impact is something like 4 years old, Fortnite is almost 8 years old, Roblox, WoW and FFXIV are older than dirt. Even a game like Rainbow Six Siege, which can still get 6 digit active users, is 9-10 years old. What keeps people engaged in all of these games is content so that’s where the development resources go. Content is the one thing is that obvious to the player and casts the widest net, the one thing least limited to system specs or player preferences. I would like to add to that. Content is of great importance, but for most of those "old" games, what's even more important is community. That can be directly (friends, guildmates, raids) or more indirect (reddit, discord). It gives the player a feel of belonging and the player would have to give that up when starting a new game until he found a new group and that new sense of purpose. So for a new game to come along they have to either make it so good and endearing for players to switch or even better, make the whole group switch | ||
|
Silvanel
Poland4700 Posts
On March 22 2025 06:05 Hat Trick of Today wrote: Examples of the later are just about all RTS/4x/turn based strategy games because the ultimate goal isn’t to emulate a human being but to provide the player with a system they can exploit without noticing. So the player can feel good about doing the thing. I call BS on that. In most strategy games AI isnt even able to properly mange its own economy. It has to be fed free resources in order to be competitive with humans. In many games AI doesnt even play by the same rules as human because noone had the resources/will to code that. The above was true for all strategy games I have played recently, meaning: 1)Stellaris - a HUGE upkeep reduction on higher diffculties + advanced start options (more things on start than player is getting) 2)Old World - as above, with more ridiculous advanced start on higher difficulties 3)Civ VI - free resources on higher difficulties + again, advanced starts AI in strategy games is also often coded to speciffcally counter human playstayles and not to play good on its own (because the former thing is easy and latter is very very complicated). Like in Stellaris some AI empires will overproduce fleets in order to try to kill humans. Doesnt really matter if they tank their economy and research and lose all hopes of success against other AIs. Their goal is to be obstacle to player, if they harm the player they have succeded, human will likely leave the game. They have achieved the goal set before them (which again is not being good at game, because thats is much harder and costier to achieve. | ||
|
Hat Trick of Today
79 Posts
I can’t say for sure if the AI in those games are especially awful to the point that they don’t actually do anything beyond the one thing they’re scripted to do. I think that’s what you are alluding to? I haven’t played those games so I can’t speak to whether or not those AI patterns are a huge enjoyment detriment to a player who hasn’t seen how the sausage is made so to speak. But these specific examples of awful AI still don’t dispute the fact developers just don’t put a whole lot of effort into AI because players want to win and want the dopamine hit you get by (subconsciously or consciously) exploiting the AI. Yes they don’t put a lot of development effort into more complex AI because a lot of the time it will only be appreciated by a smaller segment of the market - it’s not a priority for the industry at large because the wider market just doesn’t care much for it. In most games, there’s really no reason to put a lot of effort into better AI when you can just obfuscate unfair advantages by making sure the actual game’s design is aware of them. Sekiro and Elden Ring both have absolutely gross input reading bosses but it isn’t as obvious that Sekiro bosses input read because you’re rarely animation locked into getting punished. | ||
|
Silvanel
Poland4700 Posts
AI in strategic games is dumb because companies don't want to put too many resources into its development. It took many years and millions of dollars to put chess engines to the point where they are now. And chess is a very simple problem to solve by comparison. It has no fog of war, You have full knowledge of Your enemies' movement and resources, resources to don't change, it is symmetric, there is no stochastic element, there are only two players and so on and so on... | ||
|
Hat Trick of Today
79 Posts
Granted these statistics don’t have that much nuance, like people just not liking the game and quitting, but Brad Wardell did once claim that like 90% of people didn’t move on from the default difficulty in Galactic Civilizations III. It’s all anecdotal but that figure makes sense to me. I know a lot of people in the past didn’t really engage in the Civilisation games with any particularly deep thought - would these be the same people who would have a serious problem with the quality of AI in the current 4x games? I dunno, maybe it is just that bad and noticeable but most are really just happy to just do their thing. | ||
|
JimmyJRaynor
Canada16596 Posts
On March 24 2025 22:59 Hat Trick of Today wrote: All AIs in video games merely exist as a puzzle for the player, not to emulate a properly reactive human being.. that depends on who is making the game. Some game makers want more human like AI behaviours. https://www.gamedeveloper.com/console/krafton-debuts-generative-ai-companions-that-behave-in-human-like-manner- On March 24 2025 22:59 Hat Trick of Today wrote: I haven’t played those games so I can’t speak to whether or not those AI patterns are a huge enjoyment detriment to a player who hasn’t seen how the sausage is made so to speak. Pacman and Ms Pacman broke the bank with enemy AI nuanced enough to create a different personality behind the 4 ghosts. "Shadow" the Red Ghost makes decisions based on getting behind the player and out running him or her on long straight-aways. The Pacman games of 1980 and 1981 were pure genius. On March 23 2025 23:26 DarkPlasmaBall wrote: Given your moving of the goalposts, I will happily accept your revision (although, technically, it's still really just diminishing returns in most cases, as WombaT described). Put it this way... Consumers spend most time playing games in 2024 having graphics close to and possibly even worse than 2016 graphics. Graphics could be worse due to the immense popularity of the Nintendo Switch. Consumers are opting for worse graphics because the top potential for consumer graphics, the PS5Pro hardly raises the ceiling at all. Did consumers suddenly start playing Pong in 1982 with games like Donkey Kong Junior, Pole Position and Zaxxon around? We can see why that did not happen. The graphical difference between Zaxxon and Pong ... its a different world. This contrasts the years 2019 to 2023 when the Nintendo Switch did great with ancient graphics. | ||
|
Fleetfeet
Canada2512 Posts
Your argument is evolving! ... Congratulations! Your "Graphics Aren't Getting Better" has evolved into "Who Cares About Graphics Anyways, lets just play Mahjong"! | ||
|
Yurie
11742 Posts
I do not think we have gotten any major freedoms since 2016. Graphics have improved for top titles and that draws some sales. But an open world such as Witcher 3 is still good and has mostly the same freedoms as Assassin's Creed Shadows. The latter looks better if you are on good hardware but basic possibilities on what you can do has not changed much. Even Skyrim from 2011 had mostly the same ones and thus stands the test of time in a decent way. I think separating game play possibilities from graphics is relevant. For most people the game play possibilities are more important than graphics. But graphics does still sell. The options on what you can do and how much it costs at previous graphics levels are slowly improving. The cost to improve graphics over previous levels is just rising much faster so on a cost benefit ratio for players spending that extra budget on polish over graphics likely sells. Switch has a lot of development cost reduction due to being on 10 year old hardware (in cases where you aren't converting a modern game using ray tracing for lighting). | ||
|
JimmyJRaynor
Canada16596 Posts
" Facial expressions? Extensive Voice Commentary! WOW! This will never be on Intellivision, Atari computers, or Colecovision". regarding how graphics have "improved" in the last 5 years... "If you're explaining.. you're losing" On March 27 2025 12:32 Fleetfeet wrote: What? Your argument is evolving! ... Congratulations! Your "Graphics Aren't Getting Better" has evolved into "Who Cares About Graphics Anyways, lets just play Mahjong"! Graphics are not getting better. check out the PS5 Pro. Ubisoft, I wonder if this is a factor in Ubisoft selling their Assassin's Creed IP to a subsidiary. Now, the Tencent overloads can apologize and blame anyone within Ubisoft proper in order to give Japan their pound of flesh. They don't even have to fire whoever is to blame... they just do not get hired into the subsidiary that controls AC. Clearly, the #1 factor in the new Tencent/Ubi subsidiary is Ubisoft needs cash to continue its operations. How does Ubisoft manage to get Japan pissed off at them? LOL. That presentation where they called AC a realistic retelling of Japanese history was a super dumb move and a super arrogant move. Ubisoft should've said, right from the start, "it is all fantasy guys ... we're just a bunch of video game dorks making up stupid stories". Instead, Ubisoft demanded we take their AC Shadows story ... "seriously".  Part of what makes most video game stories endearing is how silly they are. A fat italian plumber rescues his girlfriend from a giant ape? Aweome. The game's name comes from a bad // impossible translation from Japanese to English. Awesome! | ||
|
WombaT
Northern Ireland24264 Posts
Graphics are getting better, but they’re not making big ‘wow’ jumps, because returns are diminishing there. If you’d, I don’t know, actually read people’s posts you wouldn’t need to make your own arguing against people who already made the exact same points and agree with the points you’re making. | ||
|
JimmyJRaynor
Canada16596 Posts
On March 31 2025 21:32 WombaT wrote: Jimmy literally nobody in thread actually disagrees with that, everyone has said basically exactly that. Graphics are getting better, but they’re not making big ‘wow’ jumps, because returns are diminishing there. If you’d, I don’t know, actually read people’s posts you wouldn’t need to make your own arguing against people who already made the exact same points and agree with the points you’re making. i'm glad we all agree. However, I'd say graphics are same because we are requiring more computing horsepower and expense in order to deliver improvements that are almost unnoticeable. Thus, when cost is factored in graphics are staying the same. Whereas, by 1990 you could play Arcade Punch Out on cheap hardware. It is debatable whether or not UE5 is better than UE4. https://forums.unrealengine.com/t/large-performance-regression-in-ue5-cpu-performance/1524868 in the past, graphics got substantially better on the same hardware. | ||
|
Fleetfeet
Canada2512 Posts
On March 31 2025 21:32 JimmyJRaynor wrote: in the past, graphics got substantially better on the same hardware. You should read the previous post(s) where WombaT said things like "Diminishing returns" and "You should actually read the words of the other people you're having a conversation with, so you can avoid saying 'but actually' followed by a point one of them have already made." But actually, graphics have gotten substantially better on the same hardware. | ||
|
Harris1st
Germany6796 Posts
On March 31 2025 21:32 JimmyJRaynor wrote: i'm glad we all agree. However, I'd say graphics are same because we are requiring more computing horsepower and expense in order to deliver improvements that are almost unnoticeable. Thus, when cost is factored in graphics are staying the same. Whereas, by 1990 you could play Arcade Punch Out on cheap hardware. It is debatable whether or not UE5 is better than UE4. https://forums.unrealengine.com/t/large-performance-regression-in-ue5-cpu-performance/1524868 in the past, graphics got substantially better on the same hardware. For the most part, more horsepower is needed because there is more stuff on the screen. More NPCs, more trees, more flowers, more houses, more mountains, wider range of view, longer range of view, more shadows, better shadows. More fucking leaves on trees moving independantly. Things are not better looking than 10 years ago. Just more of it. That is where your horsepower goes. And sometimes the games are just horribly optimized and your horsepower just goes down the drain | ||
| ||
