Messages

16

General Discussion / Re: Netcode Lag/Delay Test Reports

« on: November 27, 2011, 07:38:24 AM »

I think the best testing is done this way:

- Get a pair of players who own several recent FGs, including well regarded netcode BB, and big popular game SF4AE.
- Try playing each FG together and compare the experience with KOF13.

This makes it easier to spot issues with your setup.  If all FGs work poorly for you, then KOF cant be expected to do any better.  Similarly, if SF4 and BB work better against the same opponent under the same conditions, then its a powerful argument for Atlus/SNK that KOF13 is not performing up to industry standards.

--------------------------

Personally, regarding a patch, I think the very first thing we should demand is an "advanced network info" mode.
You turn it on in the options, and when you do:

- You see a true ping value rather than the mobile phone bars.

- During the match, there is a banner on the top/bottom of the screen, that constantly displays connection information:
Average Ping so far.
Worst ping so far.
Packet loss stats.
Current Input Delay setting.
(if there are any "special events" that occur from the netcode, like resyncing or waiting for packets or whatever, then icons will appear in the banner to indicate when it is happening.)


This will provide much needed information to help give feedback, and also allow people to properly understand what is going on under the hood in the netcode.

17

General Discussion / Re: Netcode Lag/Delay Test Reports

« on: November 26, 2011, 04:19:40 AM »

Here we go again.  Every time GGPO gets discussed people start parading the same ignorance and misinformation.

Rollback netcode works for both 2D and 3D games.  In fact, the concept of rollback was inspired by the netcode techniques used in first person shooters.

Rollback netcodes have a fairly high overhead in processing.  When a rollback occurs, the game has to process multiple simulation frames instantly to figure out the new gamestate.   Its possible that newer games, which already fail to run at perfect framerate on consoles, dont have the spare processor capacity to run a rollback system.  But it's also possible that framerate drops are due to GPU being maxed out, so there is spare CPU.

GGPO can work on any game in an emulator, because the emulator can trivially save the entire gamestate and restore it, and also run simulation frames without drawing, which is fundamental to doing rollbacks.  However, a game that wants to implement GGPO natively can be internally designed to save/restore/run its gamestate, and often it can do so much more efficiently than an emulator.  Since it can be more selective about what data it wants to remember/process.  Some games written for PC are designed this way from the beginning, so that they will run simulation frames ("ticks") seperately from visual frames.

GGPO can theoretically be added to any game.  However, the design of the game engine and other limitations can make doing so require far more work than a game dev wants to spend.  Factors include:
- Whether the engine can be easily recoded to store multiple frames of gamestate and quickly run simulation frames faster than visual frames.
- Whether the game has been optimised enough to leave a lot of spare CPU power for the netcode.

If the game engine isnt convenient for those 2 things, then adding GGPO would require a lot of work, or even be impossible. (Certainly there is a limit to how much you can optimise, so it might be that some games cannot use GGPO on current consoles without cutting physics/graphics/sound quality etc.)

2D games are actually more hardware intensive than 3D games these days - modern consoles are optimised for 3D graphics, not for rendering very large 2D sprites.  Low RAM is especially a problem.  So, if SNK had to do on the fly sprite decompression to fit the sprites into RAM, then that would use up a lot of CPU, for example.

---------------------------------

So anyways, thats the truth as far as I know it.  I'm not an expert on netcode, but I am a professional software engineer and I've been interested in GGPO and rollback netcode. So I think I have a basic understanding of some of the issues involved.

18

King of Fighters 98/UM/FE / So, uh, 98FE??

« on: September 09, 2011, 02:56:35 PM »

Wow, I dont think I've ever seen a version of a game be ignored quite so thoroughly.

Does anyone know anything about it?  Is it better/more balanced than 98UM?  Or is it worse?
Is it popular at all, anywhere?

19

General Discussion / Re: The King of Fighters XIII 3rd |OT| Thread:Console Edition (UPDATE 8/26 1st page)

« on: September 08, 2011, 03:58:12 PM »

What, fighters arent allowed to enjoy themselves at the beach?  Or are they expected to wear their fighting costumes wherever they go? :P

20

General Discussion / Re: The Future of KOF Story base games

« on: July 04, 2011, 05:14:27 PM »

Theres no need for yearly releases in the modern age.  Just patch the game every now and then with free updates, and paid DLC characters, and release a compilation disc at the end.

Do we really want to go back to the bad old days of KOF99-2001?

Starcraft is a better model than Madden - stick to refining the gameplay iteratively rather than throwing everything out all the time.


As for story, it would maybe be more interesting to have DLC "episodes", that integrate into the fighting engine.

Each episode is like a visual novel and tells seperate complete story from a team or character's point of view.  (and ditch the "yearly tournament" format, its too limiting.)  This means you can have much more freedom telling a proper story - it can be completely unrelated to the KOF tournament.

Of course, we hate DLC and having to pay for it, but considering voice acting and artwork is probably the biggest content cost besides new characters, its probably not wise to burden the cost of the main game with what many players consider to be an optional extra. (and then you dont need to buy any stories about characters you dont care about.)

21

General Discussion / Re: Questions for Atlus - READ FIRST POST BEFORE REPLYING

« on: June 13, 2011, 08:26:59 PM »

Please think of the improvements in KOFXII's netcode post-release as movement in the right direction and a positive FOUNDATION for things to come.

The online functionality in KOFXIII is still VERY early in development and is, more importantly, a HUGE priority.

As soon as we're all decompressed from E3, we'll get to work reviewing your questions and compiling your feedback!

-Aram Jabbari
PR Manager, ATLUS

If the netcode in 13 is still "early" in development then please get SNK to wipe everything, license GGPO and implement, even if it means significant rewriting of the game's code.

I don't know what market ATLUS is targeting.  Ignition seemed to be going for the mass market with 12, whereas Capcom's 3rd Strike Online Edition is obviously aiming for the hardest of the core.

For the latter audience (at least in the West) simply having the letters GGPO in the game is probably the best marketing possible.  The importance will only increase as other titles bring greater awareness of this brand / technology amongst players.


SNK has already lost an immense amount of goodwill amongst players from its previous titles - this KOF is perhaps their last chance, and they shouldnt waste it.  It seems SNK may not be in the best financial position these days, but another barebones port with barely playable netcode is definitely not going to save them, and may instead be the final nail in the coffin.  Adopting a Blizzard-like "when its done" approach, and implementing GGPO (and other industry-best-practice online features) will be a good first step towards recovering their reputation.  I don't know the business arrangement Atlus has with SNK, but if Atlus is acting as a traditional publisher (as opposed to just a distributor), then I hope some funding might be diverted purely for this purpose.


Traditionally japanese devs havent placed much priority on online technology.  Atlus has made a good name for itself as a company willing to listen to its players.  Hopefully they can succeed where Ignition had failed, and convince SNK to finally innovate (or at least match its best competitors) in this crucial area.  It might also be valuable to look outside of the narrow world of Japanese arcade fighting games, and take a look at what other genres have been doing online - Starcraft 2, MMOs, FPS games, Steam, etc.


In game features to help build local communities, like regional lobbies or chat channels, online multiplayer training rooms for mentoring, eSport features like auto-tournaments and video/replay "TV" channels, etc will aid players in promoting the game to their friends.  In Japan fighting games have survived because of the prevalence of arcade centres there.  This kind of infrastructure is missing in most of the world, so it's necessary to duplicate it online in order to aid players in finding others and growing their local communities.


You may want to start recruiting players from around the world (or at least the US) to participate in a closed beta.  Honest, real-life feedback will be key to really demonstrating whether 13's netcode is comparable to its competitors.

(ah... I realise this post has no questions and is just a rant. Oh well.)