Friday, November 18, 2005

Beyond Polycount

Todays and even tomorrows games don't look very real and pumping up the polycount won't help. What games are laking graphical wise these days are three things in my eyes:

1) proper lighting, shadows are still the sharp ugly ones which we already had in the day of Starfox on the SNES, that just isn't very realistic, HDR helps quite a bit, but what is needed is some kind of realtime radiosity to get away from that odd-computer graphics look

2) motion, stil frames these days often look a lot better then the thing in motion, for the simple reason that animation is still a huge problem, motion capturing works fine for cutscenes, but in dynamic scenes it just doesn't look very good to always see the same prerecorded animation, beside from that it often is simply the wrong animation (classic example would be a player character walking against the wall, simply wouldn't work in reallife that way). Some kind of adaptive animation system is needed here, something that not only plays a prerecorded motion, but more or less simulates the human body.

3) information density, again not really a rendering thing, but what I mean with that is that the amount of 'information' that is presented in a game is nowwhere near reality. If I look around in a real room I might find shelfs full of books, all of which readable, cabinets full of cloths, all of them wearable, computers full of files, all of them browsable. In a game on the other side I might be able to find a table, a chair and a shelf with a few empty boxes, if there ever is a book in a game, I am happy when I can read a few pages of it, if at all. Same is true for games that play outside, GTA might give you a whole city, but each house is nothing more then a textured box, you can't walk into most of them, people that walk around on the streets are generated completly random and neither have goal or purpose. Sure, an artist probally will never close this information gap, but ProjectGutenberg might be able to fill the books with text and some kind of fractal algorithm should be able to build a wide varity of houses and rooms that are explorable, Elite did that a two decades ago and presented the player with a whole universe to explore, while it wasn't the most detailed universe, todays hardware should be able to accomplish quite a bit more.

Thursday, September 01, 2005

Nintendo Revolution Controller

Since everybody and his dog is speculating on the new Nintendo Revolution Controller, here a few ideas of myself:



The basic shape of the controller would be similar to that of the Gamecube, however with the following differences:
  • instead of four face buttons, the controller would feature six (this should help N64 games)
  • the C-Stick gets replaced by a C-Trackball/C-Touchscreen device
  • the controller gets additional L2 and R2 buttons
  • Start and Select are back in there classical places, no more ugly Z-button as Select replacement
  • the center of the controller would be rotatable like the Microsoft Dual Strike or the NegCon, this gives two additional axis of movement
Some of this of course contradicts with Nintendos saying of having a controller that is especially simple, however I have a hard time to imagine a controller that is both simple and able to play the full palette from NES to Gamecube games and in general I consider a few more buttons a good thing, since it makes third-party ports a lot more playable.

Tuesday, August 30, 2005

Gnome - love it or hate it or both...

I am not sure if I should love Gnome or hate it, but I do probally both. Once upon a time there was Gnome1.4, the almost perfect environment for Unix, easy to use and extremly flexible and powerfull at the same time. I could replace all my hand crafted SCWM scripting with Sawfish in a day, even better I didn't even need to make use of its Lisp scripting to do that, since it provided a powerfull configuration gui. I loved it. Then there came Gnome2.0, which removed heaploads of features that I learned to love in Gnome1.4. I hated the new Gnome2.0, it provided zero benfit and only worked half as good. Even worse, thanks to the flawed package management systems you have on all Linuxs I was forced to upgrade to it. Yes, I know I could have compiled myself an own tree, but that is a lot of work, waste of CPU time and its not really that much fun anyway to run a unmaintained DE. So I stayed with Gnome2.0, thinking back of the bright times I had with Gnome1.4 and how those developers could manage to ruin it completly...

Well, that was some years ago. Today things look a bit different. Gnome gets closer to what was promised back then, a easy to use desktop. Gnome took a rought road to ease-of-use, brute-force removing a lot of features, displeasing a lot of user. However slowly that way seems to pay of. Today Gnome actually is a easy to use desktop, maybe not the most powerfull, but certainly very well organized. Even better, most of the usefull features that one needed back in the Gnome1.4 days are back, either nicely integrated into the GUI or available via GConf, Gnome registry clone. The new filemanager slowly gets usable via the keyboard (no more Ctrl-L uglyness needed), once you get used to the different button order it feels great and last not least Metacity actually provides some options to configure it. Today I managed to switch finally from Sawfish to Metacity, while my configuration couldn't be transfered 100% over to Metacity, I was able to restore like 95% of it, I can start programs via shortcuts, raise, lower and close windows with special keys and pick a nice looking theme out of the dozens that are offered.

Still not everything is perfect in Gnome, there are still a few features that I miss, but overall Gnomes is these days a very easy to use and configure desktop environment, that can even please somebody who has started with SCWM and Sawfish. It might not be 100% up to the old power, but its gets pretty close, probally close enough for almost all users. I don't like how the Gnome developers brought Gnome from 1.4 to where it is now, but I certainly like where it is now. Its good that Gnome takes the road to provide an easy to use DE, instead of a feature blown one like KDE. In this case its good to have a choice.

BTW. What I would like to have would be a standard way to go into and out of real fullscreen mode for games, a quite important, but still largly ignored issue, maybe we will see it in the future, who knows...

Saturday, August 27, 2005

'Viewpoints' - A form of entertainmaint beside games and movies?

Computer and console hardware has advanced to a point where it is possible to render quite realistically looking scenes in realtime, not yet the quality of Final Fantasy, but its getting pretty damn close. This made me thinking, what beside games will it offer us? There are already machinema aprearing every onces in a while, movies rendered with ingame engines. However those are still classic movies for most part, fixed camera angles, nothing happening behind you, etc. I think sooner or later we will see a new type of entertainment, neither a movie nor a game, but something inbetween and no I don't mean those crappy interactive games with filmed actors.

A classic movie is always linear, it has a start and an end. The camera angles are fixed and you know always exactly what is happening at any given minute if you have already watched it. With realtime rendered movies this could be different, instead of providing a fixed linear film to watch, a realtime rendered movie could provide a complete world to explore by the viewer. The viewer itself wouldn't interact with the happenings like in a game, the whole story would still be linear like in a movie, but it would allow the freedom to move around, to look at other stuff in a scene, to observe happenings beside what is happening to the main characters. This would allow to dive into the movie much deeper and the movie would be allowed to present a far richer setting as it does today. Instead of fixed camera angles and just a few characters with a few lines, it could provide dozens of characters with tons of text, the viewer wouldn't get everything on the first viewing, but could focus on different parts of the happenings on each viewings, if there is talking in the background, he could move over to that and listen. If there is a fight happening, he could view it from the perspective of the other oponent or from the perspective of a by-stander. The viewer had the ability to view a movie from multiple viewpoints, not only predefined ones, but also ones that he chose for himself.

The effort that would be required for such a movie would be quite huge, since not only a few maincharecters would need to be fitted with dialog, but everybody appearing in a movie would get text and something todo. It would turn a simple movie into a complete little world. It would of course also steel some artistic freedom, since the director could no longer force the viewer to a specific camera angle, do things offscreen and the like, when the user can always adjust the camera completly to his liking, but it would of course also offer tons of new ways which could be artistically explored.

You might think that this sounds a bit weird, but its really not that weird, both games and movies already have gone that route, at least to a little extend. In game The Longest Journey for example there is one point where the game branches, either you go to work or you stay at home on both places something is happening, but you see only one event life, the other is told you then from the other characters. A even more drastic example is the Last Express, its an adventure game that happens in almost complete realtime. you are on a train and have to solve a murder, but while doing so all the other characters on the train follow their way on their own, no matter if you interact with them or not, the train will always move forward and the characters will always walk around. The only real different to the 'viewpoints' I have introduced here is that you are still part of the story, there are events that you can trigger and your happening will influence the surrounding. With movies there are similar things to 'viewpoints', however much lighter, for example many movies exists in multiple version, the normal one as shown in the cinema and then the version you will find on DVD which in addition to the normal one includes additional deleted scenes, which often provide different viewpoints or more detail into what happened. The directors cut version is then using all those scenes to build up a new movie, which make given even more details into the happenings of the original movie. A movie is of course limited in that it can only work with the filmed footage, thus the camera angles are always fixed, a realtime rendered one however wouldn't have that limitation.

In the end I am not sure such 'viewpoint' entertainment will ever happen on a bigger scale, but I am reasonably sure that it will happen some day on a smaller scale. The technologie is almost there, so its just a matter till story writers to come up with enough content to not only fill a single stream of events, but also the whole surroundings.

Friday, August 26, 2005

Working Around Mailman

Mailman is currently the most used mailinglist management software for free software, its used on basically all the different repositories (sourceforge, berlios, savannah, etc.), however its seriously flawed in many aspect, just have a look at Mailman Considered Harmfull for a quick overview. One of its most annoyings bugs is that it uses a password-only login for the admin page, this means that all admins have to share the same password and that browsers will have a hard time remembering the password, since the normal user/password recording mechanism isn't triggered, beside from that sf.net doesn't even seem to have a automatic way to retrieve the password. Anyway, after a bit of testing I finally found a way to work around the braindead login-page, thus making approving held posts a easy task, instead of an insanly complicated one. The workaround is quite easy, instead of using the Mailman login form, one simply builds an own login form. To do that one creates a empty .html file and inserts the following content:

<html><head><title></title>
</head>
<body onload="document.passwordform.submit()">
<FORM name="passwordform" METHOD=POST ACTION="http://lists.berlios.de/mailman/admin/foo-devel/">
<INPUT TYPE=password NAME=adminpw VALUE="xxxxxxxxxxx">
<INPUT TYPE=SUBMIT name="request_login" value="Let me in...">
</FORM>
</body>


The ACTION url then simply needs to be adopted to the one used in your mailing list and the VALUE needs to be set to the password for the mailinglist. Finally just save the file to some place on your harddisk and bookmark it. Voila, after a click on the bookmark you will be brought directly to the configuration page of mailman, no more password required. This little dirty trick might of course work for other forms too, just copy&paste them out of the original webpage, fill out the VALUE parameter and add on onload event.

Wednesday, August 24, 2005

Death of Usenet

Usenet really is dying, these days probally more then ever. Yesterday I did a search for a C# group, what did I find? A bunch of official Microsoft and Borland group, but neither a comp.lang.csharp or a de.comp.lang.csharp. I might have understood that a few years ago, but today, 2005, where there are dozens of books on the language, Mono is starting to get really usable and Microsoft pushing really hard for the language its just ridiculous. A little further search showed that there was a vote for the group, but it failed the 100 votes limited. Seems like usenet is trapped in a vicious cycle, no users, no groups, no groups, even less users -> usenet dead soon. Not sure what exactly causes this to start, but there are numerous things that just went fundamentally wrong:
  • very complicated and time consuming to create a new group, this might have made sense years ago when there weren't web forums and the usenet was full of people, but these days people simply create a webforum, which is a matter, instead of a newsgroup
  • mailing lists, again, same as with webforums, if there is something that matters people create a mailing lists instead of a newsgroup
  • lack of servers, more and more providers fail to provide a usenet server and there simply aren't any free ones which provide NNTP, there is Google Groups, but thats basically it. I can't blame the providers, since running a NNTP server is quite a huge amount of work.
  • bad usenet software, for years there wasn't a decent newsreader on windows, not even sure if there is now one, but if webforums provide automatic email notification on reply and usenet software provides exactly nothing along the same lines, its clear why people move away from the usenet
  • nettiquette, while web forums can get rough at times, most stay pretty well readable, on the usenet however people often end up having endless discussions about faked From headers, signutures longer then 4 lines, posting to the wrong group, redirecting them, follow-up or stuff like that. On mailing lists I have not even once seen such a useless discussion pop up.
In the end Usenet is doing great at shovling its own grave and I don't see any trend to move away from that. What usenet really could need is a more distrubuted aproach, open servers and software that makes it easy to interact and cross server boundaries instead of making it increadible hard or unimpossible. The whole policy on creating a new group should be reworked and made much easier, a protocol to handle group renames might help a lot to make sorting groups later easier instead of extremly hard as it is now. A little bit of common sense would already help usenet a lot, but its stuck in its own regularia, it wasn't Microsoft, Google or Netscape who killed the usenet, the usenet did pretty well at starving itself to death by sticking to old retiuals that completly dissmissed the progress of the rest of the web. I miss the usenet of the old days, but its just to damn obvious why nobody likes it anymore these days.