This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
Messages - Trovaner
Pages: 1 ... 37 38 39 40 41 42 43 [44] 45 46 47 48 49 50 51 ... 64
861
« on: August 19, 2011, 03:04:18 PM »
Philetbabe also made a tutorial for AIing which can be downloaded from his sig, his thread, or here.
862
« on: August 18, 2011, 01:17:49 PM »
ATARI made a trailer for stock RA2 but I don't believe the DSL team has created any.
863
« on: August 16, 2011, 11:12:24 PM »
Mad was making a mod which was to be a complete overhaul of DSL (with a focus on realistic land and air forces).
Having a kamikaze flight squadron as just an AI pack or team however would be pretty sweet. This would be especially true if using purely DSL components. Using piston propulsion may even be AI-able to inhuman precision (though that doesn't appear to being saying much).
864
« on: August 15, 2011, 04:43:50 PM »
I added him to the poll and changed it so that people can change their votes.
Just to be fair, are there any additional nominations besides Starcore that people would like added? By my understanding, it must be a member that is still around and posting. I refuse to remove anyone from the poll though so don't bother asking.
865
« on: August 14, 2011, 09:50:46 PM »
The 7 people who voted for me are the coolest people on this forum :D
Although the poll is missing some of my all time favorites (namely Luigica and Rejected), I still managed to find a few.
Which is why it's "at the moment" not "of all time."
Thats why I worded it that way. Though if I was to be truthful and stretch the question a bit, I would say that Rejected is still one of my favorite stock builders ATM (even if he hasn't posted any new bots). BTW, I think Starcore posted a bot more recently than Kill.
866
« on: August 14, 2011, 02:38:09 AM »
Although the poll is missing some of my all time favorites (namely Luigica and Regected), I still managed to find a few.
867
« on: August 12, 2011, 04:56:48 PM »
I understand that you are giving the most general case as possible but here are just a few unmentioned exceptions/notes/solutions. AI is pretty basic. the main tactic is to shove whatever is nearest and keep on shoving until a timer pops up. This works, because all stock AI never had tactics like driving backwards when something gets under them, or attacking from the sides.
There actually hasn't been very many additions to the AI tactics over the years. The biggest and most universal addition was Click's Ram and modified Charge tactics. The engage tactic is by far, the most used tactic in today's AI.py files. It basically keeps moving towards the opponent until the opponent is with its radius (which is unfortunately also used for hazard and wall avoidance). The shove tactic tries to push the opponent into neighboring hazards. Underused as a result of hazards being off and the requirement of a map file. The charge tactic backs up for ramming purposes when it isn't moving very fast. Click modded this to be smarter (ex. not backing into opponent). The original stock AI actually came with a flanking tactic named reorient but it was only used by chopper.py (a hammer AI) and it wasn't very advanced. Additionally, the judge "AI" (Or mechanics for judging robots in fights) is simple, and chooses a simple factor to seperate winners from losers: Damage. Flipping an opponent out isn't as highly regarded as outright killing them. There have been steps to counter this, as one 2011 tournament has recently used non damage based judging.
We could add something to give exponentially more points the higher the bot is thrown but there would be some serious balance issues.
868
« on: August 12, 2011, 04:16:49 PM »
Nothing is stopping someone from making controllers that will automatically press forward at the right times. There are plenty of programs out there that can use timers for key presses or look up data inside of a text file (for turn speed or even the heading). Its just that the AI already has everything coded.
When we use AI vs AI, it doesn't matter how precise things are because both bots have the same capabilities (you could always make a smarter AI).
When we do human vs AI, its only fair that the AI has some advantages over the human player (since we generally have an advantage when it comes to flanking, targeting specific components, or waiting for the right times to attack).
The issue here is not that Melty Brain is unrealistic (as the topic name suggests) but that it is unfair/unbalanced (as Sage has repeatedly pointed out). Then again, the same could be said about a majority of RA2's more optimized designs (flails, popups, crawlers, etc.). Melty Brain just happens to strengthen an already powerful AI type. Though if this was the issue, you could just as easily make a smarter AI for some other AI type.
869
« on: August 06, 2011, 10:28:56 PM »
I wanted to make something like Vetearn Destroyer, but I ran out of time and had to send Overloaded Orbinaut 4.
I'm screwed. XD
Same here. After making a custom AI for one of my failed prototypes, I wasn't left with much time (to make a new bot or AI). I ended up making something quick that could use the otherwise wasted AI.py (even if it meant doing a total of 3 battles for testing/optimizing). I may add something to the AI to make it more competitive. That is to say that I'll have the time.
870
« on: August 03, 2011, 03:35:32 PM »
Map files are used to mark walls, hazards, ramps, and empty space within arenas. They don't work well for overlapping floors but they were designed to mark a multi-levelled surface. The only time that I'm aware of this data being used is when "Arenas.currentArena.IsStraightPathClear(XYZ-StartingLocation, XYZ-EndingLocation, IgnoreHazards)" is called. Which happens to be in most of the AI tactics ('Charge', 'Shove', 'Reorient', 'PushOffEdge', and 'Dethrone').
Arenas/__init__.py has methods for loading, saving, and creating map files but loading is the only one that works. It is possible to get the other two to work but it requires a bit of tweaking. Another option would be to use a text editor to make the map files by hand (it really isn't that hard to figure them out).
871
« on: August 02, 2011, 07:24:40 PM »
FRAPS won't let your FPS run faster than 60 which may be an issue. It works well for games just because it doesn't do any compression while recording. If your problem is slow playback, did you remember to compress it before watching the video (it can cause lower end computers to lag). Otherwise, you may consider manually speeding it up using whatever software you are using for compression.
What program should I be using to compress? The problem isn't just slow playback, it's that the game itself slows down to about half speed whenever Fraps is recording.
It all comes down to personal preference and what you plan to do besides compress your video. Some people use Windows Movie Maker but IMO, if you plan to do anything besides compression, your better off using something else. I usually do my video editing in Linux so I doubt your interested in what I usually use. Here's a list of Windows Video Editors that seem to get the job done well: - Adobe Premiere (proprietary) - Excellent professional software with very little competition (Elements: $100; Pro: $800)
- Sony Vegas (proprietary) - Another professional level video editor ($600)
- Debugmode Wax (freeware) - Works well for most users and is compatible with many other program filters/plugins (free)
- Avidemux (GPL) - I'm personally not too fond of this one but many people swear by it (free)
- VirtualDub/VirtualDubMod (GPL) - Both are pretty feature rich but the learning curve might be an issue (both are free)
- Lightworks (proprietary but soon to have an open source license) - I haven't personally used this but it looks very promising (free)
FRAPS will always limit your game to 60FPS regardless of your settings. Due to this, you may experience unnecessary lag when recording. However, you will always experience a slowdown in performance when recording (with or without FRAPS as your recorder). Ways of increasing in game performance while recording: - A pretty good rule of thumb for FPS is that if you drop below 20FPS, your eyes will see a noticeable delay (that is why we usually set the FPS to 30). For FRAPS, we usually recommend recording in 'half size' and at 25-30 FPS
- Setting the game's screen size to smaller dimensions will reduce the amount of data being processed (resolution will take a hit though so be weary of this)
- Don't run your screen recorder while in fullscreen. Fullscreen is terribly inefficient even when you aren't recording
- If you aren't going to be using in game sounds, don't bother recording them (on the recorder end)
- Turn off Music, Sound Effects, and/or Announcer if they are not getting recorded (on the game end)
- Consider turning graphics down to Medium or even Low
- Turn off any unnecessary background processes before recording. Remember to re-enable your virus protection and firewall if you turned them off before recording
- Remove unnecessary programs and files
- Scan and remove viruses, malware, spyware, and adware
- Defrag your computer
- Update sound and video drivers
- Look into increasing your virtual memory (Control Panel>Performance and Maintenance>Adjust Visual Effects>Advanced>[Virtual Memory]Change>System managed size [for all drives])
- If you really want to go to the extremes, turn off Window's Visual Effects (Control Panel>Performance and Maintenance>Adjust Visual Effects>Adjust for best performance)
- Another extreme would be to turn off your desktop icons (Right-Click>Arrange Icons By>Show Desktop Icons)
872
« on: August 02, 2011, 02:33:15 PM »
FRAPS won't let your FPS run faster than 60 which may be an issue. It works well for games just because it doesn't do any compression while recording. If your problem is slow playback, did you remember to compress it before watching the video (it can cause lower end computers to lag). Otherwise, you may consider manually speeding it up using whatever software you are using for compression.
PlayClaw doesn't limit FPS to 60 so IMO it works better than FRAPS. It also doesn't do any compression so performance is similar (if not better) to FRAPS.I've used the trial before but if I ever choose to record/upload something, I will likely buy the full version of it.
873
« on: July 31, 2011, 10:32:37 AM »
I made a few possibles but won't send anything concrete until later.
874
« on: July 30, 2011, 10:30:51 AM »
Or you could program the bot lab to change the icon for you automatically. I actually implemented this in my unfinished practice arena a couple months ago.
Another option would be to change the background color depending on the weight class. This way, you can have as many weight classes as you want.
875
« on: July 26, 2011, 02:37:15 PM »
You have already brought this up in the past and, to my knowledge, nothing has changed. Currently, we are still struggling to find where the EXE has it hard coded... Even if we were to find it, its unlikely that it would be as simple as changing a 4 to an 8.
876
« on: July 22, 2011, 11:51:58 PM »
For Gimp, you have to explicitly specify that you want 24 bit tga/bmp by using 'Save As' instead of using 'Save' the first time. Otherwise, it defaults to 32 bit tga/bmp which aren't supported by RA2.
877
« on: July 20, 2011, 02:21:25 PM »
This is kinda "illegal" so I'll make website for this.
Its more than kinda, it is illegal. As soon as you mentioned that you were accepting entries from people, this no longer followed our unofficial tournament rules (which have now been reposted/updated to here).
878
« on: July 10, 2011, 05:54:41 PM »
I finally got around to retesting everything.... My assumption that things were given defaults based on their bases (as opposed to their type/category) was my downfall last time. Thanks Click for doing some legitimate testing on this.
Type/Category: Hitpoints (Fracture)
Chassis: Varies (40 x3)
Power: 400 (∞)
Extenders: 100 (15)
Mechanics: 400 (20)
Wheels: 400 (20)
Weapons: 1000 (20)
Extras: 1000 (60)
Stock's only exceptions: (set within their respective text files)
Disc: 500 (20)
Rotator Bar: 500 (20)
Razor Tip: 500 (20)
Emergency Flipper: 600 (20)
Ramplate: 1000 (40)
879
« on: July 09, 2011, 03:43:41 PM »
@Trovaner: Not releasing also brings the chance of someone finding out about it & Not talking about it... Just using it without telling anyone. (Ofcourse, He wont be able to overdo it or peeps will notice)
If I found out about it myself, I would use it in a few Tourneys, Win 2/3 tourneys & Then be like: LOOK WHAT I JUST FOUND 1 DAY AGO! And release it :D
Since they would be limited to the stacking ability of normal stacking (or get DQ for .bot file editing), it wouldn't be any different than it is now. For example, I could easily .bot file edit a bot that is virtually invisible among the other fair entrants. In fact, we've had a few people already admit to doing this.
880
« on: July 09, 2011, 01:58:25 PM »
I don't see any problem with it being released or not.
Though he has many options, I would consider these to be the most probable...
By releasing it to the community:
-He gets to name it. We already have 3 different glitches/hax attributed to Zero Absolute's thread like this and nobody truly knows which one he used.
-We make some rules for stacking (which probably wouldn't be such a bad thing considering AAM and .bot file editing are so hard to trace).
-We would need to make AI packs that are built to these new standards (in the same way Starcore V1 introduced glitches into common practice). It would not ruin stock but would definitely be something to consider when building bots.
-People would no longer need to complain about stacking (since everyone would have the same limitations)
By not releasing it to anyone else: (like Zero Absolute did)
-Life goes on. Another person may find it and start this conversation over again. Then he may chose to release it or not (with the same benefits).
-Nobody would be allowed to use the glitch in tournaments since it would fall under the .bot file editing (unless its an anything goes tourney).
Pages: 1 ... 37 38 39 40 41 42 43 [44] 45 46 47 48 49 50 51 ... 64
|
Show Posts