Username:
B
I
U
S
"
url
img
#
code
sup
sub
font
size
color
smiley
embarassed
thumbsup
happy
Huh?
Angry
Roll Eyes
Undecided
Lips Sealed
Kiss
Cry
Grin
Wink
Tongue
Shocked
Cheesy
Smiley
Sad
<- 1234 ->
--
--
List results:
Search options:
Use \ before commas in usernames
Ewil: it's true it's a shame SDA runs only live on outside the game pages, but then again... "losing your run"? Everything gets old doesn't it? It may be good as a point of reference sometimes to see older runs but seldom, and it'll "live on" in your YT. What's the problem actually? You want a reward? Cheesy
The "problem" is that SDA only keeps track of the best time. There are no leaderboards. It's not about getting rewarded, it's about respecting old records that will be eventually beaten.
Okay. I agree that old runs have value and submitting on the leaderboards is smart because of that.

I'll be eagerly following this thread Diego if you decide to go ahead with it!
Edit history:
diegowar: 2016-01-28 07:06:52 am
I love YaBB 1G - SP1!
Thanks for all the input guys Smiley

So actually as far as I know spawns are random but aren't randomly placed. In Quake Live that's the same, except they added a system that tries to limit player spawning to each other (in duels at least) so it's still random but random in between all the spawns minus the ones that are too close to the enemy player.

From the training and runs I did, I actually remember some of the spawn locations in the duel maps, but there's only so much you can do with that information, and as you said on bigger maps with more bots it's even harder to try to spawnkill bots because you're always trying to get to the action, so taking the time to aim certain spots in the thin hope that a bot might spawn there slow downs your movement, and I'm wondering if it's even a good operation in the long run, compared to rushing directly to where you know for sure there are bots. That's something to be looked into further.

Finally, no I don't have the bot chat off, because exactly as you said, it can be helpful to have the bot stop to write (and it actually is in the run, at the very end of the DM5). We can't see it on the screen because I disabled the console on the screen (cl_noprint 1).
Edit history:
LotBlind: 2016-01-28 12:06:19 pm
You might wanna do a summary of the graphics etc. settings you're using (options screen - do whatever you like; console - technically has to be approved if it has gameplay effects), post it here, and see if other players and runners want to comment on it at all. Even if they don't you'll have covered your bases for any verifications.

coolkid I can see you
Mods on speedrun.com allow any setting, even things like r_fullbright. The only thing you are not allowed to use are mods like OSP/CPMA and cheats/wallhack/aimbot, of course Smiley
Yeah probably common sense is sufficient here with the knowledge that the game has a history of being consoled a ton just to get its performance up.
Earlier, hztn posted some links to some demos for Nightmare IL records, which I didn't download. I tried searching for what the IL times were, but couldn't find them. What was the best time for the Xaero level?

Here is a video of my best attempt, 30.07s (30.085s in the video).



I used sv_fps 200. Jump pads seem to push you too far with sv_fps 125.
Wait, sv_fps 200 makes frames cap at 200? What do you mean jump pads push you too far? I dunno if that's a change SDA can allow, because I know (and you're saying) it affects physics. If it was originally capped at 125, and that's a console modification... :/ Can you elaborate?

Anyway as you may know in order to submit on SDA you'd have to do the whole table (or get some demos from somewhere), but the ones you submit don't actually need to be WR times. It's kinda frustrating because with competitive games like this people will never likely submit anything ever because there's always someone better somewhere out there. It's also why we don't have good, verified, unified (as per rules followed) times anywhere. Speedrun.com is a little bit lax when it comes to that stuff...

Sorry I never saw those IL demos either. If I saw that quality of runs for every fight though (it looks like it almost doesn't matter if it's I can win or Nightmare lol) - and the A/V looks good too but feel free to post that under "tech support" to make sure - It'd be one of the easiest accepts ever Tongue
sv_fps n makes the game (server) calculate physics and other game events at 200 FPS. I think the normal value is 40 (though the console says the default is 20), but players choose to increase this to reduce latency / jerkiness in movement. I mentioned the value of 125 because it seemed to be a popular choice, but I used 200 because it provided more consistent physics behavior to the default of 40, as I will discuss below.

As I understand it, the game operates with an internal timebase of 1 millisecond ticks, so when you use the command "sv_fps 200", the game understands this to mean 1000 ticks / 200 FPS = 5 ticks per frame. As it turns out, 200 exactly divides 1000, but because it is restricted to being an integer and rounds down, all values from sv_fps 200 to sv_fps 167 (1000/167 = 5.988 ... -> 5) do the same thing.

As for why these affect jump height, this has to do with rounding errors in the calculation on how gravity affects your velocity each game frame. The game's gravity is 800 units/second^2. Under the normal sv_fps value of 40, each game frame would last 1000 / 40 = 25 ticks (which can also be thought of as 25 ms). So, every game frame, a player falling would lose 800 * (25/1000) = 20 units/s of speed in the up direction. No rounding here.
But with sv_fps 125 (or 124 down to 112), each game frame will be 1000/125 = 8 ms of game time, and so the change in vertical speed per frame would be 800 * (8/1000) = 6.4 units/s -> 6 units/s, due to rounding (to nearest, apparently). So, with sv_fps 125, gravity has less effect; in a second, instead of losing 800 units/s of speed, you only lose 6*(1000/8) = 750.

If you set sv_fps 125 and take the BFG jump pad sequence in this level, you actually don't miss the second jump pad (you still clip the upper edge of it), but instead of landing on the BFG platform, the second jump pad makes you collide with the squishing block above the platform, and then fall off. With sv_fps 20, 40, or 200 (or the others with no rounding), you hit the middle of the second jump pad, and pick up the BFG as you land on the platform, as was presumably intended.

If you set sv_fps to 333 (1000/333 -> 3 ms per frame, 800 * 3/1000 = 2.4 -> 2 units/s of speed change per frame, effective gravity 666.6...), then you have to use a lot of air control to get to the platform without falling off.
I know defraggers have 8 ms frames, so that's at least some kind of standard. I thought it was a the game-internal standard too but I guess the mod itself or the players themselves have decided they'd all go 125 fps... but I really thought this was because the game doesn't track anything more exact than 8ms, so are you sure it's 1000?

It's always very very difficult to argue for "as was intended" because at that point you might as well play Quake 1 without rocket jumps or bunnying. That's why if the game has buggy behaviour, you kinda just have to cope with it. Still it might be acceptable as a community-based thing (as with whatever graphics changes you want to make), but I'd suggest posting a short summary of what and why you want to do with the game under "SDA discussion" and your arguments for doing so. While you're there, confirm that SDA accepts demos for games other than Quake 1. You musta seen the other guy's recent nightmare SS run, with the comments where he mentions which changes he's made. He specifically didn't want to use forcemodel or nogibs, and although it would be in the spirit of speedrunning to use every little thing to your advantage, instead of being half-and-half, such community decisions also exist. With ILs though you could use those options and render it normally from the demo.

I think there have been times when runners have changed FPS even from segment to segment, so I think it's fine but you should really make sure. It's the fact that it kinda completely changes how the game works but I do see the logic with 200 myself. Would you say it generally gives you an advantage to go 200 for most or all fights?

Hope you can pull through with it!
To test the effects of sv_fps on jump height, what I did was stand on the mega health jump pad in q3dm17 (which pushes you straight up), and aim at some fixed point in the map at the peak of my jump, then change sv_fps in the console without moving the mouse and see if I end up aiming above or below the original point in the following jumps. Doing this, you can quite easily see the sudden changes in behavior between sv_fps 125 and 126, 333 and 334, etc.

If the game had a lower limit of 8 ms, then surely all sv_fps values above 125 would do the same thing. But there is a clear difference between values above 125 (the lower gravity of sv_fps 333 is obvious).
Also, sv_fps 1000 works, but sv_fps 1001 will freeze your game.

Another thing suggesting an internal timebase of 1 ms is related to how I created the video, using the cl_avidemo. cl_avidemo n is supposed to generate a video at n FPS, by putting screenshots into the screenshot directory. But if I try cl_avidemo 60, then the resulting video loses sync with audio (recorded separately) over time. The degree to which it loses sync suggests that cl_avidemo 60 actually computes 1000/60 = 16.666... -> 16, and simply advances the game time by 16 ms between screen shots, effectively 62.5 FPS. Using cl_drawSnapshot 1 to show the game time on screen (and in screenshots) can help see this. So, to make the video, I used cl_avidemo 1000, and resampled the resulting 'video' to 60 FPS myself (and again, cl_avidemo 1001 freezes the game).
The second gotcha is that the game stops making screenshots at shot9999.tga, but a 30-second video at 1000 FPS is 30000 screenshots. So, for the purpose of exporting the video, I made a copy of quake3.exe and searched it for 0f 27 00 00 (9999 in hex), and changed 7 occurrences of this to some larger number. There is also a conspicuous string "shot9999.tga" near the end of the binary which I changed to something else. The resulting binary successfully exported past shot9999.tga, but the directory was strangely missing some seemingly random screenshots above shot10000 (10080,10260,10440,10620,20069,...). Because I don't fully understand the changes I made, I only use the modified binary for demo export.

There must be a less sad way of exporting videos from demos.

As to the bast value of sv_fps to use for speedrunning, I don't know. For this level, decreased gravity simply means that it takes longer to complete the jump pad sequence, so in that respect the maximum gravity which still lets you reach the BFG platform would be preferable.
sv_fps has other small physics changes. If you attempt to walk in a direction very slightly unaligned with the level's x or y axis, then you find that the game moves you exactly in the x or y direction instead. This is more noticable at higher sv_fps values. Presumably this is because the game resolves your running acceleration into x and y components which similarly get rounded to an integer number of units/s to be added per frame. So maybe how fast you reach top speed will similarly be affected by sv_fps... and the direction you happen to be aiming at the time.
For me, the biggest effect of sv_fps is the effect on input lag. With sv_fps 40, the railgun is much harder to use. It seems like the game delays my click by 2 or 3 frames, and not just 1.

I wasn't actually thinking of doing a Quake 3 run myself; I just happened to see the other guy's run, but noticed he didn't have any fast strategy for the Xaero level, or even use peeking (at 44:17 in the video, he rails Xaero once around the right side of the pillar next to the player, but Xaero shot back, so the player really should swap over to peeking around the left side of the pillar for the finishing shot (bots seem to sometimes keep shooting at where they saw you last, even if they no longer have what they consider to be line of sight to you)). So I just thought I would put this out there, in case it does help a runner going for a full run or IL.
Seems like there's no reason one person has to do all of them - unite forces! If there's enough of you crack players, everyone can just do 1 or 2 tiers. Or ask around. That's the way a lot of segmented/ILs runs are made.

I see your reasoning for why it's 1000 is pretty solid. In that case defraggers just had to agree on something to make competition fair, but maybe 200 fps was too high for some of them or something. Or they just didn't realize that. If they changed it now, all the 10 thousand defrag maps would be affected, but there's probably no reason you couldn't make 200 the SDA standard for normal speedruns. Like I said people are already messing with fps in other games though mostly to enable certain tricks in the first place.

You can render demos using wolfcam [and wolfwhisperer], or an online service called internet game movie database which has a really neat online rendered. You have to donate 10$ to make the most of it though but it's so worth it.
Quote from LotBlind:
Anyway as you may know in order to submit on SDA you'd have to do the whole table (or get some demos from somewhere), but the ones you submit don't actually need to be WR times. It's kinda frustrating because with competitive games like this people will never likely submit anything ever because there's always someone better somewhere out there. It's also why we don't have good, verified, unified (as per rules followed) times anywhere. Speedrun.com is a little bit lax when it comes to that stuff...


Can you please be more specific what you think that Q3 page on speedrun.com lacks?


Interesting stuff about sv_fps. I didn't know about it despite playing the game for 16 years ^_^ I didn't do much defragging, though. Personally, I don't mind using any sv_fps value since it's just a thing you can easily change inside the game via simple command, much like com_maxfps. Do I understand it correctly that sv_fps 200 is the "best" value?


Quote from LotBlind:
Seems like there's no reason one person has to do all of them - unite forces! If there's enough of you crack players, everyone can just do 1 or 2 tiers. Or ask around. That's the way a lot of segmented/ILs runs are made.


There is actually a segmented/IL run of Q3. It's from 2005 iirc, but it's still really good.

Ewil: I don't know what the specific page for Q3 on speedrun.com is like, but I know there have been games where e.g. timing isn't really fully agreed upon and the videos are whatever, sometimes even missing completely. No commentary often etc. Just the usual stuff. I know they're verified by someone. I just checked the Q3 page and all it's lacking is the IL runs themselves, I suppose. And some people are giving hundreths, some are not. Also if I was to run it, I'd want to know what the exact reason is why you should go skirmish over tournament mode. Is it just the timer?

That Q3 - Done Quick - if that's the only video, I think I'm not the only one who doesn't want random music and (extremely) random 3rd p views. It's not like QDQ where the recamming made some sense. So that stuff is meh, but otherwise I guess the times would be good, simply because no-one's gonna find any tricks or skips for a game like this. Still, it could be a long-term project to get even better (or just similar) times and making a normal video out of them with any and all commentary etc. cause the runs are gonna be more interesting if you can at least hear if there were any competing strategies for any given fights, or how lucky you had to be (percentually) to get those exact spawns.

Anyway, I better focus on other stuff now.
Well, timing and rules on speedrun.com are decided either by community or mods, same for verified runs, but you are somewhat correct - there are runs with missing video proofs (or missing part of the video due to twitch disconnects etc.), wrong timing, muted sound, etc. This is all up to mods for that game to manage all this stuff. If they're lazy and don't know much about speedrunning in general, the chance is bad stuff will happen. Commentaries are pretty much obsolete thing in speedrunning. There's really no need for it if there is a guide, which explains all the latest tricks and shortcuts. If not, then it's a problem, of course.

As for the Q3 page, which I'm a mod of, I don't see anything wrong with it or if there is, nobody reported it. IL table is empty, because... no runs were submitted. Skirmish mode is preferred because bots spawn back to back in Single-player mode instead of "all-in" in Skirmish, just like you can see in Q3DQ. This is especially an issue for the very first map q3dm0 - there's about 10 seconds delay before Crash spawns in Single-player mode, unless you use map_restart 0 trick.

Q3DQ - There used to be a website with all the demos used in the video in case you didn't like the music and cams, but it's been dead for many years. You can still see what's going on the screen and copy the strats though, so I guess it's just about personal preference. Personally, I don't mind not having crystal clear HQ video with only in-game sound, as long as half of the screen is not filled with crap.

If anyone here wants to improve that Q3DQ as a team, I'm up for it, but I'm pretty sure it's gonna be hard and a lot of grinding.
"commentaries are obsolete" - That's actually kinda absurd. On SDA, you can expect your runs to get wider exposure than on speedrun.com, and many people won't have seen any recent runs at all. The guide is meant to be really detailed and cover everything in detail isn't it? Well, that's not ideal for most people, instead a commentary specifically aimed at an uninitiated audience is much better. If you want to link to a guide in your commentary, go for it, but why should the viewers all need to dig up the guide instead of the runner writing a few words him or herself? The guide also won't cover stuff like the specific events, RNG, and ad lib involved in the run at hand, so I really don't get how you can call commentaries useless :/ Audio commentaries in specific are many people's favorites, and again, have to be aimed at a different audience than your random Twitch livestream comments.

Yes, I get that's why there's no ILs Wink

The Q3DQ video isn't purposefully recammed, a good chunk of the time you're seeing the wall behind the player. That's not "personal preference". I appreciated the effort put into most other QDQ videos.

Quote from Ewil:
If anyone here wants to improve that Q3DQ as a team, I'm up for it, but I'm pretty sure it's gonna be hard and a lot of grinding.


I can only imagine. How about doing the team arena fights instead? Is there really just 4 "tournament" matches or are there more once you've gotten under par on each of them?
Quote from LotBlind:
"commentaries are obsolete" - That's actually kinda absurd. On SDA, you can expect your runs to get wider exposure than on speedrun.com, and many people won't have seen any recent runs at all. The guide is meant to be really detailed and cover everything in detail isn't it? Well, that's not ideal for most people, instead a commentary specifically aimed at an uninitiated audience is much better. If you want to link to a guide in your commentary, go for it, but why should the viewers all need to dig up the guide instead of the runner writing a few words him or herself? The guide also won't cover stuff like the specific events, RNG, and ad lib involved in the run at hand, so I really don't get how you can call commentaries useless :/ Audio commentaries in specific are many people's favorites, and again, have to be aimed at a different audience than your random Twitch livestream comments.


OK, maybe this statement was a bit over the top, but like I said, if the guide explains all the latest tricks and shortcuts (and rng, specific events etc.), then commentary for each run is really not necessary. And please, I didn't say commentaries are useless. If someone wants to write them, I have no problem with it, but as you can see almost nobody does that on speedrun.com. I completely understand why people write commentaries here on SDA.

Quote from Ewil:
If anyone here wants to improve that Q3DQ as a team, I'm up for it, but I'm pretty sure it's gonna be hard and a lot of grinding.

I can only imagine. How about doing the team arena fights instead? Is there really just 4 "tournament" matches or are there more once you've gotten under par on each of them?


Almost noone liked Team Arena, so finding a team to do IL speedrun would be really difficult, especially now since Q3/QL is almost dead Sad
I love YaBB 1G - SP1!
Here is my latest run of the game.

(I've seen discussion about sv_fps earlier in the thread, and man do I wish I had that many FPS while I did this run :). I recorded as I played with FRAPS and was locked at 60fps or below, I couldn't even get to MegaHealth on DM13 unless from above ^^)

This is probably the last run I will ever do of the game, it took me way too much time to do and I'm not completely satisfied by it :p

Anyway I'm still glad to have finished this run and be done with it. Months of fun and frustration.

With this game you can just record demos and render them separately if recording at 200FPS won't work, but unfortunately only for ILs. SS requires the menuing to be there. Well, unless the game has an in-game timer showing your finish time, but AIR it doesn't. If it's a problem for future runners, there may be room for an exception though, seeing as 200 FPS is by no means a standard.

9:28 - how did you know that quickly which spawn you'd gotten? You're moving your mouse INSTANTLY? Or wait it's the same on Campgrounds... does it always give you the same spawn in tournament mode or something?

19:00 - this first rail on Klesk... does that happen so often that you know exactly where they'll be when you drop? :O Where are all these fucking pixel-perfect flicks coming from anyway?!

23:00 - O_o

THAT LAST FRAG!!!!!!!! SwiftRage all day.

OMG I laughed so so hard on the whole intro! Jesus, that's the funniest speedrun "schenanigans" since one of the QDQ vids. Man! The Cooller quote...

Look, this is no longer a "maybe". This run HAS TO make it into the SDA collections as an official entry. Do it so I can write about it in some update in the future. I doubt you were really losing a lot of time to not forcing models, and you definitely adjusted your 1v1 strats (and I see better ones in some other maps too) to include those spawn frags. You were right in that there weren't so many of them, but it makes the run look more speedrunny nevertheless. The plasma gun in this run is MESSED UP, but I struggle to understand how you've saved 8 minutes on a run that didn't look exactly bad... Would you say you got your warm-up in? Tongue

Man o man o man.
Q3 doesn't have any final/total IGT unfortunately Sad

9:28 - This map has only a few respawn points and since there are 2 enemies, the chances you get this particular spawn is high.
The Dork Knight himself.
If you need higher than 60fps to play, try using OBS for recording. You can always use the nVidia encoder if the software x264 built-in to OBS is too system intensive.
Edit history:
diegowar: 2016-07-01 02:09:52 pm
diegowar: 2016-07-01 02:09:09 pm
diegowar: 2016-07-01 02:03:58 pm
diegowar: 2016-07-01 02:03:36 pm
I love YaBB 1G - SP1!
Yes, for 9:28 and like other maps, our first spawn is 100% always this one. On Campgrounds there is a few possibilities, this is why you see me think for a split second before going to the jump-pad on the left.

For Klesk there is a few places I can spawn, but for the majority of them I have the time to get to the rail gun and hear where Klesk has spawned. Depending on where I hear Klesk's first spawn I have a setup, in this case yep I know approximately that he was going to be there, but I have to be extremely quick because if I loose just a little time he will get the red armor.

For 23:00 Smiley , if I am able to kill all the bots quickly after getting the quad damage, then it is a good bet to camp the teleport exit and spam it. If they don't come I can go quickly chase for them from the end of the corridor thanks to the Haste power-up. In this case I had a pretty good run of teleport-exit frags, but in other runs I have had even better in the past where I could even finish the map right there :p

Hehe yeah that last frag, one more bullet (so that's like a fraction of a second) from Xaero and I was dead :p

Glad someone thought my silly intro was alright :p

For the 8 minutes difference from my previous run, there's a few things :
- First, I worked on it longer (since February)
- My aim is not what it used to be back when I was playing everyday 15 years ago, but I think it was a little more consistent (except my railgun with which I got extremely frustrated)
- I actually studied bots spawn this time (which is mostly useful when there is only up to 2 or 3 enemies, you rarely have time to spare when there is more)
- On some maps I used a bind which restarted the map (you hear "Fight !") which allowed me to either get a more favorable spawn (on DM12) or to get the bot (usually in 1v1) to spawn immediately.
- I used better strats overall (especially on the first half of the run) and the mistakes that I made cost me less times than the mistakes I made on the other run, I would say.

Now for entering this as a SDA official entry, I mean.... are you sure ? It doesn't look like an SDA speedrun, with the intro and everything Smiley


For OBS that's what I used for my 45min run but I had to record in a window and my computer didn't render it very well. But thanks for the tip.

The main point of map_restart command is to get all bots spawn instantly, which you didn't do in many maps. There's another command you could have used, which is bot_nochat 1. I also don't understand all the typing after you got killed.

IIRC you can record with Fraps without locking your framerate, but yeah OBS would be better since Fraps is known to cause input delays and other unexpected behaviour (tricks and shortcuts not working properly etc.)
Edit history:
diegowar: 2016-07-02 04:15:19 am
diegowar: 2016-07-02 04:15:13 am
I love YaBB 1G - SP1!
Yeah in many maps I don't use the restart for one or two reasons :

- The first spawn is known and I like it and don't want risking spawning somewhere else
- Having all the bots spawn at the same time might make it very difficult for me to survive the beginning of the map if a bot spawns in the same room as me, in a machinegun vs machinegun fight bots will most of the time win or leave you crippled. For example on DM6 if I use it, I might spawn at bridge stairs, and a bot will probably spawn at bridge, standing between me and the rocket launcher. I really don't want to deal with these kinds of situations.


For the few times I type after I got killed, it's a way to vent frustration, remember that I have done like a million runs, and these runs are very hard on me, both mentally as I am 100% focused at every second during the whole run, but also physically as my left fingers and right wrist have started to hurt after a few weeks of trying. It doesn't make me lose any time and keeps me sane, and keeps my hardware intact (although my mouse has withstood a certain number of kilograms of pressure per centimeter :)). When I say that I don't intend to do another run of this game it's also because of this. I have spent so much energy on this. When I type "mmm" after dying on DM17 I'm on the verge of exploding.
And I've done some very hard things in gaming, like getting A+ on all Super Meat Boy Dark Levels (and some of these are just torture), which didn't even affect me that much. Doing a million runs on Quake 3 Nightmare, seeing my mistakes and my bad luck ruins tons of 30+ minutes runs where I was invested to the core, that's why at some point I started to insult bots in the chat Smiley (which thankfully we don't see much of that here ^^). I've run other games and they feel like relaxing at the beach in comparison. The intensity is just on another level.

Now Fraps recording without locking framerate might have helped, but I'm unsure if it would have worked because I don't think I even maintained the 60fps with fraps recording at the same time Sad

bot_nochat I absolutely don't want to use, because as you can see it can be useful sometimes as a bot will just stand there and type (I think it happens on DM4 and maybe somewhere else in the run).