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
1 page
--
--
List results:
Search options:
Use \ before commas in usernames
1-Up!
Game Page: http://speeddemosarchive.com/JediKnight2.html

Star Wars Jedi Knight II: Jedi Outcast (Any %) (Segmented)

Verifier Responses

Quote:
Awesome run for an awesome game. Accept


Quote:
Not much to say about this run that hasn't been said in the forums when this first came around. Incredible improvement over LLCoolDave's run. It's sad to see that old run come off the site, but this type of improvement is worthy enough to do it.

Improvements:
Increased difficulty (Jedi Master vs Padawan)
Better overall strafejumping
Incorporates every known trick from the last 6 years
Adds in a ton of extra tricks, most of which are definite "WTF???" moments

To get this degree of improvement, which is estimated to be at least 15 minutes faster than the old run, the author uses an insane amount of segments. For those who might cringe at 100 segments, fear not cause the amount of precision required to get the time this low is justified.

Audio/Video is perfect, no signs of cheating detected.

Accepted this run is, yes Smiley

**Encoding note**
Since there are a lot of segments in this run, the author was nice enough to do each level as it's own. How hard would it be to combine all videos into one video for the final-final-final1 video?


Quote:
An amazing run for a game that terrorized my child hood. Tongue

The audio and video were both perfect as far as I could tell and there were no signs of cheating anywhere.

Accept


Quote:
This is the Outcast run we've all been waiting for. Mind blowing execution and surprising tricks throughout. So much for the blue milk run...

A/V is great. I would also like to see the run fully appended if possible.

Definite accept, the force is strong with this one Cheesy


Quote:
I originally wanted to write some indepth commentary on this run, but after rereading the supplied runner comments there really isn't much that would be of particular interest as a verification comment to the general reader that isn't already covered in them.

I'd like to quickly clarify how the trick in yavin_trials actually works, as some people have had trouble replicating it properly in the past. The little nobs on the pillar have a triggered event when they are force PUSHed to move to a predetermined spot on the map in a certain timeframe (iirc it's around 0.25 seconds, but it's been years since I checked the map files). Normally, when the pillar is fully extended and you are in the final area of the map, that spot is slightly behind where the nobs are, giving the impression of pushing them into the pillar. When you trigger that event while the pillar is still rising from the first encounter, the distance the nobs will have to cover in the same timeframe is significantly higher and thus, they appear to shoot upwards at a high speed, giving you a neat stepping stone to get onto the pillar. There, the base of the pillar and the cage on the top don't line up perfectly on the outer edge. If you stand on the edge of the pillar and get the entire system stuck by wedging yourself between the moving pillar and the ceiling, moving in certain patterns or specific angels at a high enough framerate will result in a single frame where you only stand on the pillar, lodging it into place as your body is still stuck against the ceiling, while the cage can ascend further. At that point there is enough space on the pillar below the cage for you to actually stand on properly and things continue as expected.

Beyond that, I don't think I can provide any technicalities that aren't covered by the runner comments to a satisfying degree, but I might step in and add more things to the knowledge base at some point, especially as there has been some minor interest in doing a RTA/races of this game recently, although I am still not convinced that's a good idea with the randomness involved.

A/V seems fine, although the bitrate of the Verification copies is rather low so I can't tell if there's some video issues that only become apparent at higher qualities.

High level of segmentation is very well justified in this game as tricks (and even just general movement) require very high levels of precission and enemies, especially at higher difficulties, are rather unpredictable and surprisingly accurate (especially with the E-11 Blaster Rifle, one of the worst main machine guns in FPS history and most likely the leading cause of the Empire's loss to the Rebels). There are a couple of saves that may not make up for the save penalty by themselves but lead to an overall smoother actual gameplay.

I second that a single video encode alongside the level by level format may be in the interest of the viewer as long as the segment breaks are still kept intact as on the current encode (I strongly dislike "cheating" the viewer's preception of segmentation with well timed segment cuts like some runs in the past have done). This should be possible from just splicing the current encodes into a single file without a full reencode.

Very strong accept, one of the outstanding runs of the past year.


Quote:
Having watched his original test run, I'm surprised at how much time he managed to shave off of that, so comparing it to the SDA run is just plain unfair. He hits a lot of tricks that I would have found completely inconceivable to pull off in a run. (the first level of nar shaddaa comes to mind). I'm also super impressed by how well he handles strafejumping up stairs. Having played FPS games for years I can tell you that is not an easy thing to pull off. The dark jedi fights are also really good.

Overall the level of optimisation is incredibly high, so a definite accept from me.

A/V quality is fine, I saw no disrepancies or signs of cheating in the run. I agree with [above verifier] that including segment breaks is a good thing for this run.


Decision: Accept

Congratulations to Gustav 'recon' Jensen!
Thread title:  
Might be magic...
Congrats, this was a fantastic run!
Yay. Thanks verifiers, and sorry for having you do all the extra work of verifying segments that I improved time and time again. Perfectionism, sometimes it gets the best of me.
And thanks gammadragon, glad you liked it!

Quote:
**Encoding note**
Since there are a lot of segments in this run, the author was nice enough to do each level as it's own. How hard would it be to combine all videos into one video for the final-final-final1 video?

Not hard, but well, time-consuming. I've already done it, though. It's one long video containing every segment, without visible segment breaks and with added jk2 music. That's definitely the version I prefer to have on the site. It's also similar to the previous youtube run and the new youtube run of the final run which I'll upload soon at some point.

Quote:
I second that a single video encode alongside the level by level format may be in the interest of the viewer as long as the segment breaks are still kept intact as on the current encode (I strongly dislike "cheating" the viewer's preception of segmentation with well timed segment cuts like some runs in the past have done). This should be possible from just splicing the current encodes into a single file without a full reencode.

Quote:
I agree with [above verifier] that including segment breaks is a good thing for this run.


I have to say I disagree - I personally don't prefer these kinds of videos. To me, watching a speedrun should be like watching a movie: non-stop action, no ads, interruptions or whatever. Visible segment breaks, even if they're only a second long, interrupt the flow of the run in my opinion and they take away some of the "entertainment value" that is contained in continuous action. While it's true that segment breaks are part of the run, I think the clearly visible information in the beginning of the video, and on the website, that the run uses heavy segmentation is enough. With that information, the viewer can sit back and enjoy the run knowing that its level of optimisation was reached with the help of very many segments, and that it therefore isn't as impressive as if ther were no segment breaks.

So, my argument largely comes down to this: Having one big video containing the entire run, with no visible segment breaks, gives greatest "viewing pleasure", at least for the novice person who just wants to see the run - or that's how I perceive it (and that's how I would prefer watching any run). Then, if the viewer wants to analyse the run with the purpose of knowing where segment breaks were placed, he should be able to download a copy of the run that has these segment breaks included.

However, I suspect what I've just said could spark off a great dispute, so with the intention of avoiding this, I suggest that two versions of the run should be added to the site:
- The one-vid version: one big video containing the entire run with no visible segment breaks, with added music etc. to make everything nice and smooth
- The verifiers' version: 24 videos, each containing the run of a level, with visible segment breaks and everything else untouched.
That way, everyone should be satisfied. Right?
1-Up!
If you append all of your segments together you'll need to keep the original audio, with no new music mixed in. I guess you could put it on track 2 like an audio commentary but the track 1 needs to be the original audio.
Edit history:
ShadowWraith: 2013-03-21 04:57:07 pm
.
I think he meant readding the game music in to the run since he recorded it with game music disabled. I don't like that idea as it gives the illusion that the levels themselves aren't segmented as the game music is continuous.
Run is a must see, super entertaining.
1-Up!
Quote from ShadowWraith:
I think he meant readding the game music in to the run since he recorded it with game music disabled. I don't like that idea as it gives the illusion that the levels themselves aren't segmented as the game music is continuous.

Ah. In that case yeah I'd say he can use it as a 2nd track if he wants but still the first track needs to be whatever the verifiers had. Needs some sort of mention in the comments in that case.
All right, if things have to be this strict (which is totally understandable) I'll just have one long video with visible segment breaks and everything untouched on this site as suggested, and then I'll host the version I prefer (with no visible segment breaks, music etc.) elsewhere so people can download that if they want to, or that version can be viewed on youtube.

on another note, I've made a few new improvements to this run. I couldn't resist the idea of breaking another minute barrier (43 minutes) (incl. save penalties, according to my timing), so I scrutinized parts of the run and improved segments in artus_topside, kejim_base and cairn_reactor, saving around 17-19 seconds in total (depending on whether cutscenes count in time or not). I might as well wait a little and see if can/will improve more before sending this in. And if it's not too late to send improvements in, how do I go about that? And was there made any decision on how timing is done (i.e.  whether cutscenes count toward final time)?
Edit history:
Flip: 2013-03-24 02:44:04 pm
Flip: 2013-03-24 02:43:56 pm
1-Up!
Yes, I recommend getting to the point where you aren't improving it every week and then submit Smiley

edit: submit as an entirely new run at that point. IsraeliRD can answer your timing questions.
Edit history:
ExplodingCabbage: 2013-03-24 03:30:03 pm
ExplodingCabbage: 2013-03-24 03:29:36 pm
ExplodingCabbage: 2013-03-24 03:29:27 pm
ExplodingCabbage: 2013-03-24 02:55:05 pm
Possible relevant precedent as far as music goes:

1) RandomEngy and I recorded the Jedi Knight 1 speedrun without music and edited it back in. I don't know if Flip knew about it, though, and since we were doing one-segment-per-level the illusion-of-not-being-segmented issue that ShadowWraith raises didn't apply to us. I'm not sure if Flip even knew about this at the time of accepting that run, though (we mentioned we were doing this in the planning thread, but I don't think it was in the comments and I seriously doubt it was mentioned at verification time) so maybe we slipped through something that would've been objected to if it had been pointed out. I don't see any reason not to allow this in our case, though, where ShadowWraith's objection doesn't apply.
2) The old Half Life 1 speedrun by SpiderWaffle (now removed as part of the scripted run purge) had music added back in, and did have mid-level saves. (IIRC music completely stops any time you save and load in HL1, so having it play through the loading screens would very much be a case of the kind of misleadingness ShadowWraith is concerned about).

There are probably other cases of this - perhaps quite a few - but these are the only two I know of.



As far as appending segments together seamlessly goes (i.e. trimming out save screens), that's allowed for segmented PC runs and there's plenty of precedent for it. Be aware though that some members of the community disapprove of it. If I remember right - and perhaps I don't - dex doesn't like this in part because when runs get posted on YouTube, without the notes and details present on SDA, it's not clear to the viewer that the run is segmented. I think I recall him discussing the old HL1 run in this context, and how YouTube commenters were remarking about the awesome skill the runner must have, presumably unaware that it was segmented a little over once every 10 seconds. I don't know the views of any of the other green-name-brigade members on this, though.

Quote from recon:
I have to say I disagree - I personally don't prefer these kinds of videos. To me, watching a speedrun should be like watching a movie: non-stop action, no ads, interruptions or whatever. Visible segment breaks, even if they're only a second long, interrupt the flow of the run in my opinion and they take away some of the "entertainment value" that is contained in continuous action. While it's true that segment breaks are part of the run, I think the clearly visible information in the beginning of the video, and on the website, that the run uses heavy segmentation is enough.


FWIW, in the context of this run, I agree with you. I don't agree with you extending that judgement to speedruns in general, though. Smoothness and ease-of-viewing is important for fast-paced runs of action games, sure, but for something like a slow-paced but highly technically-complicated run of an RPG featuring lots of save glitches, heavy-duty luck manipulation that's highly save-reliant, and restrictions on when or how often you're allowed to save that have to be worked around, it can easily become the case that the value in terms of added clarity and information of showing the viewer when you're saving is high, while 'smoothness' just isn't important at all.
1-Up!
Yeah I don't think there are others but I'm not 100%.

Like I said, recon is welcome to supply an appended version and use a music track on track 2 in addition to the original segmented version. Also of course he can host it on his own in whatever form he wants.
Thanks for the feedback Flip and ExplodingCabbage.
Quote from Flip:
edit: submit as an entirely new run at that point.

I don't want to seem know-all or anything, but would that not be making things harder than they have to (i.e. sending in all of the files again compared to a few)? I mean, I can send the individual improvements which are no more than maybe 4 or 5 files in total and then the verifiers can check these - and I can have these files ready in a matter of days. Or perhaps the issue is whether or not the "old" verifiers can verify again, and if not, that new are needed? In any event, I realise it's my own fault I submitted too early when there was indeed room for improvement, so I'll just do whatever is the easiest for you. Sorry about that inconvenience.

Quote from ExplodingCabbage:
FWIW, in the context of this run, I agree with you. I don't agree with you extending that judgement to speedruns in general, though. Smoothness and ease-of-viewing is important for fast-paced runs of action games, sure, but for something like a slow-paced but highly technically-complicated run of an RPG featuring lots of save glitches [...].

Yeah, good point. I mainly had action games (FPSs) in mind when preferring no save menus etc, so your example is definitely an exception.
Not a walrus
Quote from recon:
I don't want to seem know-all or anything, but would that not be making things harder than they have to (i.e. sending in all of the files again compared to a few)? I mean, I can send the individual improvements which are no more than maybe 4 or 5 files in total and then the verifiers can check these - and I can have these files ready in a matter of days. Or perhaps the issue is whether or not the "old" verifiers can verify again, and if not, that new are needed? In any event, I realise it's my own fault I submitted too early when there was indeed room for improvement, so I'll just do whatever is the easiest for you. Sorry about that inconvenience.


You don't have to send in all the files again, or even need to find new verifiers, he just means send puwexil the info as if it were a new submission and then send the changed files. We'll work the necessary magic behind the scenes.
Great. Just to be sure, I'll let you know that I've now attached a new torrent file containing the final run improvements. I put it under the "MyRuns" feature where you can upload new torrents.

And, some info on the new improvements that's probably relevant to the verifiers: (I don't mind anyone else reading, so why not put it here)


I managed to shave off about 36 seconds of the run (according to my timing) by improving segments in levels kejim_base, artus_topside, bespin_undercity, cairn_reactor, ns_starpad and doom_detention. In detail:

kejim_base (level 2): I improved everything after the second segment of the level, so you can skip to about 30 secs into the video if you only want to see  the new segments. The main reason I redid the rest of the level was that I found out it was possible to use Saratura's awesome laser mine skip later in the level, despite only having 30-some hp. It was possible with precise timing. Thanks to Saratura for that neat trick. However, to actually succeed in this very difficult skip, I felt I needed an additional segment break. Enjoy the shortest segment of the run - it's about 4 segments long. It's worth it, though, in my opinion anyway (the reason I felt an extra segment break was needed after the mine jump was that the following room is a dead hard part to complete). I also managed to complete the rest of the level significantly faster, and I got a much smoother run in the narrow room with with a ton of stormtroopers. 7-9 seconds saved according to my timing depending on whether cutscenes are counted in timing.

artus_topside (level 5): I improved the last three segments of this level, so you can start watching the new stuff from 01:07 onward. The main difference is a faster (and simpler) way of first destroying the imperial officer whose keys you need to pick up, and then getting down to that place. I do take a bit more fall damage from this method, but that's no problem since I somehow manage to not get hit despite the many enemies. Other than this, I just smoothed things out. It's not perfect, but faster. 4 seconds saved according to my timing.

ns_starpad (level 10): I could say a lot about the complexity and randomness of the annoying hangar scene but I'll try to keep this short. The improvement of this level consists only of an improved third segment (the segment containing the hangar scene), beginning at 1:09 in the video. The reason I redid this segment was first and foremost that I randomly discovered that honorableJay's awesome convo skip is applicable in this scene. By jumping to the end of Lady Luck and sticking your saber into a wall, Lando shuts up and the first wave of enemies appears - and so it begins. Other than this, the whole thing is smoothed out a lot. I also pick up more thermal detonators that some of the three-eyed aliens drop if you're lucky. With these extra detonators I can wipe out the last enemy wave way faster than if I had been using the saber.
Finally, the end goal of this segment is to make Lando say "Kyle, we've got a problem", because this makes you able to do the other stuff you need to to be able to finish the level (refuel the ship or whatever). I found that to have Lando say this, you firstly need to be finished with all the enemy waves, and secondly, you must step just a little bit inside the Lady Luck (on the end of the ramp leading to inside of the ship is sufficient). After some amount of seconds, he says the line. In this improved segment you'll not hear him say the line, and that's because he says it at a point after I've opened up the menu and saved. However, when opening this saved file, he says the line shortly after, which means that the segment was indeed "valid". Total saved time is about 11 seconds (according to my timing).

bespin_undercity (level 11): I decided to redo the first three segments of this boring level with the purpose of speeding up the "air winds / gust" scene, but overall, every segment contains improvement, the biggest one being in the third segment (the gust scene). With great luck you can have every gust appear in time, so that you're always in motion. I had that much luck, and so, the total improvement is a little over 4 seconds according to my timing.

cairn_reactor (level 16): I improved the third segment and the rest of the level from there. The new segments start at about 48 secs into the video. The improvements are mainly just smoothing things out, and this is particularly apparent at the end of that rotating cylinder room where I sort of "jump" through the door instead of slowing down before it as I would in the old segment. The last segment also contains a significant improvement that consists of a little faster way down from the highest floor. I can't explain that, but you should be able to see the difference. I also use a bacta here, which I would not do in the old segment. This is safe, though, because I end this speedrun with 4 bactas that were all picked up in the Nar Sha Daa levels, so I'm not out of supply at any time. Seconds saved: about 5,5 in total.

doom_detention (level 19): I improved the first segment of this level by smoothing things out. There's also a small new timesaver I use: In the room before "the big room" (with the spacecrafts) I force push a stormtrooper toward a door, which actually opens the door for me. This means I can maintain my speed and get faster by. Time saved: nearly 2 seconds.
Not a walrus
Just to be clear, are these videos that people have already seen, or are they new since the decision was posted?
They're entirely new, seen by nobody but me. So they're new since the decision was posted, yes.
Not a walrus
Ok, that's a bit different, then, since this run is already past verification.

When I said 'attach the new files', I mean send in a new submission and attach the new files to THAT, not to the existing run. Sorry if that was unclear.
Edit history:
Flip: 2013-03-28 04:06:41 pm
1-Up!
Right, you need to contact puwexil with your submission info first, that way we know what we're getting.

edit: here's our official statement - we'll see about getting this implemented into the rules.

"We'll accept improvements while runs are in verification but once a decision is posted, improvements to a run will have to be treated as a new, independent submission."