After crashing, our alien realizes the fastest way he's getting back out is by convincing his enemies to carry countless weapons only he can use. Including, for some reason, buildings and freaking scorpions.
Note to encoders: Please use this file. The file I'm submitting is for purposes of record keeping. The suggested file is identical, except to insert a little something at the name entry, leaving the epilogue and credits before then to play out naturally.
Chatty subtitles here. In lua script form, for watching with an emulator.
  • VBA v23 svn270
  • Aims for fastest time
    • An alternate file for name entry is provided, however.
  • Uses death to save entertainment
    • I have not found a moment where it saves actual time.
    • Unless you count unneeded luck manipulation.
  • Uses hardest difficulty
  • Abuses clouds to save time
Any tricks I can think of are in GameResources/GBA/AlienHominid. Visit that page for details on some of the tricks in this run.

About the run

The Dream Team Contest #3 used this game for the competition. The contest would have players sign up and work together as teams in order to produce the fastest movie within a time limit of a few weeks. I was part of one of the teams in there, Team 5, and things were a bit hectic. We would start by analyzing the game swiftly, produce a particularly useful script, then start crushing each stage we dash through. I still recall the reaction from a teammate when I had already implemented rescaling the hitbox display when asked to find a way to show offscreen enemies.
It would appear that we had a good strategy and did a very good job at figuring out the tricks to the game. Not really one to brag, but the fact remains, the team I was part of was the winning team. I knew there were better ways to get through the game, so I had started working on improvements. Shouldn't take too long.
Over two years later (possibly why DTC #4 requires the winning team to submit their movie), I finally complete this run. Part of the reason it took so darn long is due to having lost and regained interest over time, and various life stuff. This run never left my mind, not for very long. When things stabilized for a bit and I finally came back to my TASing computer, I finally got the finishing touches needed to get through 3-3 and 3-4.

About the game

You control some sort of yellow alien with a gun. For whatever reason, he also has very Earth-like grenades with Earth-like explosions. He collides with some sort of satellite around Earth and crash-lands right in front of the FBI, who then runs off with his spacecraft and somehow forgets that bright yellow alien thing in clear sight. If you're looking for story, this isn't the game, as it just gives an excuse why you're doing these things.
The game play is similar to Metal Slug from what I hear. You have a gun, you will knife things standing next to you, you can throw grenades, there are vehicles to enter, and there are different weapons to pick up. But while the core game has similarities, there's plenty enough differences between the two. This one being a fair amount simpler, but your yellow alien can dig into the ground or latch onto the heads of some of his enemies.
Simple doesn't necessarily mean easy, though. In hard mode, most attacks kill you outright, and you have 5 spare lives per credit. It's pretty easy to end up catching a stray bullet or losing yourself in the action. But this TAS just runs right through the game with little worry from the various hazards the game would throw at you.

Frame breakdown

Measured mostly by when the camera unlocks.
Comparisons are made against the teams from Dream Team Contest #3:
T5 - FatRatKnight, micro500, NitroGenesis, pirate_sepiroth
T1 - Xkeeper, alden, Acmlm, Kumquat
T2 - DarkKobold, adelikat, Cardboard, Tompa
T4 - Sonikkustar, MUGG, Aqfaq, Mukki
PlaceThis Run#T5T1T2T4Notes
Screen
Clear
1474 1475147814741474Last A button press before gameplay
1-1: 12674
(1200)
2675
(1200)
2678
(1200)
2696
(1222)
2701
(1227)
1-1: 22884
(210)
2885
(210)
2892
(214)
2968
(272)
3018
(317)
1-1: 33039
(155)
3040
(155)
3074
(182)
3214
(246)
3226
(208)
1-1: 43245
(206)
3248
(208)
3282
(208)
3437
(223)
3494
(268)
T5 lagged a frame here
1-1: 53653
(408)
3654
(406)
3706
(424)
3841
(404)
4109
(615)
Framed rule'd
1-1: 63814
(161)
3816
(162)
3913
(207)
4170
(329)
4357
(248)
1-1: 74844
(1030)
4846
(1030)
5005
(1092)
5329
(1159)
5515
(1158)
1-1: 85302
(458)
5304
(458)
5478
(473)
5822
(493)
5987
(472)
1-1: 95526
(224)
5528
(224)
5687
(209)
6197
(375)
6257
(270)
Frame rule'd (I'm 5 frames slow!)
1-1:105705
(179)
5707
(179)
5936
(249)
6445
(248)
6532
(275)
1-1:116463
(758)
6476
(769)
6713
(777)
7249
(804)
7629
(1097)
1-1 END7180 7193743079668346
Screen
Clear
7664 7677791484508830
1-2: 18013
(349)
8058
(381)
8289
(375)
8912
(462)
9177
(347)
T4 didn't get grenades.
1-2: 28104
(91)
8183
(125)
8390
(101)
9051
(139)
9278
(101)
1-2: 38309
(205)
8401
(218)
8606
(216)
9309
(258)
9500
(222)
1-2: 48460
(151)
8584
(183)
8769
(163)
9487
(178)
9710
(210)
1-2: 58730
(270)
8861
(277)
9054
(285)
9782
(295)
10024
(314)
1-2: 68952
(222)
9110
(249)
9312
(258)
10108
(326)
10403
(379)
Snuck in under frame rule!
1-2: 79183
(231)
9354
(244)
9552
(240)
10361
(253)
10668
(265)
1-2: 89600
(417)
9768
(414)
9970
(418)
10872
(511)
11109
(441)
Huh, frame rule?
1-2: 910048
(448)
10229
(461)
10419
(449)
11327
(455)
11600
(491)
1-2:1010566
(518)
10798
(569)
10937
(518)
11850
(523)
12178
(578)
1-2:1110790
(224)
11098
(300)
No
Lock
No
Lock
12522
(344)
1-2:1211235
(445)
11551
(453)
11616
(679)
12529
(679)
12973
(451)
1-2:1311852
(617)
12168
(617)
12234
(618)
13146
(617)
13590
(617)
T1 didn't position camera
1-2:1413263
(1411)
13629
(1461)
17013
(4779)
17926
(4780)
18370
(4780)
Eat boss. Yum.
1-2 END13589
(6409)
13955
(6762)
17341
(9911)
18255
(10289)
18698
(10357)
Screen
Clear
14125 14441178271874119184
1-3: 114505
(430)
14867
(426)
18288
(461)
19213
(472)
19658
(474)
Frame rule up ahead
1-3: 214825
(320)
15187
(320)
18607
(319)
19534
(321)
19983
(325)
Same excuse as before
1-3: 315187
(362)
15549
(362)
18971
(364)
19908
(374)
20400
(417)
1-3: 415336
(149)
15705
(156)
19130
(159)
20124
(216)
20596
(156)
There's the frame rule!
1-3: 515460
(124)
15829
(124)
19284
(154)
20311
(187)
20813
(217)
1-3: 615863
(403)
16263
(434)
19719
(435)
20786
(475)
21281
(468)
Checkpoint16409
(546)
16819
(556)
20278
(559)
21345
(559)
21847
(566)
Minigame17294
(885)
17717
(898)
21164
(886)
24472
(3127)
DNF
1-3 END17588
(3999)
18011
(4056)
21462
(4121)
24766
(6511)
Screen
Clear
18267 186912213725445 Strange minor variance
in frames, here...
1-4
Heli1
18984
(727)
19473
(782)
22901
(764)
26239
(794)
1-4
Heli2
19362
(378)
19877
(404)
23295
(394)
26667
(428)
1-4
Heli3
19733
(371)
20264
(387)
23676
(381)
27176
(509)
1-4
Heli4
19881
(148)
20413
(149)
23827
(151)
27424
(248)
Huh. Must have mis-measured.
1-4
Heli5
20024
(143)
20558
(145)
23975
(148)
27627
(203)
1-4
Heli6
20172
(148)
20704
(146)
24148
(173)
27862
(235)
I fear the harmless background object
Vicious RNG-affecting clouds!
1-4
kills
20510
(338)
21095
(391)
24613
(465)
28219
(357)
1-4
boss
20923
(413)
21513
(418)
25050
(437)
28642
(423)
1-4 END21338
(3755)
21928
(3917)
25465
(4003)
29057
(4291)
We're through with the first third of the game! Oh, there's more tables to go through, of course! But at least we're down a team.
PlaceThis Run#T5T1T2Notes
Screen
Clear
22014 226032614029732
2-1: 122276
(262)
22900
(297)
26436
(296)
30071
(339)
2-1: 222379
(103)
23036
(136)
26542
(106)
No
Lock
2-1: 322555
(176)
23237
(201)
26776
(234)
30528
(457)
2-1: 422962
(407)
23644
(407)
27183
(407)
30979
(451)
2-1: 523086
(124)
23857
(213)
27377
(194)
31257
(278)
2-1: 6No
Lock
No
Lock
No
Lock
No
Lock
2-1: 723527
(441)
24323
(466)
27813
(436)
31685
(428)
Er, uh... What happened?
2-1: 823733
(206)
24540
(217)
28049
(236)
32029
(344)
2-1: 923919
(186)
24796
(256)
28299
(250)
32250
(221)
2-1:1024068
(149)
24969
(173)
28478
(179)
32424
(174)
2-1:1124386
(318)
25447
(478)
28884
(406)
32870
(446)
2-1:1224575
(189)
25623
(176)
29076
(192)
33061
(191)
Manipulating a weapon...
2-1:1325111
(536)
26165
(542)
29785
(709)
33823
(762)
2-1 END25538
(4200)
26592
(4664)
30212
(4747)
34250
(5193)
Screen
Clear
26023 270773069734735
2-2: 127188
(1165)
28242
(1165)
31895
(1198)
35902
(1108)
2-2: 227429
(241)
28513
(271)
32136
(241)
36146
(244)
2-2: 329603
(2174)
30689
(2176)
34310
(2174)
38320
(2174)
2-2: 429766
(163)
30895
(206)
34513
(203)
38523
(203)
2-2: 530132
(366)
31307
(412)
34883
(370)
38907
(384)
2-2: 630201
(69)
31362
(55)
34948
(65)
38966
(59)
Variance in strategy.
2-2: 730255
(54)
31416
(54)
35007
(59)
39027
(61)
2-2: 830309
(54)
31481
(65)
35066
(59)
39086
(59)
2-2: 930491
(182)
31661
(180)
35248
(182)
39281
(195)
I blame said variance.
2-2:1030592
(101)
31763
(102)
35355
(107)
39394
(113)
2-2:1130784
(192)
31966
(203)
35549
(194)
39588
(194)
Timing used:
Yeti in control
2-2:1232639
(1855)
33842
(1876)
37511
(1962)
41499
(1911)
Timing used:
Victory stance
2-2 END32865
(7327)
34068
(7476)
37737
(7525)
41727
(7477)
Screen
Clear
33349 345523822142211
2-3: 134518
(1169)
35737
(1185)
39398
(1118)
43364
(1153)
T2 spent grenades.
I did not, here.
2-3: 234937
(419)
36156
(419)
39820
(422)
43808
(444)
2-3: 336110
(1173)
37509
(1353)
41037
(1217)
45069
(1261)
Checkpoint36170
(60)
37568
(59)
41097
(60)
45175
(106)
Bad camera position
2-3 END36452
(3587)
37850
(3782)
41379
(3642)
45457
(3730)
Screen
Clear
36937 383354186445942
2-4
Kills
38147
(1210)
39773
(1438)
43182
(1318)
47466
(1524)
2-4
Boss
38842
(695)
40544
(771)
44021
(839)
49063
(1597)
2-4 END39258
(2806)
40960
(3110)
44437
(3058)
49479
(4022)
Don't worry! There's only a third of the game left! Take a look!
PlaceThis Run#T5T1T2Notes
Screen
Clear
39931 416334511050152
3-1: 140412
(481)
42141
(508)
45749
(639)
50634
(482)
3-1: 240782
(370)
42533
(392)
46217
(468)
51103
(469)
3-1: 341269
(487)
43021
(488)
46705
(488)
51592
(489)
3-1: 441710
(441)
43463
(442)
47286
(581)
52115
(523)
3-1: 542526
(816)
44304
(841)
48105
(819)
52952
(837)
3-1: 643174
(648)
44972
(668)
48920
(815)
53827
(875)
3-1: 744500
(1326)
46529
(1557)
50510
(1590)
55455
(1628)
Checkpoint44582
(82)
46630
(101)
40611
(101)
55616
(161)
Victory stance
3-1 END44808
(5550)
46856
(5896)
50837
(6400)
55842
(6363)
Screen
Clear
45294 473425132356328
3-2: 145635
(341)
47686
(344)
51709
(386)
56676
(348)
3-2: 245763
(128)
47837
(151)
51848
(139)
56812
(136)
3-2: 345880
(117)
47957
(120)
51948
(100)
56915
(103)
Different
strategy
3-2: 445943
(57)
48044
(87)
52019
(71)
56986
(71)
3-2: 546319
(376)
48435
(391)
52402
(383)
57527
(541)
3-2: 646484
(165)
48591
(156)
52557
(155)
57755
(228)
Bad position, was
manipulating luck
3-2: 747039
(555)
49197
(606)
53105
(548)
58352
(597)
That same bad
position.
3-2: 847518
(479)
49728
(531)
53715
(610)
58897
(545)
3-2: 947734
(216)
50017
(289)
54009
(294)
59235
(338)
Checkpoint47751
(17)
50034
(17)
54027
(18)
59252
(17)
3-2 END47976
(3168)
50260
(3404)
54253
(3416)
59478
(3636)
Screen
Clear
48460 507455473859963No idea where I
got that frame
3-3: 149460
(1000)
51813
(1068)
55794
(1056)
61002
(1039)
3-3: 249827
(367)
52243
(430)
56138
(344)
61401
(399)
I blame the new
smashers timing
3-3: 350055
(228)
52470
(227)
56380
(242)
61673
(272)
3-3: 450376
(321)
52777
(307)
56692
(312)
62036
(363)
Failed improvement
starts here
3-3: 550652
(276)
53083
(306)
56986
(294)
62392
(356)
3-3: 650983
(331)
53410
(327)
57314
(328)
62797
(405)
Poor manipulation
3-3: 751523
(540)
53951
(541)
57873
(559)
63384
(587)
Check 51843
(320)
54297
(346)
58191
(318)
63707
(323)
Poor positioning
3-3 END52065
(4089)
54519
(4259)
58413
(4160)
63929
(4451)
Screen
Clear
52551 550055889964415
3-4
Boss
53388
(837)
55850
(845)
59947
(1048)
65382
(967)
Huh, miscounted.
3-4 END54523
(2458)
56985
(2466)
61082
(2669)
66517
(2588)

Stage by stage

I figure I should mention stuff about individual stages.
Ironically, the first and second place teams were slower than the other two here. Said other two is apparently frame-perfect.
Set difficulty to Hard. This affects your lives, continues, shield strength (does not exist in Hard), and grenades. If it has other effects, I did not notice them, but I didn't look hard. Standard "show mastery over game using hardest mode" type of thing.
I alternate A and start when and where possible, as the game specifically doesn't read input until it's needed, showing up in VBA as lag frames. Sticking with A would mean I wait out a non-lag frame with null input, whereas start can still be detected. Then again, the game only wants A at times. This is primarily how the slower teams were ahead on the menu.

1-1 . The first level

T5 did well. 12 frames suboptimal according to this run.

1-1 Scene 1

Get a power-up from helpful kid; 1 kill is needed.
The spawn timer begins when you collect that weapon pick-up from the helpful kid, and thus starts the frame rule. It is obtained by dashing on the first possible frame, and manipulated into Wpn6. The grenade pack was trivially manipulated from the building. The reason to manipulate Wpn6 is due to the length of death animations, and will see lots of use in later scenes.

1-1 Scene 2

5 kills are needed.
The game was generous, as an agent dropped another grenade pack just for us! Miraculously, very little manipulation was needed, the RNG was already good. I damage two buildings at once with two grenades, with two more destroying them separately. If the buildings are simultaneously destroyed by a single grenade, they only drop one power-up.

1-1 Scene 3

5 kills are needed.
The spawn timer was perfectly in sync, down to the very frame. When the camera locks in place, the death animation of one agent finishes and another spawns. The kill was counted.
During this scene, I am charging. I also shoot without losing this charge, and this is the first instance where I shoot while keeping charge. I also pick up Wpn7, as I will soon fight enemies with more than 1 HP. The spread shot fires 3 distinct bullets at a time, meaning 3 damage instead of 1. The graphics lie -- I'm not firing 6 shots at once, my hitboxes script claim only 3 hitboxes are there.

1-1 Scene 4

Destroy a particular enemy: FBI Van
The vehicle I hop into has a speed of 2.5 pixels/frame. Dashing averages out to around 2.78 pixels/frame. Despite the vehicle being slightly slower, I hop in because of the tremendous advantage of being able to shoot while moving, which dashing around won't allow with such frequency. And besides, even in the vehicle, I still have to wait a few frames for the camera to fully scroll in place. It's too bad I only use it for this part of the entire run.
The FBI Van took a full-power 10-damage charge and a stream of spread shots. It didn't live.

1-1 Scene 5

Destroy boss.
T5 already did a stellar job up to this point. Other than a single lag frame, I couldn't make any improvements thanks to the frame rule called the Spawn Timer. It's thanks to this frame rule that I believe it is impossible to improve even by a single frame.
I tried a cool strat involving ice shots, in an attempt to glitch the hit detection, but alas, there was far more lag than frames saved from the battle. I haven't checked if I beat the frame rule with the ice shots, but with that much lag, I don't think it would help.
I intentionally delay the kill by a few frames. I'm still getting within the frame rule for the next scene anyway, but this is so I can more easily copy what T5 did.

1-1 Scene 6

5 kills needed.
I did what T5 did: Synced up the RNG, jump, throw grenades, knife something, and have the timing down to the frame on every kill. I would like to hear if anyone can come up with a different RNG that still beats this scene in the same amount of time.
For luck manipulation, I fired the last two spread shots I had. Every shot of Wpn7 counted for something. I love this stage! Wpn6 was picked up in time for the last kill.

1-1 Scene 7

12 kills needed.
Enemies freeze on the ground and ignore your shots as the camera pans up... Or at least that was intended. Apparently the code to handle that was poorly done. So, off-screen, I get two packs of grenades. Lazy luck manipulation at its finest, when you can see the off-screen hitboxes and have an RNG scanner script.
I should hope it's obvious that I keep a charged shot ready while still shooting every agent that comes down.

1-1 Scene 8

Destroy a particular enemy: Heavy Fire Agent
This enemy takes double damage from grenades. I'm not sure why, really. Regardless, this enemy is also why I charged my shot in the first place. There is a faster way to kill this enemy, by digging into the ground and burying him, but you will die by doing this, and only get one "frame rule" ahead in return. I glanced over the RNG scanner and came to the conclusion I would not be able to recover my stock of grenades and weaponry in a timely manner.

1-1 Scene 9

4 kills needed.
Incidentally, about that frame rule... I was 5 frames too slow to reach this scene to beat it. Thus, the frame rule wasted 26 frames. Ouch.
To the extent that I can tell, the only allowed spawn position for the agents is coming down on ropes from the sky. If they could spawn elsewhere, you wouldn't be seeing them on ropes coming down in this run.

1-1 Scene 10

4 kills needed.
The ropes still cause nightmares until the camera locks in place, but it's already done its damage. I manipulated the first one to be as far right as possible to still be in sight for a needed kill.
There is another vehicle I can hop into here, but the mobile offense provided here has zero value. I manipulate up Wpn7 and take it once I fire the last shot I need to.

1-1 Scene 11

Destroy boss.
I carefully measured the dashes and made sure that my last dash stops right when I've gone far enough to lock the camera and thus summon the boss. A 10-damage charged shot, grenades, and numerous spread shots from Wpn7 happened.
There is one point during the battle where I stopped shooting momentarily, when I jump to avoid the only attack the boss makes. While I do make a few downward shots, there's an odd thing about Y velocity the game does: While I am still traveling upwards, downward shots have a minor effect on your Y velocity, while if you're in the process of falling back down, a downward shot actually pushes you up much harder. I can afford the minor pushes easily, but the major shove wastes too much time.

1-2 . This one has tanks!

T1, up until the boss, beat the other teams, being a mere 134 frames slower than this run. Then again, they didn't eat the boss, costing them 3318 frames for that oversight, but who's counting?
The spawn timer begins once you travel far enough, so some effort was made in triggering it early.

1-2 Scene 1

Destroy a particular enemy: Shielded Agent.
It's possible to jump over the shield and kill the guy behind it to pass this scene. But it turns out it's faster to just blast the 20 HP iron shield right through.
T4 did it surprisingly fast. Even this run unlocks this scene 2 frames slower. This route through was inspired by T4's efforts. While it is possible to beat T4's time here, the RNG didn't agree, as I needed a grenade drop as well. Regardless, what I have here lets me beat a frame rule, without a frame to spare, later on.

1-2 Scene 2

Destroy a particular enemy: Shielded Agent.
A scene that is not really noteworthy. Spending grenades on that building isn't possible without wasting time.

1-2 Scene 3

3 kills needed; Destroy a particular enemy: Helicopter.
This scene has an interesting combination of requirements. The 3 kills I mentioned does NOT include the helicopter -- It has to be standard agent kills. I meet both requirements within a few frames of each other. Priority is made to destroy the helicopter quick, as that's the longer requirement to fulfill. The Spawn Timer is not the limiting factor here, not by a long shot.
There are simultaneous manipulations taking place, where I have to get the right weapon from the helpful kid (Wpn7), as well as pick up grenades from the buildings. This scene turned out pretty well.

1-2 Scene 4

Destroy a particular enemy: Tank.
Kind of a breather scene following that last tricky one. Which is good, as I had to redo these parts a few times.
I still need to set the dashing into the next scene optimally, though.

1-2 Scene 5

Destroy a particular enemy: Tank.
Apparently, they thought that since the first tank didn't beat me, the second one surely will!
While I do spawn grenades off an enemy, this is actually a byproduct of manipulating both agent's death animations. I needed them to take long enough dieing for them to count in the next scene. By the way, there was no way I could take the grenades, but I didn't need them anyway.

1-2 Scene 6

6 kills needed.
When I got to this scene, my script says the camera locks right when the Spawn Timer reads 30. This was a pretty triumphant moment when I saw, once and truly, I could beat the frame rule. It was down to the very frame. I had earlier seen a possibility that I could beat the frame rule, and worked towards it, restarting this level several times, even messing with 1-1's ending for new RNGs, just to see if I could. And I did.
Had that earlier kill finished one frame early, it wouldn't count. Had it finished one frame late, it would have blocked a new agent from spawning. Imagine how I must have felt when I had hit such a tiny target.
Alas, would have been perfect if I had Wpn6, but I have to make do with grenades, Wpn7 shots, and manipulate the shortest death animations.

1-2 Scene 7

Destroy a particular enemy: Tank.
That first building I blew up to mess with the RNG. I take Wpn7 from it anyway, but it was unneeded. The second one I blow up is apparently a grenade shop. It ignores the RNG and always drops a special grenade pack containing 30 grenades. This serves for the rest of the entire stage.
The tank was nothing special. Their third attempt just doesn't work out!

1-2 Scene 8

6 kills needed.
The frame rule comes in full force here. As I did not believe I could save anywhere near enough frames to beat it by one "frame rule", I let this attempt through. Nor sure how I'm slower than other runs here.
During this scene, I manipulate Wpn6 and Wpn7. Although Wpn6 use was brief, there are no more "kill x agents" scenes later on. I may as well save every last frame I can.

1-2 Scene 9

Destroy a particular enemy: Heavy Ice Agent.
Same deal as with the Heavy Fire Agent in 1-1.
Although, I should have checked if I could bury him instead. I would eat a death, lose my grenades, and possibly have a difficult time handling luck later on. I may mark this as "possible improvement." All I can say is good luck, as manipulating it for needed things will be tough.

1-2 Scene 10

Destroy a building.
Here's the reason why I pick up Wpn1, the green cloud weapon: The building handles hit-detection poorly. There's no check in place to prevent lingering hitboxes, such as those from green clouds, fire, or ice weapons, from hitting repeatedly. It happens that Wpn1 has a larger hitbox than Wpn2 or Wpn3, meaning I can start damaging the building earlier.
The building will always drop something. I manipulate it into being Wpn1 so that the item drop doesn't interfere.

1-2 Scene 11

Destroy a building.
Same thing, but rather than charge through, I jump over the building. I carefully checked, jumping over saves a few more frames than just dashing straight through, where it won't help with the previous building.
On the way to the next scene, I knife an agent. What luck that he should have Wpn7, the spread shots! Exactly what I needed!

1-2 Scene 12

Destroy particular enemies: 2 Tanks.
The tanks will always shoot with the exact same timing. There are two ways to stop this: Destroy the tank before it shoots, or stand next to one. I can't fulfill either on both tanks at once, so I must jump to avoid an attack. It otherwise goes smoothly enough.

1-2 Scene 13

Wait out an animation.
Some sub-pixel manipulation of the camera on the previous scene in order to let it lock one frame earlier here. Nothing major.
With time to kill, I show off how many grenades I picked up, and how lucky I can be. Not a lot to do other than set myself up to fight the end boss. I had taken the spread shots just in case I needed to manipulate the RNG. Turns out I needed to avoid as many RNG calls as I could, meaning the weapon wasn't fired at all.

1-2 Scene 14

Destroy boss.
It's instant pudding! Though, despite being pudding, only one team had known to eat the darn thing. In fact, only one person among the entire DTC3 found this out. That person... Is NOT me. No, even I'm not that awesome. pirate_sephiroth of T5 is the one to have found it.
In order to eat it, you must first hit the hydrant so that it soaks the pudding up just right for eating. It won't let you hop on its head if you don't do this. Then hop aboard its head just like you would on any regular agent. You can bite 5 times before you are forced to jump off, but there is a 1 frame window where you can jump right back on and repeat the eating process.
The alternative of hitting the hydrant over and over is 3318 frames slower than chowing down. Mainly due to the fact you have to wait for the pudding boss to stop holding it down each time.
However, Mugg did find a trick with this boss. You can hit the hydrant and have it spray more water at this boss. In order to register the attacks to the hydrant, the giant pudding monster must not be in a "swat nasty alien away" animation. To avoid the repeat of this action, I must travel to the left. I made sure to shoot at the earliest possible frame where my hits register, and even manipulated the luck of my bullet positions to do so.
Oh, yeah. There's also a cooldown between hydrant sprays, justified by that pudding monster holding it down. Must wait for that, too.

1-3 . More of the usual, plus a minigame

T5 was pretty fast. This run is only 57 frames faster than T5. This is also the level where T4 had stopped. It's alright, you had a shining moment that helped improve this run.

1-3 Scene 1

Destroy a particular enemy: Wall on wheels
First, standard manipulation of Wpn7 and grenades. I don't mind being "slow" here...

1-3 Scene 2

Destroy a particular enemy: Helicopter
Next, standard anti-air routine of jumping. I'm not too worried about every last frame...

1-3 Scene 3

Destroy a particular enemy: Heavy Fire Agent
One of these days, I should remind myself to check if burying him is faster than conventional shooting. Please label this as a "possible improvement". Ammo is certainly far less strained here than in 1-1, that is certain.

1-3 Scene 4

4 kills needed.
Spawn killing scene. This therefore means the Spawn Timer works in full force, so any frame "losses" earlier don't count thanks to the frame rule here.

1-3 Scene 5

4 kills needed.
Standard spawn killing scene. Nothing special.

1-3 Scene 6

5 kills needed.
Standard spawn killing sce- Yeah... No. This demonic scene was taunting me about its frame rule during the DTC3, which I failed to get. But at last I've come back for it!
I can't avoid grabbing the helpful kid's weapon, or else I don't lock the camera in time for a spawn. Meaning I must manipulate Wpn6 to prevent frame loss. This affects where the spawns come from. And, of course, I must also time the kills along the way carefully, not to mention have them spawn correctly even before I enter the scene. That last grenade of mine proved its importance, alright.
Basically, a number of luck manipulation miracles had to come together at once. Among said manipulation involved shooting the mailbox using a charged shot. You see, the mailbox shaking advances the RNG, but shooting downwards pushes me upwards. However, a charged shot, for some unknown reason, does not affect your momentum in the least. So that is why you see the all-powerful charged shot disappear into the infinite depths of the mailbox.
After this scene, on the way to the minigame, a few downward shots lets me avoid having to jump a second time. This allows less of that slow air time than two jumps would have, meaning I get more of those speedy dashes along the ground.

1-3 Minigame

The minigame is simple enough: Drag 10 fireballs back into that shooting vehicle. Or rather, drag the same fireball into it 10 times, thanks to a glitch. Catch the fireball at the same frame it hits that vehicle. Your beam overrides the "destroy fireball" routine. How fun!
The fireball's X-velocity depends on your position when it is fired, while the initial Y-velocity depends on the RNG (manipulate lowest possible). For clear reasons, manipulate the fireball so that it is as close as possible to that vehicle.
The little agents on the ground affect the RNG. Play around with 'em if you need to adjust the RNG.
Finally, upon winning this minigame, if you drive yourself into the fireball at the same frame, you retain control over your ship, and further agents will spawn. This counts as a death. The best way to maintain entertainment, as more stuff happens, but it's also a good way to get last-minute RNG manipulation as well. I don't exactly NEED this last-minute manipulation, but I figure I may as well use it.

1-4 . Flying stage #1

This run is 162 frames faster than the leading DTC3 run for this stage. More savings than I expected.
I don't like this level. The RNG is so involved here, and is difficult to affect. T5 was the fastest run until this one was created.
The RNG advances thanks to harmless background objects vicious luck-eating clouds. As none of the enemies, and neither the player, can affect the RNG while alive, the only things left are harmless background objects vicious luck-eating clouds. For quite some time, they escaped even my script-assisted detection, but I finally realized the source of my frustrations at long last: Harmless background objects Vicious luck-eating clouds.
The RNG advances as thus: It rolls once per frame in an attempt to produce a harmless cloud when below the limit of 4 harmless clouds. When there are 4, the RNG stops advancing. These harmless clouds have absolutely no mechanical effect other than messing with the RNG. As such, they are key to luck manipulation, and often are completely invisible off the screen during said manipulation. They truly are harmless background objects vicious luck-eating clouds.
My strategy was basically search a bunch of RNGs and see which one gives me the fastest rush to the first helicopter. Helicopters try to escape the player diagonally, but by moving back and forth around one of X or Y axis of a helicopter, it will keep accelerating around zero speed for one axis, thus making the copter somewhat slower. This saves plenty of frames off the DTC3 runs.
The first helicopter gives you your weapon. A front-firing 11 damage shot. As you can't attack until then, the game provides a squad of helpful kids already smashing the helicopter in 300 frames for you, but they'll only start smashing after they're within a certain distance, so I manipulate the helicopter to be in sight as soon as I can.
The second helicopter gives you armor, setting your ship's HP to 20 instead of the initial 5 you begin with. As worthless as this may be in a TAS, the game forces you to pick it up -- It won't even spawn the next helicopter until you get the power-ups in order. So, naturally, I have no choice and must pick it up anyway.
The third helicopter gives you a rocket, speeding you up greatly. This uses a fuel meter to give you the extra burst. However, after one accelerates to maximum speed, one can alternate pressing and releasing R in order to maintain near-top speed without losing any of the fuel meter. Naturally, it is quite abused in order to avoid slowing down to basic speed. Speaking of which, using the slower engine instantly slows you down if you're at the boosted speed.
The fourth helicopter upgrades your front weapon. It now deals 13 damage per shot and has a larger hitbox. The graphic may look like 3 distinct shots, but it really is only one shot. Relatively a pathetic bonus, but it still destroys the next helicopters each in one shot less than before.
The fifth helicopter gives you back shots. A backward-firing pair of 9 damage shots. Since you shoot two at once, it's 18 damage if you hit with both. Amusingly, it looks weaker than the front shots, but the two projectiles combined deal a healthy chunk more.
The sixth and last helicopter gives you homing missiles. Spawns two missiles that home into nearby enemies, for a pathetic 2 damage per hit. The hitboxes for these things are extremely wide, in that you can fit like five of these things' graphics side by side, and you'd approximate the hitbox pretty well then. The Y-hitbox size is more sensible, but still a touch larger than intuition would tell you. Despite the insane hitbox, 2 damage isn't much compared to your other weaponry.
After you get all your powerups, clear out all enemies. Then the boss shows up. Kill boss to win. While inside the boss, the weapons combined deal 35 damage every time you shoot.

2-1 . Messin' with the KGB

I had plead with the clouds back in 1-4 to give me the perfect RNG. I didn't stop pleading until I got what I wanted. And I did stop, meaning I was quite satisfied.

2-1 Scene 1

4 kills needed.
The result of said pleads gave me Wpn6. Death to enemies came swift. Very swift, in fact. I beat the DTC3 runs by half a second at least for my expedience with this weapon.

2-1 Scene 2

Destroy a particular enemy: Tower guard
I walk a few frames before dashing. This is so that once I finish my second dash, I can release the charged shot upwards to hit the tower guy. Alas, I didn't kill him fast enough to stop the screen from locking for a moment, slowing it down from entering the next scene sooner.

2-1 Scene 3

6 kills needed.
However, even with this slow down, I still make it to this scene under the frame rule. The Spawn Timer ticks at the exact moment of the screen lock, so a well-timed kill let me avoid waiting out an extra 31 frames.
Though, I would have appreciated at least one frame of error. Oh well, make it frame-perfect or bust, then. Regardless, I do appreciate beating that frame rule here! Very much so.

2-1 Scene 4

8 kills needed.
Fairly standard spawn killing scene. Notably, I do have some time to manipulate things I need. Before leaving Scene 3, I had 13 grenades. I walk out of this scene carrying 43 grenades. Bah, still not enough.

2-1 Scene 5

4 kills needed.
This... Was a very tricky scene to handle. First, before this scene locks, enemies can spawn up to four. So in order to clear this scene fastest possible... Saw that first enemy appear when I cleared scene 4? Yes, that one had to die in time for this scene.
This scene required some acrobatics to pull through without wasting a single kill. As a result, this scene unlocks almost instantly!

2-1 Scene 6

Destroy a particular enemy: Tower guard
What luck we need to jump to get around an obstacle, and joy! We're at the perfect height to kill the guard! What fun!
A short bit of walking was done following a jump. This is to lock the next scene with one fewer dash needed. Ultimately, it's so I can shoot upwards earlier.

2-1 Scene 7

Destroy a particular enemy: Flying machine
Despite the rather short time given to charge, I abuse the auto-firing aspect of Wpn6 to give me as much time as I can for charging, yet still shooting bullets. When I finish my dash, I shoot and charge. When I jump, this gives me another moment where I can safely release B without losing charge, to then fire another volley of auto-firing death. Then the 5-damage charge is ready and I fire that.

2-1 Scene 8

6 kills needed.
While the timing of the enemy kills is pretty strict on scene entry, it's still not quite as acrobatic as Scene 5.

2-1 Scene 9

6 kills needed.
Oh, how fun. I need to jump to clear an obstacle in the scene, with the enemy grenadier to shoot an annoying projectile! The jump makes one enemy impossible to kill in a timely manner without use of a grenade.

2-1 Scene 10

Destroy a particular enemy: Brick wall
Grenades, charged shot. These scenes make me wish I can freely switch between weapons, but I'll just have to deal with handling Wpn6 all the way through.

2-1 Scene 11

6 kills needed.
More acrobatics. Which made me somewhat discouraged about having to go through this scene a second time when I realized the RNG was not going to cooperate in my first attempt. I count myself lucky that I only had to redo back from Scene 7. Scene 5 was also quite the acrobatic scene.
These acrobatics involves manipulating enemies such that they don't take up the spawn limit while certainly far out of reach off-screen. They eventually spawn good enough for me, and on I go.

2-1 Scene 12

Destroy a particular enemy: Brick wall
Yes, I still would like to switch weapons at will. The best substitute being manipulate an enemy agent along the way instead. I lose several frames by dedicating them to the RNG, but at least I get the preferred weapon with only that loss.
My first attempt had the RNG in the wrong spot by a long, long ways. It... Wasn't pretty. My only course was to go back and redo a bunch of scenes, but once again, I'm glad to avoid redoing Scene 5.

2-1 Scene 13

Destroy boss.
This boss flies. It also gets only one attack at me before blowing up. The hitbox you are to shoot at is the head. However, once it loses 100 HP (half dead), the new hitbox is its entire body, making it much easier to hit with later attacks.
While I did walk in with 30 grenades, it still wasn't enough. I ran out, and could have thrown more if I had them. In fact, ammo in general was tight -- I ran completely out of both Wpn7 and grenades. The boss was down to 1 HP when ammo ran out. The basic shot finished it off. Couldn't ask for a better timing for when ammo runs completely out -- The exact moment where one more would have, in fact, wasted time.
Then again, I was still in need of grenades. Still, I beat the DTC3 runs.

2-2 . Bombing us away!

Oh... This level. I don't like it. Because of all the forced waiting. So... Just be entertaining, of course. As tempting as it is to do only what is necessary to survive, doing nothing in particular just isn't TAS-like. Doing interesting nothings in particular can get kind of tricky at times.

2-2 Scene 1

Wait out the bombs.
According to my measurements, T2 cleared this scene 57 frames faster than this run. Honestly, I have no clue how it happened, and I even analyzed their .vbm carefully. Same can be said of T1, and I'm just as confused as to how they're slower than my run here, by 33 frames.
While the kill counter reads 17 when that van stops, the counter isn't used to unlock this scene. Instead, it's just used to determine whether that van has done its job and drive away. I still must wait out those bombs.
As I'm forced to wait, I think it's time to reveal just how lucky a TAS can be.

2-2 Scene 2

Destory a particular enemy: Flying machine
I checked to see if jumping off that stone in the way to throw grenades at it was faster. I did indeed destroy the machine the moment it spawned, but it was three frames too slow. Reason being I didn't scroll the camera quickly enough.
Instead, just jump on the stone and dash right off. Dashs are faster than walking, and so scrolls the camera faster. Thus locking this scene earlier and making that flying machine spawn earlier.

2-2 Scene 3

Wait out the bombs.
... Oh, another waiting scene. How fun.
I figure weapon drops and more is a good idea. Don't you? Of course you do!
However, I leave the right building alone. I discovered a little trick that saves a fair amount of time involving that building.

2-2 Scene 4

Destroy a particular enemy: KGB copter
The trick is the fact I jump high enough that my shots are created off-screen, and thus placed beyond the screen boundaries to reach the off screen copter.
I need a building intact long enough to make the well-timed jump. And now you know why I left that right building alone in the previous scene.

2-2 Scenes 5 to 8

Destroy a bomb
For each of these four scenes
On the way here, I can't avoid the weapon orb from that helpful kid. So I manipulate it into being Wpn7, the spread shots, keeping the same weapon as always.
Each bomb has 15 HP. For obvious reasons, I use a charged shot on the first one. I tried to make use of grenades here, but I just wasn't positioned right for a toss to provide any assistance.
Instead, I just shoot the spread shots. I duck with each shot so that the upper bullet of each spread shot actually hit the bombs.

2-2 Scenes 9 and 10

Destroy a particular enemy: KGB Van
Nothing noteworthy here.

2-2 Scene 11

Free the Yeti.
Grenades and a single shot for the first peg. Charged shot goodness and a few shots for the second peg. These things have 10 HP, no trouble getting them down.
The yeti himself is invincible. He walks slowly, but you must let him pick you up in order to unlock this scene. And you're stuck with him until you reach the cliff ahead.
There is also a glitch I do at the cliff. When you would drop out of the yeti, you can have him grab an agent at the same time. The presence of this agent gives me something to jump on, providing a lower bounce and enabling me to duck under a bullet. Not only is seeing two yellow aliens rather amusing, but it's also a great time saver.

2-2 Scene 12

Destroy boss.
This boss is in two phases...
For the first phase, the boss has a collective HP. It doesn't matter which one of the four you attack, as any damage counts toward the whole phase. Nothing special needs to be done as far as damage needed, standard TAS-speed damage is done.
However, there is an effective speed-limit in terms of how much damage you can do, thanks to a glitch. If you reduce the collective HP to zero before the fourth "boss piece" shows up, it never spawns, and destroying the other three pieces in the second phase will not end the battle. In effect, gameplay comes to a screeching halt when the player is stuck in a snowy wasteland without the ability to continue. Thus, I delay the last bit of damage until the fourth one shows up.
In the second phase, the boss comes in smaller pieces. Since the fourth one takes so long, I put priority in killing the fourth one ASAP. I destroy the other pieces in such a way to minimize the delay in destroying this fourth one. The middle-sized ones each have 15 HP, and the small ones 10.

2-3 . The ammoless gauntlet

Bah, this stage has no real amount of ammo. The only buildings you see me destroy really are the only ones that drop stuff for me. There's also a helpful kid near the end who gives a weapon, so I only have three drops to work with.
I take a pack of grenades and 2 spread shot orbs from the resources the game provides. I'm still stuck relying on basic shots on the end.
My script does find a building that I don't destroy. It's the one seen at the start of this stage. When I blew it up, what luck that the building's placement drops the item inside the impassable crates! The resulting item from this building can't be taken, due to terrain issues.

2-3 Scene 1

Destroy boss.
On the way here, I destroy two buildings. The first one needs to drop grenades -- How can I possibly destroy the second building without grenades? The manipulation worked out well, and I pick up spread shots trivially from building 2, by throwing grenades while up against the box.
As for the scene itself, if the boss lived long enough, it would jump straight back up and wait for the missiles it launched to do their job. Thankfully, I kill the boss at the speed of TAS and this annoying wait does not apply.
I tweak the RNG on the way out of this scene. This is my only free time to do so until the next weapon pick-up I get. Thankfully, due to the lack of RNG stuff needed between here and weapon orb later, this is pretty trivial.

2-3 Scene 2

Destroy a car blockade.
Nothing special. Just run and shoot.
There is one little thing I do up ahead, though. I slow down, just ever so slightly during my jump, in order to allow an enemy bullet to disappear off-screen. This is so it won't interfere as I dash forward. Due to earlier manipulation, no need to slow down getting the spread shot.

2-3 Scene 3

Destroy boss.
With only three items in the entire stage, it's certain you'll run out of ammo, being stuck finishing off this thing with basic shots. There are plenty of times here where I can manipulate the RNG, so I have pretty good control over the start of 2-4.

2-4 . The last flying level

You need a certain number of kills to spawn the boss. No need to collect your ship parts this time, you start with it all.
Most of it is making sure that you fire whenever that cooldown finishes, and always hit something with each shot. Some of it is making use of the backward shots whenever it's possible. I still end up delaying one of my shots for a short moment at one point, though.
As for the boss, it will always try to maintain a certain distance away from you. You can try to rush towards it, but it will simply back away, out of reach. I finally pull off my tactic of luring it to move towards me, then while it has momentum towards me, I rush inside it to slam both back and forward bullets into it. I wanted to do this in the DTC3, but time constraints got in the way.

3-1 . Area 51. There's also 52, 53, ...

The scorpions around here actually drop stuff at 1/4 rate, as opposed to the 1/16 from earlier agents. Good thing, as there are no buildings in sight.
I also loved how this run went, too. Every scene, I improved by at least one frame, no matter whose team you compare me to. Things just.... happened.

3-1 Scene 1

Kill 4 scorpions.
The spawn timer doesn't begin until you lock the camera into this scene.
The nice thing about these scorpions are that, despite 4 HP, they still die from a single grenade, for whatever reason. Another nice thing is that I don't have to worry about different death animations, as all scorpions will die and vanish in the same amount of time, regardless of method of kill.
Oh, and their 1/4 drop rate. That's a very nice thing I make great use out of. I needed around 10 grenade packs and a few spread shot orbs.

3-1 Scene 2

Destroy 1 vulture.
This thing has 60 HP. Due to that durability, it's largely why I need to manipulate so many grenades. Actually, there's more of these things ahead, explaining my desire for crazy ammunition.

3-1 Scene 3

Wait out the tornado.
You can either dig into the ground to avoid the traveling tornado, or die. Either way, the camera remains locked until it passes. I want my 39 grenades and 100 spread shots. Mostly why I chose to live.
Of course, this is a perfect time to manipulate a bunch more grenades. Can't be too careful.

3-1 Scene 4

Destroy 2 vultures.
Both of these also have 60 HP each. The fact I'm entering this scene with 59 grenades isn't paranoia, it's preparation. I exit with 40 grenades. I swear, it's not paranoia!

3-1 Scene 5

Wait out the tornado.
This is exactly like the first tornado. Only thing you can really do is just be in position to dash into this scene as soon as possible, and to dash out quickly as possible.
Oh, and manipulate more grenades. 78 isn't overkill, is it?

3-1 Scene 6

Destroy 3 vultures.
As with other vultures, abuse the charged shot for its properties of not interfering with your weapon cooldown, shoot using spread shots like crazy, and force-feed grenades into these things.
I carry 51 grenades out of this scene, to face the boss. Also, along the way, I pick up ice shots without wasting time.

3-1 Scene 7

Destroy boss.
Before the boss, I had to deal with a scorpion. No worries, just kill it before it stabs me for dashing into it.
Ah, the giant worm-like thing. T5 used spreads and grenades, but this wasn't enough for a single cycle kill, as we weren't aware of that CloudStacking glitch as I call it. T1 and T2 didn't manipulate grenades quite like T5, and therefore didn't have enough for a single cycle kill.
But this run... There was no such shortage of grenades. I also picked up ice shots for that glitch. I proved the single cycle kill was indeed possible, saving a few seconds. I also noted that the boss may take longer to die if multiple segments are above ground, so I actually delay a moment to minimize this wait. Seems the tail tip takes 19 less frames to finish exploding compared to the head.
For fun, I dig in where I killed the boss. There was exactly enough frames to pop back out for the victory stance without wasting time. I finished with 16 grenades. Huh, so I did pick up too many grenade packs. Seems I could spare just one out of the eleven I picked up.

3-2 . Outer base.

Aw, now we're back at the 1/16 drop rate of the usual agents. And there are still no buildings in sight. That's fine, I just glare at the RNG real hard. It even rained grenades from the sky for me!

3-2 Scenes 1 to 4

Destroy or wait out a certain enemy: Bug bot
Waiting for it to just fly off is obviously slow. Charged shots, basic gun, and grenades prove that their 10 HP each disappears quickly as always. I need to jump anyway to avoid mines, as being dead doesn't assist in dashing.
Although, I didn't set aside time to manipulate any drops yet... I might need to have grenades rain soon...

3-2 Scene 5

Destroy a particular enemy: Heavy fire agent
How fun. I'm walking into this scene with the basic gun and no grenades. Well, time to get to work, eh?
Some acrobatic weapon manipulation took place here. The one I knifed dropped a pack of grenades, which I immediately used. I shot someone out of the sky to rain grenades down onto me. I timed two kills to take place simultaneously to adjust yet another into becoming grenades. A total of four grenade packs dropped this scene, and I needed every last grenade. Oh, on the side, I pick up spread shots.

3-2 Scene 6

Destroy a particular enemy: Heavy ice agent
Unlike the previous scene, I'm walking into this one armed to the teeth in potent weaponry. He never stood a chance, just like the last guy.
I took a look at 3-3 to see what sort of RNG I need. Something frighteningly far away. I worked on the RNG pretty swiftly starting now.

3-2 Scene 7

Destroy boss.
Another revisit to a boss we've seen before. And, like before, it goes boom before it has a chance to attack.
Manipulate the RNG real hard, then be in position for the last enemy in this stage.

3-2 Scene 8

Destroy boss.
The jump is because I don't need the dashing speed, and so the lower bullets don't hit the ground before reaching the gate. It's tall enough I don't have to worry about the upper bullets.
Notably, alien victory poses are immune to bullets. No need to avoid them.

3-3 . Inner base

Sorry about taking as long as I did completing this. Over a year passed before any progress was revealed.
The RNG was in a rough spot. I did, however, manage to find something that worked out. Worked out well enough to beat the DTC3 teams, anyway.

3-3 Scenes 1 and 2

Destroy a particular enemy: Wall on wheels
I threw a grenade in order to make one kill on the way to the first scene. Said kill handed me spread shots. I seemed to have navigated the stompers swiftly enough to beat the DTC3 runs significantly.
Grenades do not appear to save any time against these walls, so it's just fine that I only had one Wpn7 drop for the first scene. The grenades I got just before scene 2 weren't used on this other wall.

3-3 Scenes 3 and 4

Destroy a particular enemy: Packaged robot
I am well aware I fell through the first box, as though it weren't there. Trust me, that thing was solid like a wall before the camera locked in place, so I couldn't keep dashing through. You get to see me land on the second box for a very brief moment, though.
Their 30 HP disappears quick. They are destroyed even before they hit the ground. The second one could die sooner, by having me toss that last grenade from a closer position, and this would unlock the camera sooner. This would save a few frames, but my attempt did not end with a favorable RNG.

3-3 Scene 5

Destroy particular enemies: Packaged robots
You'd think that dealing with two at once would slow me down a bit more. Thanks to a better position than the solo bots, they don't really last significantly longer.
I ran out of ammo briefly. That's fine, I was right next to a manipulated wpn7 drop to replenish my weapon. I had some difficulties making sure that drop existed in time for me to take it, but it is necessary.
After this scene, there are numerous psychotic walking bombs with 10 HP each. Jumps take more than a full second to complete, but four spread shots can be done across 25 frames. Thus, I opt to shoot every one of them.
... But I wouldn't have enough spread shots for all of them, so I needed to get another spread shot drop. This isn't the fastest I could do, but unfortunately, the improvements I made left me in a very bad RNG I can't get around.

3-3 Scenes 6 and 7

Destroy a particular enemy: Unpackaged robot
Ah, I can shoot them right away! I did not try to lock the scene immediately, rather I started walking and shooting when I was close enough for it to count.
Most of the trouble was due to enemy agents being in the way when trying to shoot through the psychotic walking bombs. This attempt didn't manipulate most of them out of the way, but somehow has the favorable RNG I need for 3-4.
Towards the very end, you just need to travel far enough. Replacing the last dash with a shorter lasting backflip is far enough, just a matter of good positioning.

3-4 . Inner-inner base

This is where you fight the final boss.
I had originally planned to manipulate so many drops that the CloudStacking glitch would break the hit detection so thoroughly as to deal 5 damage per frame. Turns out I somehow messed up figuring out the panic runner's drop rate. It's not 1/8, it's 1/16. Whoops. 1/16 generally isn't enough for the crazy flood of weapons that I want. So, normal methods it is.
By the way, the panic runners die from grenades. The grenade itself, not just the explosion. I used this fact to get two kills at once in order to tweak a drop. One less grenade for the boss, ah well.
I didn't kill the boss as quickly as T5, but T5 left with a much longer death animation. Part of why the slow kill was because the weapon drop I had on "standby" would disappear too quickly. It otherwise would have made my grenades very deadly for a short moment.
Scratch that. I miscounted frames. I did kill the boss faster. I believe it's from more efficient use of the CloudStacking glitch. T5 kept staying close enough that the bullet hits immediately, doing about 3 damage every time they shoot. I made sure the bullet took at least one frame to reach the boss, doing 4 damage. Still, what I have is better than my attempted improvement's RNG, where I would only have access to two decent drops at best.

Possible Improvements

To start, 3-3 has a potential 28 frames to be saved (.vbm). Unfortunately, this puts me in a very bad RNG spot. I was unable to undo just four of these RNG calls, just four necessary, in order to salvage my current 3-4 strategy. I can't be at more than 23446 RNG calls, and I haven't found a way to be below 23450 RNG calls while keeping even a few of those 28 frames.
I checked the RNG. Even two hundred more is no good. Anything out to around 23660 are all bad. And there aren't any good ways to get an extra 100 in this stage anyway. It has to be 23446 or below, or lose a few seconds to 3-4.
The game is very sync-stable, if you can line up the RNG. When Mugg found a way to save frames at the 1-2 boss, I checked it out, and have indeed saved those frames. I was working at 3-1 or 3-2 at the time, so I was hoping I don't have to redo 1-3 all the way back to where I was at. I made sure to tweak the RNG to keep what I had for 1-3, then copied the rest of my movie in place, and sure enough, nothing broke. I was able to save those 50 frames without redoing large portions of my run.
Even so, most of this run should be pretty optimal. There is a spawn timer that is used to determine when new agents show up, and there are limits depending on the scene as to how many of these agents can exist. In many cases, I kill the last agent necessary at the earliest possible moment, using a weapon with the shortest death animation. Due to the spawn timer working as a 31-frame rule, improvements are generally unlikely to take place in such scenes.
Regardless, learn about lua scripting. Then learn about the lua script I've developed with some help during the DTC3 as T5. You'll need to know about hitboxes, where the random spawning takes place, the RNG, at what points in the RNG you can expect a drop, and a few other random tidbits of information. The tools are there, and I used them extensively in this run.

Suggested screenshots

In descending order of preference:
11992 - Mouth of behemoth pudding
31676 - Something really silly here...
3319 - Heavy use of weaponry
29344 - Some might call this... Lucky.
20377 - Activity in a flying stage.

feos: Judging...


TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 15544
Location: 127.0.0.1
This topic is for the purpose of discussing #3867: FatRatKnight's GBA Alien Hominid in 15:12.85
Player (212)
Joined: 9/14/2011
Posts: 349
You've got through all this like nothing. Great run! You just won yes vote from me. By the way, I couldn't find VBA v23 svn270 on here, so I used vba-rerecording-svn232 instead and it syncs perfectly.
mklip2001
He/Him
Editor
Joined: 6/23/2009
Posts: 2227
Location: Georgia, USA
Maybe I'm missing something, but what are these "teams" like T1, T2, ..., T5? Was this submission the result of a contest I didn't know about? Regardless, I look forward to seeing this when I have my emulators. I haven't had time to follow up with as many runs lately, but then these past two weeks have seen a lot of nifty runs in the submission queue :-) EDIT: Ok, I've seen this. I think the game is a little visually confusing, since there's not a ton of outlining to separate all the sprites. However, it's really good for GBA, and the action is frantic. Good job!
Used to be a frequent submissions commenter. My new computer has had some issues running emulators, so I've been here more sporadically. Still haven't gotten around to actually TASing yet... I was going to improve Kid Dracula for GB. It seems I was beaten to it, though, with a recent awesome run by Hetfield90 and StarvinStruthers. (http://tasvideos.org/2928M.html.) Thanks to goofydylan8 for running Gargoyle's Quest 2 because I mentioned the game! (http://tasvideos.org/2001M.html) Thanks to feos and MESHUGGAH for taking up runs of Duck Tales 2 because of my old signature! Thanks also to Samsara for finishing a Treasure Master run. From the submission comments:
Shoutouts and thanks to mklip2001 for arguably being the nicest and most supportive person on the forums.
Skilled player (1737)
Joined: 9/17/2009
Posts: 4979
Location: ̶C̶a̶n̶a̶d̶a̶ "Kanatah"
mklip2001 wrote:
Maybe I'm missing something, but what are these "teams" like T1, T2, ..., T5? Was this submission the result of a contest I didn't know about? Regardless, I look forward to seeing this when I have my emulators. I haven't had time to follow up with as many runs lately, but then these past two weeks have seen a lot of nifty runs in the submission queue :-)
I assume it's from something called "Dream Team Contest" mentioned here.
Editor, Skilled player (1198)
Joined: 9/27/2008
Posts: 1085
The Dream Team Contest #3 was a competition held here at TASVideos. I probably should do a better job at explaining that in the submission text. Also, I've spotted a few errors in my frame counting. I've been checking over a few things, and, well... I seem to be pretty error-prone at the moment, more than usual. Said error made me think that I killed the final boss slower and the resulting animation was different or something. Just an error in counting frames, nothing more. Along with a few other things, I believe I shall have to fix up the submission text. I even forgot to point out a few of my favorite frames to use as the image here. I think frames 11992 or 31676 are good.
mklip2001
He/Him
Editor
Joined: 6/23/2009
Posts: 2227
Location: Georgia, USA
Ah, I'd forgotten that DTC #3 ran Alien Hominid. So, in that table, you're comparing your run against the other teams from that competition? I get it now.
Used to be a frequent submissions commenter. My new computer has had some issues running emulators, so I've been here more sporadically. Still haven't gotten around to actually TASing yet... I was going to improve Kid Dracula for GB. It seems I was beaten to it, though, with a recent awesome run by Hetfield90 and StarvinStruthers. (http://tasvideos.org/2928M.html.) Thanks to goofydylan8 for running Gargoyle's Quest 2 because I mentioned the game! (http://tasvideos.org/2001M.html) Thanks to feos and MESHUGGAH for taking up runs of Duck Tales 2 because of my old signature! Thanks also to Samsara for finishing a Treasure Master run. From the submission comments:
Shoutouts and thanks to mklip2001 for arguably being the nicest and most supportive person on the forums.
Joined: 2/3/2013
Posts: 320
Location: Germany
A (very ugly) quick encode using the originally submitted movie file This is my first time capturing, encoding and uploading a video of a TAS. So please be lenient when seeing how bad its quality is, even if it's only a temp encode. I did not quite figure out, how to work with the respective tools. Edit: This isn't needed anymore. Oh and I'd vote 'Yes', if I could. A very worthwhile run of a very worthwile game (I only played an HD remake on XBox; I think of it as a faster Metal Slug).
All syllogisms have three parts, therefore this is not a syllogism.
Editor, Experienced player (734)
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
I'm not usually a fan of Contra style games, but this was colorful and fun and action packed (and also very very bloody?) I could tell from reading your forum posts just how much work went into this run, and I think also reflects my distaste for the Dream Team Competitions, where you can clearly see that a rushed project is never going to be as satisfying in quality as a fully dedicated project without deadlines. (Mostly I feel like trying to get a TAS product done as quickly as possible ruins the spirit of TASing.) Yes vote.
Homepage ☣ Retired
Player (80)
Joined: 8/5/2007
Posts: 865
An outstanding run that unfortunately will live in the shadow of the Yoshi's Island run while it's on the workbench. Because it's FatRatKnight's work, you know it's of the highest quality. Easy yes vote. Will we see a version of this run with commentary? Edit: Just saw you uploaded a Lua commentary file (haven't looked at it yet). You wondered about how encoders will handle it. I know YouTube has features allowing the video to be paused for a few seconds at a time. As long as it can be turned off, I hope that encoders use that feature.
Editor, Skilled player (1198)
Joined: 9/27/2008
Posts: 1085
Bobo the King... You just saw my uploaded lua file. You have quite the timing, considering I only just now uploaded it. I found a few more errors in my submission text... I forgot to mark a scene as slower, though the numbers said so already. I didn't dash under a bullet in 2-2, I ducked under. I was successful in the backflip at end of 3-3. Wonder why I remembered differently... I also need to set up an IRC client on this laptop at some point... RGamma, thanks for the encode! Although, you didn't include the high score entry. Not as though it's important, but now people have a YouTube way of seeing this run now.
Player (80)
Joined: 8/5/2007
Posts: 865
FatRatKnight wrote:
Bobo the King... You just saw my uploaded lua file. You have quite the timing, considering I only just now uploaded it. I found a few more errors in my submission text... I forgot to mark a scene as slower, though the numbers said so already. I didn't dash under a bullet in 2-2, I ducked under. I was successful in the backflip at end of 3-3. Wonder why I remembered differently... I also need to set up an IRC client on this laptop at some point... RGamma, thanks for the encode! Although, you didn't include the high score entry. Not as though it's important, but now people have a YouTube way of seeing this run now.
And that's how I found your upload. I'm currently on the #TASVideos channel asking about the topology of the human body and it announced your upload. Good to hear you're back at work, and with a new laptop no less! I'll check out your subtitles in a day or two, or whenever an encode pops up.
Site Admin, Skilled player (1250)
Joined: 4/17/2010
Posts: 11473
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
Lua drawings don't go into video capture. How am I supposed to publish these subs then? They're not in normal subtitles format ether, are they?
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Editor, Skilled player (1198)
Joined: 9/27/2008
Posts: 1085
The subtitles aren't in a consistent spot. They are also bordered by different colors. I also show a box at one point. Some subtitles overlap in terms of time displayed. An internal memory address is read and output onto the screen at times. If that complicates subtitling a video, I apologize for my choice of style in my lua script. I know nothing about video making. There is a reason I never posted a work-in-progress on YouTube. What, roughly, would be something that makes the subtitles easier to deal with? I am aware that a lua script isn't something that would convert to some subtitles format as the lua script is arbitrary code, not just display. The fact you can't make a separate encode with lua display over it makes things more difficult to do. Runs fine on emulator, of course. But if the difficulty is too great, perhaps I should take the time to learn some format that makes your task easier next time.
Editor, Expert player (2328)
Joined: 5/15/2007
Posts: 3929
Location: Germany
feos wrote:
Lua drawings don't go into video capture. How am I supposed to publish these subs then? They're not in normal subtitles format ether, are they?
In text display options you can make it "in game" I think? I know that you can record Lua output into an AVI. Anyway, just here to say this is a sick job of a TAS. Well done
Site Admin, Skilled player (1250)
Joined: 4/17/2010
Posts: 11473
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
Yeah, I recalled that option, MUGG. FRK, can you make the box canvas less transparent in all the script? I think it's too transparent now, and the text is so small, so it gets mixed with the ingame objects all the time.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Editor, Skilled player (1198)
Joined: 9/27/2008
Posts: 1085
In the script, at lines 330, 331, and 332:
Language: lua

local RGf,RGb= 0x7F7F0060, 0x7F7F00C0 local RBf,RBb= 0x7F007F60, 0x7F007FC0 local GBf,GBb= 0x007F7F60, 0x007F7FC0
Just change those 60 and C0 into FF if you want fully opaque boxes:
Language: lus

local RGf,RGb= 0x7F7F00FF, 0x7F7F00FF local RBf,RBb= 0x7F007FFF, 0x7F007FFF local GBf,GBb= 0x007F7FFF, 0x007F7FFF
I can't say I know whether it'll look better if I change the opacity on my end, since it looks perfectly fine to me. So I would ask you to change it yourself, and hopefully it's a simple matter with my directions. You can also tune the colors darker if necessary, by replacing the 7F to some smaller value. There does exist text in there that does not have any background border, in particular the boss HP indicator. That's slightly more complicated to handle, and probably would be best if I make the changes myself, unless you want to take the time to figure out how to add the function call yourself.
Site Admin, Skilled player (1250)
Joined: 4/17/2010
Posts: 11473
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
Changed to CC. As for the HP meter, it's for a short time and isn't a text to read normally, so nevermind.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Post subject: Movie published
TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 15544
Location: 127.0.0.1
This movie has been published. The posts before this message apply to the submission, and posts after this message apply to the published movie. ---- [2322] GBA Alien Hominid by FatRatKnight in 15:12.85
Skilled player (1651)
Joined: 11/15/2004
Posts: 2202
Location: Killjoy
This movie was everything amazing I was expecting it to be.
Sage advice from a friend of Jim: So put your tinfoil hat back in the closet, open your eyes to the truth, and realize that the government is in fact causing austismal cancer with it's 9/11 fluoride vaccinations of your water supply.