Jazz Jackrabbit in 26:41.92 by Ilari, Noxxa & slamo

Goals

  • Emulator used: jpc-rr r11.8
  • Game version: v1.0
  • Clears all main episodes (episodes 1-6)
  • Takes damage to save time
  • Plays on hardest difficulty (Turbo)
  • Aims for in-game time.

About goal choices

Version choice

There are four versions of this game, 1.0, 1.1, 1.2 and 1.3 (plus maybe some prototypes).
  • 1.1 doesn't appear to be easy to get good version of. We haven't been able to find one.
  • 1.2 has an extended version of Tubelectric 2, some airboard ditching events are missing and Marbelara 3 (episode 5 secret level) is impossible to complete.
  • 1.3 has totally broken airboard ditching events and a physics change which makes Orbitus 2 almost impossible to clear.
Now, the problems with airboard ditching events are the show-stoppers. In many levels, missing the ditch event is a massive time loss (especially in Orbitus 1) because airboard climb/descent is so slow.
The Orbitus 2 being almost impossible wouldn't be that bad, because the strategies one would pull off in a TAS still work. However, there seems be other changes to bounce beam physics too, which if true, would cause large time losses.
Marbelara 3 being broken is of no consequence, as it won't be visited anyway.
For the ditch trigger problems alone, 1.0 was chosen. It also happens to be the version Vortale used in his RTA.

Aiming for in-game time

There are basically three possible times to aim for:
  • Fastest real-time.
  • Maximum time at level end.
  • Least in-game timer.
Why we chose least in-game timer:
  • Fairness to configuration:
    • PC hardware is not fixed. Varying hardware speed varies loading times and lag.
    • Much easier to judge runs involving dissimilar CPUs if this can be ignored.
    • If we had as fast emulated PC as the RTA Vortale did had, this run would be about 2 minutes shorter in real-time.
    • The time left is decremented from timer interrupt, which means it does count even lag frames.
    • But in-game timer only counts non-lag frames.
  • Perverse incentives:
    • Shooting the exit sign stops the time. Thus there would be incentive to shoot it from as far as possible, leading to very long scrolling sequences.
  • Unreasonable difficulty:
    • Optimizing lag in this game is hard and relatively few lag frames can be eliminated. One would get greatly increased effort for very little gain.
Unfortunately, this timer is not available in the status panel, so one has to memory watch it. It has the following properties:
  • Does not count lag frames
  • Increments by 1 every in-game frame.
  • Resets to 0 (but may increment before a frame is rendered) every level entry.
  • Is the timer that controls frame rules.
  • Starts counting a few frames before screen begins to fade in and stops a few frames before screen completes fading out.
The quoted lengths of levels are the final values of this timer (with times calculated using the nominal frame rate of 125875/2108 fps).

Difficulty

Turbo difficulty adds +1HP to every enemy (except bosses, which get +10HP) over hard. The +1HP means even the lowest-tier mooks take 2 hits to kill. Time is set to 4 minutes, and Jazz dies if the timer runs out.
Hard difficulty adds some more enemies, has -1HP for Jazz and +10HP for bosses over medium. Time is set to 6 minutes, and Jazz dies if the timer runs out.
Medium difficulty adds a lot more enemies, has -1HP for Jazz and +20HP for bosses over easy. Time is set to 8 minutes on medium, although nothing happens if the timer runs out.
The +1HP from Turbo difficulty makes TASing the game a lot more interesting.

General notes

  • Normally if you clip a corner or run up a hill, you slow down, so this is to be avoided. Many places that don't seem very uneven still count (e.g. ends of bridges) and conversely many places that look really uneven aren't actually.
  • The way Jazz moves is on frame rule and Jazz can move multiple pixels in one frame. This causes the fastest way to do turns to be highly variable. Sometimes hugging walls is the fastest. Sometimes using arcing path is fastest. Sometimes even running into wall ahead is fastest.
  • Holding down while on air causes Jazz to accelerate down faster.

Comparison to previous TAS (episode 1 by Ilari) and RTA (30:54 by Vortale):

The RTA length is calculated from timestamps of first visible fadein frame, first visible score countdown frame plus fudge factor of 2.8 seconds.
The fudge factor includes:
  • 0.05 seconds for fade-in to become visible.
  • 2.75 seconds for score countdown cancel + screen fade out
    • The game actually stops calculating the level midway in fadeout.
LevelLengthOld LengthΔlengthRTAΔRTA
Diamondus 100:23.9700:24.0300:00.0724.530.45
Diamondus 200:33.8800:35.8700:01.9839.665.61
Tubelectric 100:52.5800:53.5000:00.9256.123.28
Tubelectric 200:32.9700:33.7800:00.8236.933.80
Medivo 101:01.4201:02.9000:01.4870.248.53
Medivo 200:37.3200:37.8700:00.5541.443.94
Guardian 100:12.0000:12.2800:00.2814.742.68
Episode 1 total04:14.1304:20.2300:06.10283.6628.31
Letni 100:24.88N/AN/A33.648.64
Letni 200:40.00N/AN/A52.0811.89
Technoir 100:35.42N/AN/A42.296.70
Technoir 200:30.80N/AN/A51.6820.73
Orbitus 100:38.97N/AN/A48.279.12
Orbitus 200:39.63N/AN/A50.1110.29
Guardian 200:09.68N/AN/A15.585.85
Episode 2 total03:39.38N/AN/A293.6573.21
Fanolint 100:31.67N/AN/A36.764.94
Fanolint 200:25.18N/AN/A33.247.94
Scraparap 100:18.13N/AN/A22.213.99
Scraparap 200:25.67N/AN/A34.939.14
Megairbase 100:32.25N/AN/A45.6113.20
Megairbase 200:59.78N/AN/A78.7918.72
Guardian 300:08.85N/AN/A12.693.80
Episode 3 total03:21.53N/AN/A264.2361.73
Turtemple 100:23.12N/AN/A38.5115.28
Turtemple 200:36.03N/AN/A43.136.92
Nippius 100:19.63N/AN/A27.017.28
Nippius 200:27.83N/AN/A46.9719.00
Jungrock 100:28.18N/AN/A30.962.64
Jungrock 200:43.02N/AN/A48.124.90
Guardian 400:08.38N/AN/A9.811.39
Episode 4 total03:06.20N/AN/A244.5157.42
Marbelara 100:33.82N/AN/A43.779.79
Marbelara 200:58.90N/AN/A75.1816.00
Sluggion 100:22.17N/AN/A31.469.19
Sluggion 200:36.92N/AN/A52.5015.41
Dreempipes 100:48.23N/AN/A67.3818.91
Dreempipes 201:22.73N/AN/A106.0622.93
Guardian 500:10.55N/AN/A13.723.12
Episode 5 total04:53.32N/AN/A390.0795.34
Pezrock 100:52.97N/AN/A73.5220.30
Pezrock 200:41.75N/AN/A50.748.79
Crysilis 100:27.38N/AN/A38.8411.33
Crysilis 200:35.65N/AN/A55.1019.28
Battleships 100:31.55N/AN/A42.0510.35
Battleships 200:37.00N/AN/A34.29-2.89[1]
Guardian 600:10.78N/AN/A19.828.98
Episode 6 total03:57.03N/AN/A314.3676.19
Grand total23:11.60N/AN/A1790.48392.19
[1] This level is slower due to having to wait the bird to shoot 10 extra shots in order to kill the level boss (none of your own weapons can hurt it unless you go inside, which would be much slower).

Stage-by-stage comments

Episode 1 ("Turtle Terror")

Diamondus 1

Length: 1438 frames (24.08 seconds).
[00:17] In this drop, it is possible to get so far to the right that Jazz will miss the slope and run into a turtle in the bottom, so one must slow down a bit in order to jump off.
[00:31] Jumping over this pillar in one go is not possible, so one must jump on top of it and then drop down to the exit.
[00:35] About 140 in-game frames (bit more in real-time as the game lags a lot when in the score countdown) after the score countdown begins, one can cancel the rest of the countdown by hitting fire.

Diamondus 2

Length: 2033 frames (34.04 seconds).
[00:36] The loading times are a lot longer than in the RTA run. This loses about 2-3 seconds per level.
[00:44] There are two paths out. The upper path via spring (that the last run took) and lower path. Turns out lower path is faster.
[00:49] For some strange reason, shooting that turtle that makes a brief appearance makes the bee move a bit, making hitting it inevitable unless one slows down.
[00:54] When in fast feet, running up slopes doesn't slow you down.
[01:03] Collecting that one disc and hitting the spring optimally didn't seem to be possible so the disk was skipped.
[01:05] The shield becomes useful here: There is an off-screen enemy on the way. One can't shoot it because bullets instantly disappear off-screen, but the enemy can still damage Jazz.
[01:08] This is the first place where holding down to accelerate faster actually saves time. Yes, one can use it earlier and yes, it can locally save a bit of time, later there will invariably be some place where frame rules and precise positioning will just kill the savings.
[01:10] When you shoot the exit sign, the camera will scroll the sign to certain position on the screen. This scroll is slow, so the fastest way is to shoot the exit sign when it is as close to that position as possible.

Tubelectric 1

Length: 3155 frames (52.83 seconds).
[01:16] The beginning is just weird. Some unknown factor seems to be causing subtle alterations to physics, which can randomly gain or lose time. Worse yet, there's no way to manipulate it that is known.
[01:20] Jumping into where the pipe turns from horizontal to vertical usually saves a frame.
[01:21] Hitting down can be used to cancel ejection from pipe. Doesn't save any time here though.
[01:27] This turn from right to down is just weird. The drop is very slow due to a mapping error (this section is coded to push right instead of down).
[01:29] These "downward" pipes aren't really pipes but just free spaces. Thus, using down to accelerate faster is useful here.
[01:54] Obtaining a launcher orb from here when running requires certain positions. Otherwise Jazz won't pass through a frame where shooting would hit one.
[02:00] This pipe is seriously unpredictable. The frame rules and positions here can easily mean a dozen frames gained or lost.
[02:06] One can shoot the exit sign below with a launcher. Saves 2 frames by allowing hitting it in a better place (less scrolling).

Tubelectric 2

Length: 1978 frames (33.12 seconds).
[02:13] This pipe is weird. If you enter it too far right, it will warp you to the left by 10 pixels and where you enter it affects your position on the top of the pipe.
[02:16] There would be a secret passage in the wall here, but entering it from the pipe is impossible.
[02:41] Hitting Down+Jump while passing a bridge during a spring jump cancels the spring effect, converting it into normal jump. Faster to do that here than use full height jump.
Version 1.2 would have a second checkpoint replace the exit sign and the level continues, taking some 15-20 further seconds to complete.

Medivo 1

Length: 3685 frames (61.71 seconds).
[02:59] One can't avoid this sword without slowdown, and it is by far the biggest wait, so best to use the shield here.
[03:05] Making these jumps on top of swinging balls is very hard. The problem is that one pair is just about as far from each other as possible when one makes there, making jumping from one to another just barely possible (the low ceiling restricts the distance one can jump).
[03:11] This drop is seriously annoying. The gap on the bottom is very small and thus difficult to hit.
[03:19] Collision detection with ceilings on airboard is bit buggy, allowing Jazz to get about halfway into the ceiling.
[03:20] The climb on an airboard is slow, so one wants to get as many corner-boosts as possible.
[03:23] Corner-boosting here and then shooting the sword requires certain positioning / frame rule. In general, taking damage while on airboard is a bad idea, because it freezes Jazz in place for a while.
[03:25] This sword is positioned just so that it can't be shot without slowdown, forcing a loss of 2 frames.
[03:33] Originally this run was to be done using version 1.3. However, it was discovered that 1.3 had bug that caused airboard ditch events not to work. Because it would have been utter disaster in some later levels (and sizable time loss even here), the run was redone on 1.0.
[03:44] Running on a swinging ball while it is swinging in direction of your motion pushes you ahead.

Medivo 2

Length: 2239 frames (37.49 seconds).
[04:02] Getting that fallen star means spending most of the level in invincible state.
[04:06] The other passages here are dead ends.
[04:29] The fallen star would run out just about on top of that sword, forcing one to slow down a bit to shoot it.

Guardian 1 ("Devan ship boss")

Length: 720 frames (12.05 seconds).
[04:40] Starting with a blaster saves a bit of time, because it is so slow that the boss will move to it even if shot before getting firing line with missiles.
[04:41] It is actually possible to kill this boss before it switches phase to jumping and punching out blocks.
[04:45] Boss levels have the screen scroll the boss into a certain position, so one should have the boss as close to that position as possible.

Episode 2 ("Ballistic Bunny")

Letni 1

Length: 1493 frames (25.00 seconds).
[05:01] These magnet beams are weird. Even if vertical, in addition to pulling Jazz up, they also allow Jazz to run faster than normal 6px/f, even without fast feet. When exiting the field, the speed instantly drops.
[05:07] Faster to take damage here than to shoot the barrier. The corridor is so narrow that Jazz recovers from being stunned almost instantly (and with 4 speed from the damage knock-back).
[05:10] The fastest way here is to use the chip enemy to knock-back Jazz into the barrier.
[05:12] This room is a bit prone to cause the game to crash. Fortunately, no crashes were seen when actually attempting to go for speed.
[05:17] It is faster to slow down a bit here in order to land on the small horizontal segment, as this causes Jazz to latch onto floor, and he then can run down slopes at 12px/frame, instead of falling at 9px/frame.
[05:19] Now, this floor is quite unpredictable. Jazz may or may not cling to it, and even complete stop doesn't guarantee clinging to it. If he doesn't cling, when running at full speed he will end up hitting the ceiling. The only observed pattern is that if position x+y is 3919, then he clings, otherwise not. That is the largest possible x+y there.
[05:20] 1:2 slopes up with fast feet are a bit buggy, with Jazz bobbing up and down rapidly.

Letni 2

Length: 2400 frames (40.19 seconds).
[05:29] Those ceiling chips are quite difficult to kill because those are difficult to hit (e.g. Toaster would need to be almost at the top of the jump). Launchers work acceptably.
[05:35] Going down here is faster than right (despite RTA going right to grab a bird). The difference is over a half second.
[05:39] Faster to take damage here, as stun recovery is almost instantaneous and there isn't enough time to shoot all the stuff in there otherwise (especially due to that ceiling chip).
[05:53] This breakable wall shortcut was previously undocumented and saves about 2 seconds.
[05:54] Best to use the shield here, as it is the biggest slowdown to destroy this. While a glitch attack would likely have worked, that bug was discovered far later when run was greatly past that point (and it would have saved very little at most).
[06:01] One needs to take a bit of running jump here in order to make it to the top due to the jump height being proportional to the running speed.

Technoir 1

Length: 2125 frames (35.58 seconds).
[06:14] The items in this level are buggy. The event type is "give extra time" instead of correct "item", which causes the time remaining to go crazy.
[06:16] Using the bird to shoot the exit sign appears to be impossible because the drop continues too far down and the exit sign is too far right. One can't TNT it either because Letni does not have TNT packs. This level does, but one can't return to here after getting one.
[06:23] These counter-moving conveyor belts cause Jazz to be unable to accelerate, but still hold speed. Therefore is is fastest to jump a bit so Jazz is not in contact with one and can get more speed.
[06:27] Faster to take damage here, because damage knock-back gives instant 4 speed and stun ends almost instantly.
[06:34] This belt moves with Jazz, letting him to move faster than 6px/f (still resets immediately upon losing contact with the conveyor).
[06:39] This moving spring is almost too far to the left.

Technoir 2

Length: 1848 frames (30.94 seconds).
[06:54] The mines in this level take 6 hits to kill instead of 4 like in the previous level.
[07:07] It is just barely possible to jump off this spring without waiting for almost a complete cycle (would be nearly 10 seconds).
[07:07] The bird is manipulated to shoot the shield monitor. While the shield itself will be useless, shooting the monitor for some reason causes some enemies in the section ahead to move a bit, allowing passing through without slowing down or taking damage.
[07:12] Specifically, if Jazz didn't get the shield, these mines would be bit more to the right, making passing through without damage (and stun would lose a large amount of time there) or waiting impossible.
[07:15] This turtle is intentionally hit with shields because it is faster to take damage a bit ahead.
[07:16] Specifically, here taking damage is faster (due to knock-back setting speed to 4). The bird is lost, but it is almost useless anyway and another can be obtained later.

Orbitus 1

Length: 2338 frames (39.15 seconds)
[07:32] Accelerating off the vertical bounce beam is faster because of instant maximum speed, even if one starts moving later.
[07:36] Faster to pass this wall than go over the top, even if the wall throws you upward (just a few frames).
[07:38] One wants to bounce against walls as much as possible because each bounce gives extra 10px of height.
[07:44] Getting Jazz knocked back on airboard into a vertical bounce beam causes Jazz to shoot off diagonally at (6,-13) px/f, a massive increase from the normal vertical speed. The length of the shooting is difficult to control.
[07:45] Even if Jazz is sent far into the wrong direction, the horizontal motion is fast enough for this trick to save time.
[07:52] One can cut some bounces short by intentionally hitting platforms. The RTA (apparently accidentally, judging from the runner's reaction) also pulls this off.
[07:57] Those snakes have 22HP(!) on turbo, and as such are almost immortal for TASing purposes.
[07:59] Faster to take damage here and then use mercy invincibility than to dodge some of the snakes below.
[08:03] There is supposed to be a fourth mine to the left, but for some reason it failed to spawn.
[08:03] Passing through those horizontal bounce beams is unpredictable. If the horizontal speed is high enough (barely anything), they will bounce Jazz. But even if falling through one at 0 horizontal speed, Jazz still can either bounce or pass through.

Orbitus 2

Length: 2378 frames (39.82 seconds).
[08:11] Slowing down a bit here allows contact with the bounce beam for a bit longer, giving some extra pixels forward.
[08:16] Shooting the mine in that pit is quite difficult due to the mine above having to be destroyed first and it also takes more than 2 hits to destroy.
[08:20] This would be the infamous place that is "impossible" in some CD version (actually 1.3). We didn't notice a difference between 1.0 and 1.2.
[08:21] One can either bounce from the vertical beam or spring up from horizontal beam below, very carefully aiming for the hole in wall. Bouncing from the vertical beam is a bit faster.
[08:26] These platforms are on a level timer, and as such tend to be in wrong places. Faster to take damage here and run through first two pits without caring about platforms.
[08:28] Faster to take the fallen star here as the platforms tend to be in wrong places anyway, avoiding damage knock-backs and allowing taking the bird to boss-fight.
[08:31] Getting that bird and bringing it to the boss fight saves ~10 frames due to it giving fire support and causing some damage.
[08:32] This "run up the walls" requires very careful positioning or it won't work.

Guardian 2 ("Ostrich boss")

Length: 581 frames (9.72 seconds).
[08:54] At 60 base health (100 health in Turbo), this boss is surprisingly tough for still being relatively early in the game - it's the second toughest boss in the game, behind the final boss in episode 6. In addition, it is quite speedy. Of course, it still goes down in just a few seconds, but during those few seconds still some time had to be spent to dodge the boss.
[08:55] Detonating the two TNT packs that were unavoidably in the way in Technoir 2 and didn't find any use in Orbitus saved 4 frames. Yes, detonating TNT packs only does 2HP damage to a boss.

Episode 3 ("Rabbits Revenge")

Fanolint 1

Length: 1900 frames (31.81 seconds).
[09:14] One can't jump here due to the fans then blowing Jazz up.
[09:16] Even through getting the fire shield is ~35 frame loss, due to horrible enemy positioning in this and the next level, it is worth it.
[09:20] The flying flower enemy is exclusive to Hard and Turbo difficulties, and as such it doesn't appear in the RTA. The biggest PITA in this world.
[09:25] This first flower costs about 3 frames to dodge. It is dodged, because there are more beneficial places to use the shield later on.
[09:26] In contrast, this flower is about 20 frames to dodge. One of the biggest time-wasters in this world, and thus killed with a fire shield.
[09:31] If one does this turn as fast as possible, one gets unavoidably hit by something, in this case by a cannon shot. Now, getting shot in shield is quite buggy: Jazz gets mercy invincibility that isn't shown and the fire shield disappears for the duration.
[09:32] These water flows work like conveyor belts in Technoir.

Fanolint 2

Length: 1511 frames (25.30 seconds).
[09:50] For some reason, it is in some cases momentarily possible to touch spikes without taking damage. This is likely because the spike tile's collision map takes up the entire block, and the game registers the jump before Jazz and the damage event intersect.
[10:05] Jazz is just about to hit that flying flower, but fan physics blow him away just before that.
[10:08] That flower is just in a bad place and would be one of the biggest time-wasters without using a fire shield hit on it.
[10:10] And another one which would lose a good bit of time without shields. Good thing one gets a new fire shield near the start of the next level.

Scraparap 1

Length: 1088 frames (18.22 seconds).
[10:25] Magnet physics are annoying (based on same "technology" as Letni magnet beams, just much stronger). Those spikes are hit because avoiding them would have lost much more time.
[10:28] This level doesn't have springs, it has bouncing caps. Unfortunately that means some waiting for them to fire (which occurs a set number of frames after they spawn).
[10:33] This is a good place to use the second fire shield hit because dodging would have been much slower.

Scraparap 2

Length: 1540 frames (25.79 seconds).
[10:54] This place is good for the third hit: Those rods down are solid, which makes jumping over it hit them, stopping Jazz.
[10:57] Fortunately fast feet upgrades the airboard vertical speed to 5px/f.
[10:58] Diagonal missiles are practically designed for corner-boosting. There is a flying turret in the way, good idea to use the last hit on it as avoiding it would have been a big time loss.
[11:01] One can't corner-boost off 3rd of these platforms and then avoid the 4th.
[11:02] There's a magnet beam here. It's useful to be in it because it allows going up at 6px/f.
[11:03] This is the only level in the game that is supposed to end with airboard.

Megairbase 1

Length: 1935 frames (32.40 seconds).
[11:12] One needs a tricky shot in order to kill one missile below. Otherwise it would be an 8 frame time loss to dodge (and much more to hit).
[11:18] This spring is on a level timer. In this case, in quite a bad place, losing some time.
[11:19] This stalker ghost is in a place where avoiding it without luring it somewhere else is impossible, so it's pretty much a given that it should be killed with a fire shield hit.
[11:22] This second stalker ghost would just have been a few frames to avoid. However, it was determined that the entire rest of the episode only requires taking 2 hits, so one might as well sacrifice the hit.
[11:37] In contrast to the previous flying spring being in a bad place, this spring and the next one are in very good positions (both are on level timer; this first one even suddenly spawns on-screen).

Megairbase 2

Length: 3587 frames (60.07 seconds).
[12:15] Now things just get nasty: One has to choose how many boots one collects. Each boot increases jump height, and higher jumps allow for more strategies.
To make things extra nasty, the major differences only come at the end of the level. There are three basic strategies there:
  • Run across room and then go up and left. Needs 6 shoes (actually, supposed to need 11 due to earlier section, but due to fast feet lasting just long enough, 6 suffices).
  • At start of room, jump up. Needs 14 shoes (saves running across the room first).
  • At start of room, jump up and right. Needs 16 shoes (also saves time from being able to move to right when going up).
Now, considering the extra time to obtain more shoes, it was determined that picking 16 is the fastest overall.
[12:19] Now jumping up this shaft is supposed to require all 22 shoes, but because fast feet (Jazz got here so fast that there are about two seconds of it left) increases jump height a bit and together with ramming walls to get further boosts up, 16 is just enough to make it.
[12:31] Now this zig-zag down is just a pain. The turtles are positioned exactly so that if you just run down, you WILL get hit. Fortunately, after like two of them, one can get the screen to scroll down to mess up their spawns, so they can be shot.
[12:34] That stalker ghost is just in a nasty place. Fortunately, careful launcher shots can kill it before Jazz reaches there. There are three strats to the part near the end of the level, requiring different numbers of jump shoes.

Guardian 3 ("Devan tank boss")

Length: 531 frames (8.89 seconds)
[12:51] It is possible to defeat that boss in 114 frames, but that was done on top of an earlier test-run. In the final run when getting there, the physics subtly changed (and couldn't be manipulated) forcing a loss of one frame (so 115 frames).
Bosses are also invincible to glitch attacks and one doesn't have to dodge attacks, so no use to bring a shield here.

Episode 4 ("Gene Machine")

Turtemple 1

Length: 1387 frames (23.22 seconds).
[13:17] Fast feet makes the first half of the level fly by very quickly. The pyramid with the zig-zag section is skipped by doing a running jump over the tallest pillar.
[13:25] The boulders have strange mechanics and are basically like rolling platforms. They can be ridden to gain/lose speed and approaching them from the bottom causes Jazz to be warped on top of them.
[13:26] In the section with the boulders rolling on the cacti, it's faster to just take damage rather than riding them safely. The spike-jumping trick doesn't work here because the cactus tile doesn't take up the entire block, forcing Jazz to intersect with the overlapping damage event.
[13:30] Amazingly, the fast feet bonus is kept until the end of the level.

Turtemple 2

Length: 2162 frames (36.20 seconds).
[13:43] This level starts off with some straightforward sections, although the 45 degree angle hills are deceptively hard to optimize.
[13:53] The airboard section has a lot of opportunities for corner-boosting. Jazz can even corner-boost on spikes without taking damage if done carefully. The boulders also help by warping Jazz to the top of them.
[14:06] Getting the fast feet bonus here uses a small trick to not lose any time. With a specific button sequence on specific frames, it's possible to turn around and shoot without losing speed.

Nippius 1

Length: 1178 frames (19.72 seconds).
[14:19] One of the shorter levels in the game. On the second turn, it's possible to barely reach the snowman platform without hitting the floor below, but it requires getting Jazz jammed into the corner and losing all his speed.
[14:24] There is a bird here that's not too far out of the way but is skipped. It's not particularly helpful yet and there's a better positioned one in Nippius 2 anyway.
[14:30] In the spring section it's possible to use the first set of springs to reach the necessary platforms.

Nippius 2

Length: 1670 frames (27.96 seconds).
[14:42] Scrolling the screen down is used here to make sure a launcher hits the speed bonus without disappearing. The bonus can be hit a bit earlier, but the fast feet actually runs out before the level ends, and the extra speed is more meaningful later in the level than it is on that turn.
[14:50] This jump to cliff requires some very careful positioning. It is very close to pixel-perfect.
[14:51] The ascending spring section has problems with off-screen events. Events that aren't visible have no effect, so since Jazz rises so fast, he moves off the screen and completely passes where the top spring should be. The best way to deal with this is to hit one of the ice blocks and wait a bit for the camera to catch up. This is also a good opportunity to pick up the bird without any time loss.
[14:53] Another off-screen problem immediately arises with enemies. Jazz appears on the camera perfectly positioned in front of a penguin, and since bullets disappear off-screen, it's impossible to kill while going too fast. One actually has to slow down to manage this situation.
[14:54] Launcher strategies once again come in handy before zig-zagging down, killing a turtle that would otherwise be in the way. There's a fire shield here that's out of the way, and it was determined to not be worth getting because there aren't any significant threats for the rest of the episode.
It would actually be possible to shoot above the platform while staying below it, but due to one enemy there, it would either hit that enemy or Jazz would hit a wall, causing a large time loss (more than the fire shield would save).
[14:59] The moving platform section offers some small speed boosts by running along co-moving blocks. The section after can be skipped almost entirely by doing a long jump over to a thin ice platform.

Jungrock 1

Length: 1691 frames (28.31 seconds).
[15:16] Long hills followed by a jump-turn. These are just plain nasty to optimize.
[15:20] One can just barely hit the ceiling right of the hole, instead having to hit the ceiling to the left of it, or worse, getting Jazz squeezed into that hole.
[15:24] Bit of an odd route but seems to be the fastest (faster than dropping down to bottom level immediately).

Jungrock 2

Length: 2581 frames (43.22 seconds).
[15:56] This and the next jump are just nasty with positioning and frame rules, essentially creating a frame rule that will erase all previous gains unless you beat it. Also, one of those flying things would be absolutely wonderfully positioned if not for the fallen star.
[16:20] The third platform has the spring to exit cave. The other platforms would connect to the shaft on the right (which would clearly be a longer path).

Guardian 4 ("Lizard boss")

Length: 503 frames (8.42 seconds).
[16:32] One needs to stand in a certain position. More to the right and you get hit (you still will get hit eventually, but you can kill the boss before). More to the left and one doesn't get a clean firing line.
[16:37] The position also seems to have the perfect camera position: There is no scroll at all. Furthermore, it seems that running around causes the score countdown appear later, even if you return to that position.

Episode 5 ("The Chase Is On")

Marbelara 1

Length: 2029 frames (33.97 seconds).
[16:54] It doesn't appear to be possible to directly jump on the left trunk cut.
[16:56] Collecting some RF missiles. This episode has somewhat of a shortage of those.
[16:59] Fortunately those flying dragons take much less hits than would be expected (only 6 on turbo).
[17:04] It does not appear to be possible to jump off the gaps blasted into pillars. Thus one must fall down to bottom and jump up.
[17:05] Easter egg: Shooting the statue of female rabbit subtly changes it.

Marbelara 2

Length: 3534 frames (59.18 seconds).
[17:35] One can take a shortcut through the room with the bonus gem (just don't collect it). There also would have been a fire shield nearby, but it was determined that collecting it is not worth it.
[17:56] This spring jump is just too high, and there is nothing to cancel it with.
[17:57] Destroying that secret wall allows pushing Jazz bit further to the right, allowing earlier start of acceleration to left without hitting the wall.

Sluggion 1

Length: 1330 frames (22.27 seconds)
[18:35] Faster to run on this block than jump over it (it pushes Jazz a bit).
[18:37] The shortest route is through a breakable wall here. The RTA uses a longer route to pick up invincibility.
[18:38] Going right here instead falling through shaft on the left is a bit faster.

Sluggion 2

Length: 2215 frames (37.09 seconds).
[19:07] This breakable floor is a previously undiscovered shortcut, skipping a small section.
[19:16] Those flying worms take 10 hits to kill. Here's killing one for demonstration.
[19:17] That section above has items, but is seemingly unreachable (solid walls on all sides everywhere, no warps on the level, etc).
[19:18] That slug is in an absolutely bad position, so taking damage is by far the fastest way through it.
[19:27] There are two paths through this maze, one clearly faster than the other.

Dreempipes 1

Length: 2894 frames (48.47 seconds).
[19:38] Even though those turtles appear to be asleep, touching one still hurts.
[19:40] One turtle is used for a damage-boost here.
[19:43] The water is annoying because swimming is much slower than running. As a consequence, jumping a bit to stay above the fuel a bit longer saves time.
[19:47] These big mines take 7 hits to kill.
[19:51] The items here have a similar bug as in Technoir 1: They give time when collected, messing up the time remaining.
[19:58] Apparently swimming has similar hit detection to the airboard. As a consequence, the same ceiling tricks work and in this case enable squeezing through the gap. The free spaces only touch diagonally between one pixel from each side.
[19:59] That rod is solid, as a consequence one can't shoot the switch from far away, nor do launcher antics work.
[20:02] Don't hit that switch. Hitting it would lower the fuel, making progress impossible.
[20:06] Damn waiting for fuel level to rise...
[20:09] One could get into the space to the left by using similar trick as mentioned above, but Jazz would then get stuck until the time runs out.
[20:21] Lose the shield in preparation to damage boosting.
[20:22] That mine can be damage boosted off of, skipping the topmost switch and waiting for the fuel to rise.

Dreempipes 2

Length: 4964 frames (83.13 seconds).
[20:33] Same sort of time bug here as in Technoir 1 and Dreempipes 1.
[20:36] One can't avoid dropping into fuel here, resulting in the speed getting capped to 2. That stud is too far to reach.
[20:38] We want to leave the leftmost small mine and not take damage here from the big one in preparation for damage boosting.
[20:39] Damage boosting off a mine into the pipe saves 160 frames here by skipping a switch.
[20:44] However, we didn't raise the fuel level previously. This saves 130 more frames by staying above water, for nearly a 5-second time-saver.
[20:45] The third-from top controller has the "change water level" property. It is also unavoidable. Doesn't change the fuel level by much.
[20:49] There is a counter-current in bottom half of that pipe (which would cause speed to drop to 1), so it's better to stay in the upper half.
[21:12] By squeezing against the ceiling here, one can bypass the big mine without killing it.
[21:25] The middle controller here is another of those "change water level" events. However, the TAS is so fast that it will reach buttons that change fuel level before the change completes.
[21:25] The pipe is a dead end, but contains fast feet, which save a good chunk of time, since those upgrade the swimming speed to 4px/f and increase the running speed (there will be a running segment near the end).
[21:33] More squeezing against the ceiling to avoid mines.
[21:40] This place is supposed to be dry, but the fuel level is still dropping, not having reached here yet.
[21:42] Looking carefully, one can indeed see that the fuel level is dropping here. That's the effect of that one strange item collected about 17 seconds earlier.
[21:42] We only need to hit the second from top switch and then wait for the fuel to rise back up a bit, then we can jump to the ledge.
[21:44] Finally just running toward the exit instead of slow swimming.
[21:50] Fast feet actually expires mid-air between the pipe and exit sign.

Guardian 5 ("Owl boss")

Length: 633 (10.60s).
[22:00] After the approach salvo, we do a short jump, fire into the owl, then at last moment turn around, causing damage knock-back to go right, landing Jazz into the owl's nest (the reason for short jump is that it allows faster recovery). Then we just bombard the owl with missiles.
[22:01] We turn moments before the boss dies. This is because otherwise the spawning shots no longer hit the boss.
[22:05] The game wants to scroll the camera towards the left of the nest. One can't stand underneath it without being on the floor below, and Jazz does not jump high enough. As a consequence, one must do some sort of weird moving jump to minimize scrolling.

Episode 6 ("Final Clash")

Pezrock 1

Length: 3178 frames (53.22 seconds)
[22:20] Can't get first the two orbs without clipping a corner, causing a slowdown.
[22:23] Those spikeballs could in theory be destroyed, but not in practice due to the ridiculous HP those things have (115 BaseHP + 1HP (turbo) = 116HP(!!!) ).
[22:26] One of the snakes must be destroyed or you will get hit by something, and that requires some careful launcher shots.
[22:38] One can't jump over that spiked ball.
[22:42] Diagonal clip through blocks. The collision masks are diagonally connected between one pixel from both sides, similar to the clip in Dreempipes 1.
[22:49] Those springs only move when you face away ("stalker walker" AI).
[22:52] Not quite the intended route, but faster (RTA uses this too).
[22:53] Those spikes are so low that even the shortest jump will hit them.
[22:54] Those platforms are on a level timer.
[22:56] Faster to take damage here than wait for the platform.
[22:58] Jumping from spikes somehow. The invincibility should have already ended.
[23:01] These two spikeballs are positioned so that one would either have to take damage or wait. We take damage.

Pezrock 2

Length: 2505 frames (41.95 seconds)
[23:22] The left side has a spiked ball in a very nasty position.
[23:47] Going via lower route, because the upward one has lots of spiked balls, many in bad positions.
[23:48] These platforms are on a level timer, and so far off phase that it is pointless to wait instead of taking damage.
[23:49] However, this is in a good enough position to be faster to ride it than take damage.

Crysilis 1

Length: 1643 frames (27.51 seconds).
[24:02] Dropping down is a faster route here than going right.
[24:03] Jumping on spikes. I guess Jazz goes inside the event box just for one frame.
[24:05] Careful here, those spikes are so low even the lowest jump can hit them.
[24:09] Those spheres have ridiculous HP (100 baseHP + 1HP (turbo) = 101HP).
[24:15] The platform pushes Jazz into wall. One has to be careful not to get stuck in wall.
[24:17] Taking damage on previous mines lets one use mercy invincibility to pass these mines (shooting would be almost impossible).

Crysilis 2

Length: 2139 frames (35.82 seconds)
[24:35] This episode has a severe shortage of RF missile ammo. This level is the only one where missile ammo is gathered, and only from smaller +2 orbs.
[24:42] Some damage boosting for a faster turn. It's the most productive use of a hit. Also saves having to wait to blast the following mines.
[24:52] Blowing up flying turtle in bad place by using some launcher antics.
[24:54] Those spikes have no solid box and are so low that jumping over that turtle is impossible.
[24:59] The end section is such a gauntlet that grabbing this fallen star is well worth it.
[25:01] This spring is in a nearly optimal position (it is on a level timer).
[25:01] More nicely positioned platforms (unknown if these are on level timer or not).

Battleships 1

Length: 1893 frames (31.70 seconds)
[25:17] The path branches here. The right one is substantially faster.
[25:17] The spikeballs are such that you can't make it under the second one to the right without taking damage.
[25:18] Some frames faster to take damage here.
[25:22] Yes, those barrels can be blown up.
[25:25] Those pipes here move Jazz at 12px/f, unlike the pipes in Tubelectric/Industrius, which move Jazz at 10 px/f.
[25:28] These vertical pipes in effect act as 8-frame rules. One can lose up to 7 frames here (this run loses 5).
[25:30] If one gets damage-knocked into the pipe, Jazz flies at 13px/f speed, but that is not useful here.
[25:31] Those spikeballs are pretty durable, 11HP (10+1HP). The last spikeball is phased such that one can't get around without slowing down.
[25:32] Shoot backwards to save the bird, followed by some tricky launcher shots to blow up a barrel (one and two hits is easy, three required to blow it is not).
[25:33] Holding backwards has Jazz try to slow down while pushed forward. Useless graphical thing for fun.
[25:35] Because missiles spread, shooting from this far away would not be effective, so start softening the boss with toasters (which shoot straight ahead).
[25:35] Release forward so that Jazz will slow down and stop just before the boss, allowing blasting it at point blank range.
[25:36] One of birdshots hits the boss, saving a pair of missiles (2 frames).

Battleships 2

Length: 2217 frames (37.17 seconds)
[25:57] Intentionally slow down 8 pixels to get better positioning (saves 2 frames later).
[25:58] Using the bird to destroy the boss and skip the entire level. Altough it takes a while, it is much faster than going through the ship (which is much bigger than the first one).
There is a version difference here. v1.2 has bit different physics here, allowing bird to hit 3 times per 64 frames. In v1.0 that appears to be impossible, only allowing 2 hits per 64 frames. This trick would be several seconds faster on v1.2.
The boss also has more HP than the one in RTA, which causes ~2.85s time loss.
Also, getting the bird to start hitting as early as possible (A=692) isn't as easy as it looks (for instance, without those 8 pixels lost, the earliest is A=694).
[26:17] Finally, after 00:18.15 seconds of bombardment, the boss finally starts to explode.
[26:21] It is not possible to look up or down with the airboard. Thus one can't get right camera position, but must instead just get the nearest one.

Guardian 6 ("Rabbit boss")

Length: 674 frames (10.84 seconds).

Screen-shot suggestion

VGA Frame 25922 / Timecode 428.19

Thanks

Nach
Vortale (routes, RTA times)

Nach: Nice run, nice feedback, accepting as improvement to existing single episode run.
Nach: Publishing.


TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 15629
Location: 127.0.0.1
This topic is for the purpose of discussing #4360: Ilari, Noxxa & slamo's DOS Jazz Jackrabbit in 26:41.93
Spikestuff
They/Them
Editor, Publisher, Expert player (2656)
Joined: 10/12/2011
Posts: 6449
Location: The land down under.
ILARI LIVES! I KNEW YOU DIDN'T DIE!... I'll vote in a bit. Quick, insane and complete. Yes Vote.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account. Something better for yourself and also others.
Site Admin, Skilled player (1255)
Joined: 4/17/2010
Posts: 11495
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
Spikestuff wrote:
ILARI LIVES!
I could tell. http://tasvideos.org/userfiles/user/Ilari
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.
Emulator Coder
Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
feos wrote:
Spikestuff wrote:
ILARI LIVES!
I could tell. http://tasvideos.org/userfiles/user/Ilari
Those are hidden, you can tell because of your senior rank.
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Editor, Skilled player (1441)
Joined: 3/31/2010
Posts: 2113
Both this and the extra episodes were a very exciting watch. Nice fast and fluid movement. Also, the trick in Battleship 2 was really cool. Voting yes on both.
Experienced player (691)
Joined: 11/23/2013
Posts: 2239
Location: Guatemala
Yes for the Star!
Here, my YouTube channel: http://www.youtube.com/user/dekutony
Warepire
He/Him
Editor
Joined: 3/2/2010
Posts: 2178
Location: A little to the left of nowhere (Sweden)
As expected this was extremely entertaining. Voting yes, and recommending that this shall obsolete the "Episode 1" TAS currently on site.
Editor
Joined: 3/10/2010
Posts: 899
Location: Sweden
I am just wondering how this can be slower than the RTA for one level. Could you please clarify how this is?
Noxxa
They/Them
Moderator, Expert player (4128)
Joined: 8/14/2009
Posts: 4090
Location: The Netherlands
henke37 wrote:
I am just wondering how this can be slower than the RTA for one level. Could you please clarify how this is?
Our run plays on the hardest difficulty ("Turbo"), whereas the RTA plays on Medium. On Turbo, bosses have quite a bit more health (20 more hit points to be exact), and the monitors in the Battleship stages count as bosses. Normally this hit point difference is negligible because we can blast most things to bits in mere seconds, but the monitor in Battleships 2 is killed slowly by a bird helper. It has to fire 10 more shots compared to the RTA to destroy the monitor, which ends up costing us a few seconds in that stage.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Skilled player (1417)
Joined: 10/27/2004
Posts: 1978
Location: Making an escape
Isn't there a cheat you can enter to make the game even harder?
A hundred years from now, they will gaze upon my work and marvel at my skills but never know my name. And that will be good enough for me.
Warepire
He/Him
Editor
Joined: 3/2/2010
Posts: 2178
Location: A little to the left of nowhere (Sweden)
There is, DOOM gives you even harder enemies, I had forgotten about that. Reference: http://www.gamefaqs.com/pc/565038-jazz-jackrabbit/cheats
Patashu
He/Him
Joined: 10/2/2005
Posts: 4045
Glitches Self-destructing enemies If you enter the code ''DOOM'' 20 times in a row (code used to make enemies/items harder) then instead of all the enemies becoming impossibly hard, they destroy themselves! Whenever you see an enemy, they self-destruct! Also, whenever you see a weapon pick-up, or checkpoint sign, it will automatically act like you've already shot it!
It looks like 'DOOM' makes the enemies arbitrarily stronger every time you enter it (until overflow). So entering it any particular number of times is quite arbitrary, except for 0.
My Chiptune music, made in Famitracker: http://soundcloud.com/patashu My twitch. I stream mostly shmups & rhythm games http://twitch.tv/patashu My youtube, again shmups and rhythm games and misc stuff: http://youtube.com/user/patashu
Editor, Expert player (2480)
Joined: 4/8/2005
Posts: 1573
Location: Gone for a year, just for varietyyyyyyyyy!!
Silence! The KTHULHU is now activating the triangular portal! Activate Jazz Jackrabbit cheatcode: DOOM Sequence 1/3 completed. Bu-ha-ha-haa... Activate DOOM cheatcode: IDSPISPOPD Sequence 2/3 completed. Bu-ha-ha-haa... Activating Amazin' SPISPOPD cheatcode... What the...? No. No way... THIS CANNOT BE! Everything is ruined... If only Amazin' SPISPOPD had had the cheatcode "JAZZJACKRABBIT" as planned (instead of #$@&%ng "JAMULDOOM"), then the triangular circle would have been completed, the portal would have opened between the meta-dimensions of intervideogamespace... And we, The Old Ones, would have entered your realm... Danm these clumsyLovecraftian typewritters of hel, we almsot suceedead! Who is responsible for this #$@&%ng "JAMULDOOM"??? HOW MANY TIMES HAVE I SAID: KTHULHU NEEDS BIGGER KEYBOARD, SO THAT THE TENTACLES WON'T STICK ONTO MULTIPLE KEYS SIMULATENOUSLY! BUT NO, ALL YOU WANT IS BIGGER LEATHER WINGS AND HORNS AND TENTACLES! HOW DO YOU THINK WE CAN HACK THE WORLDZ WITH #$@&%ng TENTACLES???
Joined: 5/2/2010
Posts: 41
Location: the Netherlands
Definite Yes vote. Love this game, and love the TAS. Thanks for this!
Editor, Skilled player (1824)
Joined: 8/25/2013
Posts: 1200
interesting game, yes vote from me.
effort on the first draft means less effort on any draft thereafter - some loser
Player (26)
Joined: 8/29/2011
Posts: 1206
Location: Amsterdam
Excellent work, and for one of the PC classics that deserves a full run.
Joined: 9/27/2011
Posts: 207
Location: Finland
Now hack the game to show a greater screen area so we can have an Atlas video (isn't that what they're called?) To emphasize this run was a blast!
Patashu
He/Him
Joined: 10/2/2005
Posts: 4045
Svimmer wrote:
Now hack the game to show a greater screen area so we can have an Atlas video (isn't that what they're called?) To emphasize this run was a blast!
Atlas videos aren't created by hacking the game, they're created by taking the video footage and super-imposing it on a created map of the level it's going through, moving it around, etc. I hear they're a lot of hard work ;)
My Chiptune music, made in Famitracker: http://soundcloud.com/patashu My twitch. I stream mostly shmups & rhythm games http://twitch.tv/patashu My youtube, again shmups and rhythm games and misc stuff: http://youtube.com/user/patashu
Joined: 9/27/2011
Posts: 207
Location: Finland
^^ Yeah I knew that. However, if the game would draw any more of the screen with the moving enemies, that's what I was looking for. Not to say the hack wouldn't probably be a lot of work itself.
Patashu
He/Him
Joined: 10/2/2005
Posts: 4045
Svimmer wrote:
^^ Yeah I knew that. However, if the game would draw any more of the screen with the moving enemies, that's what I was looking for. Not to say the hack wouldn't probably be a lot of work itself.
If making the screen larger caused more enemies to be loaded, pretty sure the TAS would desync right away.
My Chiptune music, made in Famitracker: http://soundcloud.com/patashu My twitch. I stream mostly shmups & rhythm games http://twitch.tv/patashu My youtube, again shmups and rhythm games and misc stuff: http://youtube.com/user/patashu
Joined: 9/27/2011
Posts: 207
Location: Finland
Hmm... because of the extra lag? Making the screen larger would only make the screen larger though. I guess it depends on how the game handles it but possibly it wouldn't affect monster loading at all? Anyway I'm speaking very hypothetically here. Just throwing something out there... that this might look pretty cool as a larger scale video.
Noxxa
They/Them
Moderator, Expert player (4128)
Joined: 8/14/2009
Posts: 4090
Location: The Netherlands
Enemies and objects (referred to as "events" in map editors of this game) only load, appear, do anything, or just outright exist at all when they're close enough to the screen. (Most games work this way, to save on unnecessary processing time for faraway objects). Thus, if one would play with a larger view screen, enemies would appear on screen sooner, thus start acting sooner, and their behavior won't match. The only way they'd stay in sync is if the boundaries for activating remain the same, but in that case, you'd see them randomly pop into the screen as Jazz gets close enough to them. Incidentally, the way events spawn when close enough to the screen is integral in the TNT skip glitch in the extra episodes run and one of the Holiday Hare stages. It's also why in Megairbase some flying springs can appear out of nowhere: their position is dislocated from their event (spawn) position, so by the time Jazz is close enough for the event to be loaded, the spring itself is already quite close to Jazz. Also, rendering the extra tiles and/or enemies would probably screw up timing (lag), so even if the enemy behavior would sync, the movie still likely wouldn't. So far, atlas videos have been created by using premade maps from e.g. VGMaps, and syncing the map coordinates to the video every frame. That way, the game or run themselves don't require any hacks, although it's a lot of manual work to create a video this way.
http://www.youtube.com/Noxxa <dwangoAC> This is a TAS (...). Not suitable for all audiences. May cause undesirable side-effects. May contain emulator abuse. Emulator may be abusive. This product contains glitches known to the state of California to cause egg defects. <Masterjun> I'm just a guy arranging bits in a sequence which could potentially amuse other people looking at these bits <adelikat> In Oregon Trail, I sacrificed my own family to save time. In Star trek, I killed helpless comrades in escape pods to save time. Here, I kill my allies to save time. I think I need help.
Joined: 9/27/2011
Posts: 207
Location: Finland
Okay, then it's probably a utopian idea. Was there never an automated/semi-automated solution (i.e. why does it require manual positioning?) for combining the TAS video and the game map e.g. by reading the coordinates of the camera view from memory each frame and feeding this into some kind of video editing program? Sorry this is fairly off-topic for this discussion page but on the other hand I don't suspect this will add anything new to the discussion. Just me being curious.
Emulator Coder, Skilled player (1114)
Joined: 5/1/2010
Posts: 1217
https://archive.org/details/jazzjackrabbit-tas-ilarimothrayasslamo - Primary encode (with embedded subtitles). - 512kb encode. - WebM (mark II) encode. - External subtitles file. Regading atlas encode. Reading camera coordinates from memory is no problem (the existing memory watch scripts do display it, as it is useful for avoiding post-boss scrolling). However, the following things seem pretty hopeless (and would be needed for automated atlas encode): - Reading current level number from memory. - Reading which events are currently active and where those are (looks like coordinates are not on pixel granularity, unlike Jazz coordinates).
Post subject: Movie published
TASVideoAgent
They/Them
Moderator
Joined: 8/3/2004
Posts: 15629
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. ---- [2642] DOS Jazz Jackrabbit by Ilari, Noxxa & slamo in 26:41.93