TASVideos

Tool-assisted game movies
When human skills are just not enough

Submission #2670: Comicalflop's SNES Donkey Kong Country 2 - Diddy's Kong Quest in 40:44.73

Console: Super NES
Game name: Donkey Kong Country 2 - Diddy's Kong Quest
Game version: USA v1.0
ROM filename: Donkey Kong Country 2 - Diddy's Kong Quest (U) (V1.0).smc
Branch:
Emulator: (unknown)
Movie length: 40:44.73
FrameCount: 146684
Re-record count: 176984
Author's real name: Joshua Nyer
Author's nickname: Comicalflop
Submitter: Comicalflop
Submitted at: 2010-05-15 03:05:03
Text last edited at: 2010-05-17 22:13:22
Text last edited by: sgrunt
Download: Download (13004 bytes)
Status: published
Click to view the actual publication
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:

Donkey Kong Country 2 - Diddy's Kong Quest Any% TAS

This is an improvement to Arne_the_great's Any% run of Donkey Kong Country 2. I've always wanted to do the any% category, and when his run came out in 2007, I enjoyed his run alot, but spotted a few minor little time savers here and there, but kept them out of mind for a while. After completing the 102% with NxCy, we checked and many of our strategies would be time savers for the any% as well. So I began working on the any%, making improvements here and there. Then as more and more tricks were being discovered, a significant amount of improvement began to present itself. I had to restart this run from scratch a couple of times, most notably when I had previously gotten to Castle Crush and then had to go back all the way to Barrel Bayou and redo because of 2 new glitches.

Overall, I saved 3,857 frames or 1 minute, 4.28 seconds. The new length is 146684 frames, 40:45, and the in-game timer of the game is now 36 minutes.

I was very impressed with the final time, especially considering how optimal Arne's movie already was. Eight of the stages don't even have more than 10 frames saved, and 29 stages aren't even 1 second faster. It was very challenging to save time in some places.

  • Emulator used Snes9x 1.43+ improvement9 (Snes9x 1.43+ v12 Beta also works)
  • Movie length: 146684 frames (40:45)

  • Use WIP1 Timing (Yes)
  • Allow Left+Right/Up+Down (Yes)
    • Volume Envelope Height Reading (No)
    • Fake Mute desync workaround (No)
    • Sync samples with sound CPU (No)

  • Aims for fastest time
  • Manipulates luck
  • Uses Warps
  • Takes damage to save time
  • Abuses programming errors in the game

There is a huge amount of new stuff in this movie. If you want to be surprised, read no further!!!

New Tricks and Glitches

Many of the new tricks are ones discovered and utilized in the 102% movie. I'm only going to describe new tricks that are present in this movie.

Acceleration

I don't know why everyone doesn't do this. It's a basic premise of acceleration. Anyhoos, at absolutely every opportunity when turning around, I test not only how far to hold one direction and then the opposite, but also holding no direction in between the left/right or right/left button presses, For example:

Going to the right, then curving around to the left:

> > > > > > > > _ < < < < < < <

This shifts the subpixel positions of your character, and allows you to uber-optimize *every* turn.

Everyone should get in the habit of doing this.

The Jump Delay Property

This is not a proper term, but is used to describe the effect of the delay that is caused when jumping in this game. The physics are different from Donkey Kong Country; the frame after you jump, there is still a 2 frame moment where you are touching the ground. This can be used for a few subtle effects.

You can even be rolling in the air, jump for two frames and then team up. This is also used throughout the run.

Team-Up

When you press A, the other Kong mounts onto your back. With the delay property, you can press B for 2 frames, then press A, and you will be "on the ground enough" to trigger the teamup. However, you are still in the air. When you jump for two frames, then team mount, you get a little boost that saves time, and allows for higher jumps than a normal teamup jump would allow. There's a speed boost because you're initial Jump speed is repeated; so you go from -2000ish speed, to -1900ish speed, then team up, then go back to -2000, -1900. (The numbers are not exact, but you get the idea.) This means 2 extra frames where you're jumping at normal max speed, and allows for faster team up jumps, and higher jumps allowing access to places normally not accessible.

Teamup Dismount

If you utilize the jump delay property, you can jump for two frames, A to team up, jump for 2 frames, then press A again to dismount, then hold B. You get a mini-super jump, which can be used to get some added height, as well as grab ropes that you normally can't get to. A useful technique.

Teamup Jump Cancel

On some goals, rather than landing on the ground and jumping again, it is faster to do a goal switch. However, if you still need the Kong, there is the team-up jump cancel. This abuses the jump delay property. By holding jump for 2 frames, then pressing A, but not jumping again when the other kong gets on your back, you are technically in the air and will fall. You have enough height to trigger the goal normally. This is used occasionally, such as in Bramble Scramble and Gusty Glade.

Door Entry Delay

This was implemented in the 102%, but fully utilized here. Going into doorways puts your speed at 384, and not holding left or right cancels this. So rolling into a doorway, without holding right, then pressing right at the exact moment will save time. This is used in any doorway to ensure as little 384 walking as possible. However, I have made a little addition to this trick, done specifically in the warps in world 1 and 2. I roll sooner than normal, but by using the jump delay, I can jump, wait a frame, then start holding right, so that I can be in a rolling state for a longer period of time. You can hear this in effect in Rattle Battle; I use the trick to cancel Rattly's jump, so you can hear what is happening in world 1/2 warps.

Enguarde

I found a small timesaver, where rather than cancel his charge and start lunging with B in order to start moving up or down, I press the opposite direction, then hold the direction I want to go in while enguarde's normal charge speed decreases slightly.

Squawks

I found that when moving towards a barrel horizontally, shooting an egg backwards extends Squawks' hitbox enough to save one frame. Used in Screech's Sprint.

The big jump via damage

There's a big jump possible in this game, and it works by taking damage. You get a normal type of boost if you jump and get hit by an enemy, but you can extend this jump to enormous heights if first you team throw a Kong into a blast barrel. This enables super jumps, which is used in Barrel Bayou to skip one of the big pits.

Big jump and float

By combining the big jump properties with the already discovered trick to float as Diddy (enter water, switch Kongs very quickly, jump out of water) You can get the super jump, but also float at the same time. This enables Diddy to reach insane heights, as can be seen in Slime Climb.

Invisible Barrel wackiness

The Golden Rambi glitch has been taken one weird step further. Apparently, at a special spot in Castle Crush, if you land on a porcupine that is a certain distance into the level, and the porcupine gets crushed by the ceiling, the goal target will appear in front of you. Very bizarre, and it requires very strict parameters to work properly. It also can only be performed at the spot where it's done in this TAS.

Corner Boost Underwater

In underwater levels, if you swim against a 90 degree corner going up, you'll get boosted up a couple of pixels, saving one frame. Very hard to perform.

There are various other tricks that are used, but these are the major ones.

Level by Level

Pirate Panic: Smoothed things up overall. New door delay combined with jump in the warp, and using the jump delay, I was able to save 1 frame when rolling onto the goal target.
25 frames faster

Mainbrace Mayhem: Faster via warp and difference in goal target switching.
7 frames faster

Gangplank Galley: I changed entering the warp barrel by team jumping before I jump on the barrels, which saves time since on such short hops, Dixie's lack of friction slows things down.
23 frames faster

Lockjaw's Locker: Only faster via the warp.
6 frames faster

Topsail Trouble: Only faster via the warp.
8 frames faster

Krow's Nest: I got very, very lucky here. This is a little tough to describe. I got perfect luck to do what I wanted. Arne waited 7-8 frames before entering to get his perfect setup, but I was able to wait 0 frames to get a setup where the 3rd egg dropped on first bounce, but its bounce was far away from where I wanted it. However, I calculated that team-throwing up there would save some time; turns out it did. 8 saved from entering earlier, and the rest from the teeny amount of time saved from getting to jump on top of egg #3. (Plus 1 frame saved on the first hit of Krow.)
12 frames faster

Hothead Hop: The first 1+ second saved level. The Team-throw delay properties allowed me to jump up onto the rock, and get up to the barrel, all while teaming up only once and jumping only twice. Sexy.
90 frames faster

Kannon's Klaim: Team-throw combo up into the barrel saved time, as did low rolling into the warp.
28 frames faster

Lava Lagoon: Jump-delay team-throw, and better warp.
12 frames faster

Redhot Ride: Various improvements. Better Helicoptering, and also used the camera trick to delay the spawning of the Krusha so that I don't have to wait but can jump around him. Also, when you team up on a balloon, a weird glitch occurs where it takes twice as long for the Kong to get on your back, so I use the jump delay property to mount in mid air-ish and jump the balloon.
35 frames faster

Sqauwk's Shaft: Slightly faster. 'Nuff said.
4 frames faster

Kleaver's Kiln: Same strategies as the 102%.
34 frames faster

Barrel Bayou: I made many changes here. First, jumping into the barrel blast you along and saves a few frames. Next, I found a few route changes with Rambi, especially those barrels. (I thought that landing earlier = more Rambi charge speed= faster. I was right!) Next up is setting up the super jump glitch. To do this, I team throw into the blast barrel. I tried doing this into the barrel that I first jumped into at the beginning of the level, but since the fastest way to handle this segment is similar to how I'd team throw into this barrel anyways, it's faster to do it here. Once the glitch is "stored" all I need to do is hit the Kloak, and hover over the large gap that you normally can't cross. I tested bringing Rambi past the sign, but unfortunately it wastes 14 seconds. There was supposedly a glitch that showed a player doing this and accessing Rambi in a future level where he's not in it, but it's cheat-codes so no dice. The damage here is going to slightly alter Glimmer's Galleon.
268 frames faster

Glimmer's Galleon: I start out with Dixie this time around. Dixie performs exactly the same as Diddy underwater (except apparently corner boosts are harder to do.) Having to rescue Diddy and doing a goal switch sets me back 46 frames, however I saved 26 frames over Arne's route (Most notably by not making the Puffup blowup at the end) which brings this stage to 20 frames slower. An acceptable trade-off considering Barrel Bayou.
20 frames slower

Krockhead Klamber: Low rolling proved to be very beneficial here. I just took the level more aggressively.
23 frames faster

Rattle Battle: I changed how I handled this, especially with the route. It's not beneficial to slow down and waste time just to slide off a ledge. I used the A triple jump technique from the 102% to save a little bit of time here, but I mostly just handled things differently and had less slowdowns.
25 frames faster

Slime Climb: Now this is bizarre. It's creepy watching Diddy go "BOING!" and then float up like a ghost. Anyhoo, I initially found a team throwing method, and then the teamup/dismount trick to grab some ropes, but the new floaty glitch beat it by like 3 seconds. If you're wondering about the first DK barrel I skipped, I tested getting that one and it's slower. I also improved the ending, by setting up the team throw jumps earlier, but also aiming to get that lower ledge with the DK coin, and doing a normal jump onto the goal target.
321 frames faster

Bramble Blast: Its almost all the same except for the through the roof glitch. Again, just like the 102%, I don't take damage, jump on the right side of the kroc for an earlier setup, and do a goal switch. That way, I end up saving an additional 29 frames with Kudgel.
97 frames faster

Kudgel's Kontest I hexed the 102% fight; I liked that fight so much, it made a comeback, and I didn't want to redo my least favorite fight. Faster because of not needing to pop a DK barrel.
29 frames faster

Hornet's Hole: Tiny little improvements here and there, hardly noticeable.
15 frames faster

Target Terror: Nothing much new.
0 frames faster

Bramble Scramble: Tackled this level way more aggressively. I tried using the route of taking damage as Squawks to avoid becoming Squitter (via the DK coin route) but it was like barely 12 frames faster, and that's without the DK barrel I'd need to pop (which is -29 frames.) 78 frames faster

Rickety Rage: A lot of hard work to get a very minuscule improvement. Ah well. Subpixel optimization abound here.
6 frames faster

Mudhole Marsh: Inserted even more more 'scroll camera left' tricks. This was another level that I handled very aggressively. It's important to note that the first 1/3rd of the level was TASed by NxCy. Thanks buddy!
39 frames faster

Rambi Rumble: Followed Arne's route, but I decided to go rogue and not use Rambi at all. Mwuhahahaha!!!!!! There is a tiny gap above that bee (which I might add is sooooo close to being to normally roll and jump through, but alas...) and because we don't spend 10 seconds transforming into Rambi, it comes out to barely more than a second faster. Props to NxCy for testing this idea out. His route was fun, but wasn't faster until I decided to let Diddy take damage to avoid having to team throw across that last pit. We're going to be at a boss fight where we can pick up a DK barrel with no slowdown, so why not I thought? And it worked. Talk about completing a level not as intended.
86 frames faster

King Zing Sting: Nothing new here. The frame count is going to lie; It says 1 frame lost here, but I saved 6 frames in Ghostly Grove before I was hexing. After completing this level that desynched, Ghostly Grove was showing 7 frames faster. So weird.
(Actually) 0 frames faster

Ghostly Grove: Very tiny differences. I think there's only two; camera tricking the porcupine, and jumping in between the 2 krocs. It says this is 7 frames faster, but it's really 6.
(Actually) 6 frames faster

Haunted Hall: No, I could not replicate the disappearing ghost glitch from Arne's run; same as 102%. I'd rather go with a big frame improvement than showing a glitch someone already demonstrated. Used the tricks found in the 102% to save mucho time.
160 frames faster

Gusty Glade: Various improvements. At the end, a low roll underneath the barrels let me only need to use one barrel, rather than two. I also used the jump delay teamup cancel to save time as well.
57 frames faster

Parrot Chute Panic: Only real thing different is the one trick used in the 102% that was used as a faster replacement to Arne's way of avoiding damage. Interesting, NxCy found one possible route change (a way to skip one more parrot) but the extra lag it created caused it to not save or waste any time. Bizarre.
34 frames faster

Web Woods: First off, utilizing the camera trick, I get to roll for a long amount of time (thanks NxCy!). As Squitter, I really tried watching my speed and not using mroe webs than necessary. I only use them for entertainment when I can launch a web while I use LR to make a platform, which does not slow me down at all (since my speed was already dropped at 0.) I think a few spots I managed to save time by not using a web in places where Arne did. One weird thing is when I did my old 102% movie, Arne scorned me for not taking damage twice. Turns out taking damage is, yes, 4 frames faster for reaching the next platform... but it loses 29 frames in Kreepy Krow. So not taking damage here ended up saving 25 frames.
I am quite sad about what is coming up next, aka the worst case of bad luck ever. When optimally ending Web Woods, I got HORRID, and I mean HORRID luck for entering Kreepy Krow. I had to purposefully delay the fadeout of Web Woods, and delay the entrance to Kreepy Krow before I got anything close to a good run for that stage. More on that later.
35 frames faster

Kreepy Krow: Uch. Hate this level. Before I had to go back and hex, I managed a near perfect (nearly 0 delay entrance) run of this stage, with perfect luck. After all the hexing, I was completing Web Woods and meeting atrocious luck with Kreepy Krow. Needless to say, I had to settle for Middle, Right, Middle barrel setup. However, I at least was able to maintain the time saved from not taking damage in Web Woods. I could save more time with better luck. Ah well. Enjoy the antics of my boss fighting anyways.
29 frames faster

Artic Abyss: Jumping gives you better acceleration on ice. Enguarde was used better. But I honestly don't know how I saved so much time. Shrug.
38 frames faster

Windy Well: This was another level made faster by the various things I found in the 102%, as well as playing more aggressively.
53 frames faster

Castle Crush: Big time saved here. If you thought Golden Rambi was weird, you ain't seen nothing yet. It was discovered by a japanese player that you could use the invisible barrel to pull the goal target down to you. This gives me more chances for entertainment like in the 102% video. I'm kind of thankful this saved so much time; I was having a very, very, very, very hard time replicating Arne's Golden Rambi route (not the execution; I pulled that off 5 frames faster). Of course, this means the start of Clapper's Cavern needs to be changed, just like Glimmer's Galleon.
1,472 frames faster

Clapper's Cavern: After suffering a 128 frame loss from ending Castle Crush as Dixie, I managed to save 50 frames over Arne, thus bringing my frame loss to 78. Various little improvements here and there. 78 frames slower

Chain Link Chamber: Utilized Arne's route, since I couldn't find any major way to save time. But from better precision and less lag, I saved 18 frames. Not bad.
18 frames faster

Toxic Tower: Rattly is a ninja. He saved even more time here. I found a few little time shavers vs. the 102% run, and smoothed up the Squawks and Squitter sections.
677 frames faster

Screech's Sprint: Followed Arne's route, but made some differences when handling the camera tricking to avoid slowdown (max horizontal vine speed is more than max running speed). Screech's skip was easier than I thought it would be. There was only one chance to use the vine zip thing, but I also smoothed out Squawks' movements. Also, I changed the ending; 10 frames were saved over precision, but 30 frames are saved from being able to pop as Squawks and than instantly land on the target, but without triggering the song. I figure Diddy can beat Dixie in K. Rool with a 30 frame headstart.

I was right.
40 frames faster

K. Rool Duel: Diddy is just as capable as Dixie in this fight. There's only one moment, when you need to carry the cannonball over to K. Rool's cannon ASAP and drop it in there, btu Diddy is almost nearly as fast in that regard. Besides that, there's essentially no differences to using Diddy instead of Dixie. So the 30 frame headstart, couple with ending input even earlier than Arne does, leads to 65 frames saved.

About the earlier ending; I know getting hit seems weak, but Castlevania does it, and since this is the any% I think any and all means to get the fastest time should be used. I myself am always amused by endings where the main character gets hit, yet still gets to the credits, especially by the extra thought taken to devise fastest strategies for ending input as early as possible. For example, NxCy was only able to beat one of my attempts at ending earlier by 1 frame.
65 frames faster

Author’s Comments

I've always been waiting to destroy the any% category, and it feels so relieving to finally have it done. I can get back to Trauma Center now that this big puppy is out of the way.
NxCy was once again an immensely fun partner to work with, even if he didn't end up TASing many segments as he did with the 102%.

Many great thanks to:

  • Arne_the_great, for his any% TAS and for all the tricks he discovered.
  • The 3 japanese players whose vids led to the big 3 discoveries of tricks in Barrel Bayou, Slime Climb, and Castle Crush (I forget their names, but the tricks were found and/or brought to my attention by them.)
  • All the patient viewers

Enjoy!


cpadolf: Impressive improvement and very good viewer response. Accepting for publication.


FractalFusion: Set to delayed by author's request.


Baxter: Setting it back to accepted.

Similar submissions (by title and categories where applicable):