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

Submission #6529: DrD2k9 & BrunoVisnadi's Coleco Jumpman Junior in 04:37.19

Console: ColecoVision
Game name: Jumpman Junior
Game version: USA
ROM filename: Jumpman Junior (1984) (Epyx) [!].col
Emulator: BizHawk 2.3.2
Movie length: 04:37.19
FrameCount: 16610
Re-record count: (unknown)
Author's real name:
Author's nickname: DrD2k9 & BrunoVisnadi
Submitter: DrD2k9
Submitted at: 2019-10-08 15:09:48
Intended tier: Moons
Text last edited at: 2019-10-14 01:47:15
Text last edited by: Masterjun
Download: Download (6376 bytes)
Status: decision: accepted
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:

Jumpman Junior

Some of this is duplicated from BrunoVinadi's previous submission notes. Some is new.

Story (from game manual)

On The far side of Jupiter lies the Jupiter Command Substation, a valuable outpost of planet Earth. The ALIENATORS, dreaded enemies from beyond our solar system, have already attacked Jupiter Command Station and are now invading the Substation! Can it be saved?

Jumpman, Jupiter's super secret agent, is desperately trying to restore communications at the Command Station. As the promising apprentice to Jumpman, you, JUMPMAN JUNIOR, have received an emergency request to drive back the ALIENATORS. You must defuse the bombs planted throughout Ihe Substation's 12 levels before they explode, crumbling the outpost instantly. Your speed, Jumping ability and courage are your only weapons against the enemy.

GOOD LUCK! Jumpman — and the entire planet — are counting on you.

Temp Encode


The previous submission was done using BizHawk 1.11.4 which utilized a different input format for Coleco games than BizHawk 2.3.2. Due to this, inputs from that submission had to be manually converted to the new format to even be read properly and copied into modern BizHawk. When appropriate, these converted inputs were used for some stages of this submission.

Seven of the 12 stages utilize different routing/strategies. Six of these are significant improvement; one unfortunately resulted in lost frames due to RNG. I have included BrunoVisnadi in authorship credits as his stage inputs served as the basis for the remaining 5 stages. Some of these have minor (less than 5) variations in routing/frame count. If he'd prefer not to be included in authorship, he can remove himself.

Game objectives

  • Emulator used: BizHawk 2.3.2
  • Aims for fastest time
  • Uses hardest difficulty
  • Uses death to save time
  • Uses luck manipulation


To finish a stage, you have to collect every item. When Jumpman dies, he falls randomly to the bottom of the level, and he can still collect items. If he collects all remaining items this way, the level still finishes, so sometimes it is faster to do this.

Stage by stage comments

Nothing To It (Frames Saved = 7)
Minor change to the end of the route yielded a faster completion time due to slightly less backtracking.

Fire! Fire! (Frames Saved = 17)
Again minor route change yielded faster completion due to less backtracking.

Sreddal (Frames Saved = 38)
As with the previous I only collected the items in the right side then went to the upper left corner of the level and died to collect all the items while falling to the bottom. The improvements to the previous run come from the order in which the top two bombs on the left are collected, an earlier initiation of the suicide plummet, and use of pause to alter RNG fall pattern mid-fall.

Hellstones (Frames Saved = 44)
Entirely new route including death plummet for last bomb.

Figurits Revenge (Frames Saved = 1)
Some route variation yielded minor improvement.

Walls (Frames Lost = 4)
Due to RNG differences, I could not match the previous submission but only had to sacrifice 4 frames.

Zig-Zag (No Frames Lost or Saved)
Identical to previous.

Spellbound (Frames Saved = 29)
By reversing the direction of bomb collection, none of the resulting letters ended up in the correct place in the letter column. This saved frames that are otherwise taken to count bonus points for correctly placed letters. This would have been a 30 frame savings, but I sacrificed a frame to yield better RNG for the next stage.

Blackout (Frames Lost = 51)
As the position of the bombs in this level is RNG based, I was unable to manipulate all the bombs to spawn on the right side of the screen as they were in the previous submission. This unfortunately resulted in much longer routing.

Here There Everywhere (Frames Saved = 3)
Almost identical to previous submission.

Hatchlings (No Frames Lost or Saved)
Identical to previous submission.

Hurricane (Frames Saved = 159)
Here is the biggest single improvement to the run. I spent much less time running into the wind and instead rode the wind currents more to finish the stage much faster than previously. I was also able to collect more bombs after the last input than the previous version.

Total Frames Saved = 243

Potential improvements

There were only two effective ways I found to manipulate RNG: 1) Delayed movement - Either delaying frames when beating the previous level, or delaying steps pre-suicide to alter a fall pattern. 2) Pause/Resume - Pausing and resuming the game alters RNG (likely in a similar way to delaying movement, just using a different mechanism). This method allows for manipulation of fall patterns mid fall instead of just manipulating when the fall begins.

As a few of the levels have significant RNG dependent variables (i.e. bomb placement in 'Blackout'), it's possible that beating prior levels earlier or even later could yield better RNG patterns than I was able to manipulate. This could result in a net faster run even if earlier stages themselves would be slower. Testing this theory would be rather tedious in my opinion.


To those unfamiliar with the game, a picture of any of the various stages is going to look mostly similar to any other; so I don't have a particular suggestion.

Rerecord Count

The listed number is inaccurate due to the bug in BizHawk 2.3.2 that artificially inflates this metric.

Masterjun: Judging.

Masterjun: Removed inaccurate rerecord count.

Masterjun: Accepted to Moons as an improvement to the previous run.

Similar submissions (by title and categories where applicable):