Fiendish Freddy's Big Top O' Fun is a circus game. The story is about a circus that is going bankrupt. We need to perform in 6 events to gain $10,000 and save the circus.
It is a nice game made by a completely normal game company. This is a completely normal TAS and not a part of any double feature submission.
(Rename the file extension from .sav to .zip to open the zipped archive containing the savestates.)
The game name in the gamedb.txt is: Fiendish Freddy's Big Top o'Fun (1990)(Mindscape International)(128K)[aka Fiendish Freddy's Big Top of Fun]
This long name may cause emulator errors, so to be safe you should edit the gamedb.txt and replace the game name with something shorter.
I believe this game has the longest loading time of any game on the site. Maybe the longest loading time for any game ever? The loading time is 14:50. That is almost 15 minutes, even though I skipped all the skippable intro cutscenes.
The main challenge for the TAS is to calculate which events give money the fastest, and which events can be failed early to save time. This version has no unique losing cutscenes for each event and the events give different rewards compared to the Amiga version, so the money route is different. Some events have a little bit of randomness that needs to be manipulated to save time or to get more points.
In this version scoring is simple and the judges give consistent score based on performance. Almost.
Game Objectives
- Emulator: BizHawk 2.11.1
- Aims for fastest time
- Manipulates RNG a little bit
- Genre: Sports
- Long loading times
THE CIRCUS EVENTS AND FRAMECOUNTS
EVENT 1: HIGH DIVING
Swan
Recline | Pike
\|/
Yoga-- --King
/|\
Ballet | Nosepinch
Dive
I suspect that vertical sprite size is causing lag because the 'recline' pose is large but only horizontally and it does not cause much lag.
Poses rated from best to worst based on in-game frame lagging, may not be 100% accurate:
6 yoga, recline
7 ballet
10 swan
king/dive (very slow)
pike/nose (these two poses are instant run killers)
This event gives us random pose prompts and we need to perform the correct pose.
There is a programming error in the game. The poses 'nose pinch' and 'pike' are mixed up, so when the game asks for nose pinch you must do pike, and vice versa. Score is not awarded for these poses, but you must complete the pose or you will not progress to the next pose. This means that we lose $110 if we get one of these poses.
We can wait before starting the event to manipulate RNG and get more favorable poses. Some poses cause more lag than others.
We get $110 per correct pose. We also get $110 for every jump that hits the target. (No variation in score whether we hit the center or edge of the target.)
There is a weird $22 that appears that I can't account for so it may be due to random judges or it may be the default lowest score from event 1. It looks like the score is always divisible by 22, which may be the smallest unit of score.
44579 start level 1 RNG manipulation
44899 jump 1: recline, king, swan, yoga, recline
46234 jump 2: recline, yoga, king, yoga, yoga, yoga, ballet, recline
49724 tape load message
I can do the first jump faster, but the luck for the second jump is good, so I am not going to try it.
I suspect there might be a limit for the number of poses we are allowed to do during the event. I do 5, 8, 3, and no more score is awarded, so 16 poses might be the limit after which no score is awarded. If this is the case then it would save time to spread out the poses more, so that we have more room for manipulating RNG for good poses. So maybe we should do one fewer pose in the first two dives and get the same amount of poses 4, 7, 5. But I don't know if it works like that or not.
I tested doing all unique poses, but that did not increase the score.
Freddy's airfan use has no effect on the score either.
But the game is even more broken. In level 4 of the first event the target can't be hit. Any landing will register as a miss.
If we disregard waiting frames, then there are 32768 different pose combinations for the first dive. That is because we do 5 poses and there are 8 poses to choose from and repetition is allowed. In reality we do something simpler than that because there are maybe only two poses that we want to get, recline and yoga. Our pose strategy is simplified, but RNG manipulation is more or less impossible to optimize manually.
To manipulate RNG we can add waiting frames to the beginning of a level, and between each pose. If we wait 0-9 frames and have to do it about six times, then there are about 10^6 or 1000000 possible combinations of waiting frames. So I believe the whole event has to be tased at least a million times to reduce all the lag frames from sub-optimal poses.
But then we have to also manipulate short judge animations after the event.
One more thing that is broken with the high-dive event is that left-side poses are better. This is because the last press left to 'gain momentum' also counts as input for making the pose. So left-side poses can be started one in-game frame earlier.
EVENT 2: JUGGLING
This is a long timed event, so we want to lose immediately. I tried to do the event differently but was not able to make it faster or get more money than $22 from instant fail. It looks like the score is based on how long we survive without exploding. Juggling the balls in the air does not increase score.
69299 event 2 (last red lag frame after loading)
71769 explosion (took 2470 frames)
It would take about 3233 frames longer to get + $550. This is worse than completing one level of trapeze.
EVENT 3: TRAPEZE
This event has 3 levels with increasing difficulty. One full level gives $726, but we can also fail earlier at varying points to get less score.
In this version we get $88 from instantly losing compared to $240 in the Amiga version.
It takes about 989 more frames to get + $176.
It takes about 3214 more frames to get + $726.
I complete two levels and then lose the two lives as fast as possible: $726 + $726 + $44 + $44.
87192 event 3
89525 standing
89600 press a button to begin level 2
92388 standing
92464 press a button to begin level 3
92984 press a button to begin level 3 again
93756 tape load
101125 judges fadeout +$1540
EVENT 4: KNIFE THROW
Winning is faster than losing, so we are forced to win maximum score. This event is done 3 times with increasing difficulty. Each iteration gives us a maximum of $726, so we get a total of $2178.
This event is also broken to the point of being unplayable. It is almost impossible to hit the balloons with the knife. The cursor moves very slowly and inconsistently.
111289 last red lag frame after loading
111292 press a button to start event 4
Events 4 and 5 have no randomness, so they should always sync with any changes made earlier in the run.
EVENT 5: TIGHTROPE
This event is done 3 times with increasing difficulty. Each iteration gives us a maximum of $726. So we get a total of $2178.
This event is faster and easier in this version of the game.
129420 event 5
129946 win pose
130942 win pose
131962 win pose
132352 event 5 fadeout
139822 judges fade out + $2178
EVENT 6: CANNONBALL
This version asks for player input after each shot, so we can't end input early. This event is also fast, so we would perform all 3 shots anyway.
The final shot is best with power level 3, so that we do not need to move the safety net. This allows ending input a little bit earlier.
157783 event 6 begins (last red lag frame after loading)
157785 start level 1 RNG manipulation, waited 5 frames
158417 target deformed
158898 start level 2 RNG manipulation, waited 8 frames
159508 target deformed
159989 start level 3 RNG manipulation, waited 4 frames
160050 end input
After the judge screen we see the good ending and a fine lady who saves the day. (The big Lady Fortuna has no symbolic meaning of any kind.)
CANNON POWER LEVELS
The cannon power level is randomized from 1 to 5. The power level seems to have no effect on score.
We can move the safety net left or right one step without losing time.
One shot takes 600 frames with power level 1.
One shot takes 627 frames with power level 2.
So power level 2 is okay if manipulating RNG for a better result wastes more than 27 frames.
I tried 200 different starting frames but no 'power level 1' appeared for the second shot, so it is faster to use power level 2. It seems the RNG is sometimes in a loop where it never gives 'power level 1' no matter how many frames we wait.
MONEY ROUTE SYNOPSIS
1 High Dive $2002
2 Juggle $22 (+$550 in 3233 frames)
3 Trapeze $1540 (+$176 in 989 frames or +$726 in 3214 frames)
4 Knife Throw $2178
5 Tightrope $2178
6 Cannonball $2178
TOTAL: 10098/10000
We also get + $4 from random cents so our real total is 10102, but this has no effect on the money route plan.
The cents value is randomized and independent of player performance. I have seen a value of .99, so the 'maximum score run' should aim for manipulating the total 5.94 from cents.
We are close to a value that would allow skipping one pose in the high diving event. Each pose gives $110. 10102 - 110 = 9992. So if you can find a way to get $8, then one pose can be skipped and we reduce some lag caused by a diving pose. But the problem is that the fastest method for getting $8 is to not cancel one pose in the high diving event, so maybe the money route can't be improved?
In this TAS events 4, 5 and 6 are the same as in a maximum score run. Event 2 is the same as minimum score run.
The maximum score category is almost the same as this run, but I am currently not interested in doing it because it involves more RNG manipulations on top of the already very difficult event 1 pose manipulations.
Possible improvements
- Event 1: Manipulate RNG to get poses that cause less lag?
- Event 6: Manipulate RNG early in the game to get good cannon power levels with fewer waiting frames? Manipulation can begin at the end of event 1, which happens about 40 minutes earlier than the diceroll. This is why optimizing the game to the last frame is hard.
- Manipulate RNG to get shorter judge animations in all judge screens? Some frames can be saved maybe?
Co-authors are welcome, if you want to optimize a ridiculous and broken game.
Best places to look for improvements are events 1, 3 and 4. Any event can be edited and then copy/pasted to the rest of the run. Events 2, 4 and 5 should sync always because there is no randomness. Not sure about event 3. The event 6 cannon power level needs to be re-manipulated if improvements are made in earlier events, but that's not difficult in itself.
This game has very long loading times, so to save encoder time and resources, please leave this as 'delayed' for 6 months. If it gets accepted and if nobody wants to improve it, then publish it in December 2026. Or you can decide another timeline if you want.
I now have a newfound awe and respect towards people who make long tases. It is very hard.
nymx: Claiming for judging.
nymx: It took me a while to see what I needed from this. After finding a run, I
found one that helped me to analyze this. This comparison video obviously cuts out the long loading times, but it helped me to see where you had cut times to make it optimal. Now that I understand what is going on, I can see the tactics that you use to winning with the least amount inputs. At first, I was confused. I thought the game was desyncing but it was intentional bad moves to end a particular event.
Accepting to "Standard".
