Donkey Kong Strategy > Basic Donkey Kong Strategy
using bonus as guide for type ii wild barrels
ww:
These are the numbers I use for the bonus when we get the type ii wild barrels on the first four levels. I tried to find these numbers wherever, but I couldn't find 'em. I've seen numbers communicated as after the first 66 seconds, after the first 33 seconds, et cetera, but I've tried to translate these to the bonus.
If anyone's interested in confirming or refuting these numbers, fire away. Thanks!
L1barrels-->type II after 1800
L2barrels-->type II after 4000
L3barrels-->type II before 3600
L4barrels-->type II before 6000
ww:
What a bad post. Finally got off my butt & looked into this. If you want a shorter version, may I recommend the last 2 paragraphs. This longer one explains details and could be a good laugh if I went through this trouble & missed something obvious or duplicated efforts. I have some trouble reading wild barrel types, especially in the danger zone, and I prefer to zip through the first 4 levels, so I like to know what's going on.
I already noticed the numbers seemed to change from game to game, then noticed the tie-in to the difficulty level being displayed on the trainer version. And yes, it changes from game to game. So I ran some trials.
For each level, I did ten trials grabbing no hammers, and ten trials trying to smash lots of barrels, to see if and how smashes made a difference.
I also checked to see if the timing of the bonus started the same way each time, and it is basically the same. The details for L=1: the 4th frame of full graphics, the difficulty seconds on the trainer changes to 34; jumpman appears on the 32nd or 33rd frame (5 times out of 10 each), and the bonus changes to 4900 on the 33rd frame. Since it was so consistent, I didn't bother looking at this on the subsequent levels.
So on L=1, with no hammer grabs, the internal difficulty changed to 3 when the bonus was between 1400 and 1600, inclusive, averaging 1520. However, trying to smash da barrels, it changed when the bonus was between 1700 and 2200, inclusive. Plotting the data points onto a graph, the line of best fit came out to be b=1511+0.5324s, where b is the bonus at the time of difficulty change, given s as the total number of barrel smashes, fire smashes, and barrel-smashes-in-progress at the time of the difficulty change. So I could expect it to change difficulty at about 1500 for zero smashes, and add 100 for every 2 smashes, plus about another half of a bonus tick (a tick being 100 points) accumulates by the time I get up toward 11 or 12 smashes. And so my overall range for L=1 was 1400 to 2200, for 0 to 12 smashes. I would insert a 65kb-ish image here of the graph if I could & knew how. Maybe at the bottom.
On L=2, with no hammer grabs, the range was 3900 to 4100, averaging 4010. The line of best fit came out to b=4010+0.7688s. So the smashes increased the bonus more on level 2, but it's harder to get many smashes in, having just 33 seconds. I could look at level 2 as having a baseline of 4000 for no smashes, increasing a tick and a half for every 2 smashes. Or 3 ticks for every 4 smashes. Or whatever, it's more awkward. So I'm starting to think I'll just keep in mind the range of bonus and smashes and then approximate it on the run. Especially since the variability from the line of best fit just seems to be about 100 points for the bonus. Overall range for L=2 was 3900 to 4600 on the bonus, for 0 to 7 smashes.
On L=3, with no hammers, there was one trial where the bonus changed on the same frame as the difficulty level, and I thought it was arguable that it changed on either one, so I partly arbitrarily gave it a value in between...making the range between 2850 and 3000. Making the average 2975. The line of best fit was b=2973+0.6933s. Overall range for L=3 was 2800/2900 to 4000 for the bonus, for 0 to 14 smashes.
On L=4, with no hammers, there range was 5800 to 6000, averaging 5930. The line of best fit was b=5929+0.7011s. Overall range for L=4 was 5800 to 6500 for the bonus, for 0 to 8 smashes.
Summarizing, since these were the lines of best fit, and ranges of the bonus for my playing (in)ability,
L=1 --> b = 1511 + 0.5324s
L=2 --> b = 4010 + 0.7688s
L=3 --> b = 2973 + 0.6933s
L=4 --> b = 5929 + 0.7011s
I'll use the following as my guide.
L1barrels-->type II after [1400-2200+] {0-12 smashes}
L2barrels-->type II after [3900-4600+] {0-7 smashes}
L3barrels-->type II before [2800-4000+] {0-14 smashes}
L4barrels-->type II before [5800-6500+] {0-8 smashes}
For anyone that skipped to this paragraph, the number of smashes closely correlates (within 100) to the point in the range I would expect the difficulty level to change. So on level 4, with 6 smashes, I'd expect the type IIs to stop about 3/4 toward 6500, or about 6300, plus or minus 100.
In conclusion I figure that during smashes, the difficulty clock keeps running, while the bonus is put on hold. Clearly, I'm assuming that the difficulty indicator on the trainer is either accurate or our best attempt at getting it. I hesitate posting the simple charts (ha, Jry) I created to see how solid the correlation looked, and to look for outliers, but I will. And I expect that I'm not sharp enough to delve further for other causes of the variability. And I wouldn't be surprised if there's something obvious that I'm missing. But I guess the smashes are the biggest variable in changing the timing of the difficulty switch versus the bonus. If anyone knows other factors at work here, I'd love to hear about it. Thanks.
ww:
just figured out how to edit a post, so i've changed the errors i found. der.
xelnia:
I love it.
And the quick response is: yes, the internal difficulty timer keeps running during hammer smashes while the Bonus timer pauses.
It looks like you worked that out and the really interesting bit is that you worked out how those smashes affect the resulting Bonus timer value. I have done this experiment before but never took it that far. In fact, I was fiddling with this last month and completely forgot about this thread until now. But your Level 4 values seem to line up with what I was seeing in my own tests. In my tests I was using the MAME debugger and using a breakpoint to immediately pause the game when the difficulty changed. That way I could get "frame-perfect" counts for each change. I actually thought I started seeing a pattern, but I'll have to go back and look...
Another important thing to consider, as your tests implied, is that even without smashes there is still some variability. This is a result of the not-exactly-consistent release of barrels, wild barrels, Kong pauses, etc. This variability doesn't exist on the other boards.
I think you would want to run more than 10 trials...hundreds maybe...with no smashes to get a reliable distribution. Since hammer smashes are fixed in length, you could then apply any number of smashes to that data to get a sort of modified distribution of what to expect per number of smashes. But I'm not a mathematician (shhhh)...
Anyway, great stuff, great post. Good work.
ww:
Thanks. Yeah, it would be nice to do more than 20 trials per level, but it already took a lot of time. But I was pleased with how each distribution looked pretty balanced and normal anyway.
Navigation
[0] Message Index
[#] Next page
Go to full version