Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
Thank you for trimming the ending. I had left those frames in because the final "Thanks for playing" took a few frames to show up, which you included in the encoded video.
I'm glad you like the subtitles! I spent 3 evenings upgrading my scripts to include a subtitle output so you can actually watch the granted fiefs increase the AP and compare it to the next target Fief.
I'm so happy it's done! Now I just hope it gets accepted.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
ars4326 wrote:
CogneatoSwitch, to elaborate on what Spikestuff addressed, using auto-fire isn't the best method to use when TASing. You'll want to get into the habit of using frame-advance and manually attempting to advance menus, and such, at the earliest possible frame. I made a brief test run, up to the stat creation screen, that saved 6 frames just on optimizing input.
Also, and I don't know how this would affect your scripts, but I saved over 300 frames at stat creation by getting values that were just below the max allocation of 109. I stopped before deciding on IQ since finalizing from there will determine the initial RNG seed (otherwise, the AI sent a Ninja toward my fief for the game's very first move!).
Anyhow, this isn't necessarily grounds for rejection; just me chiming in and sharing tips for better TASing. I'm still in the process of judging your run, and should have a final decision made later on in the day.
The opening sequence is the least optimized, as it is the rarest script used. Also, I purposefully wait frames in the beginning to speed up the stat selection; the RNG seed starts as soon as the game begins, and so waiting frames earlier results in faster stats selection.
I believe you are correct that I can save some time by ignoring Health and Drive to an extent; Health leads to sickness and (I think) plague, and Drive keeps enemies from attacking.
Luck is tricky, because it's difficult to change after the game start, and if it is 30 less than the enemy's, ninjas will fail. Since Fief 13 has the highest luck and is one of the quickest killed, I believe it is faster to spend time at the stats screen rather than use seasons to increase luck.
Char+IQ are directly needed to assassinate. I spend a majority of the time increasing Char so I can assassinate Oda. Thus, spending extra time at the beginning to maximize the stats saves me a huge amount of time later in the game.
Overall, 6 seconds are not worth "saving" if only to spend 6+ seasons recouping the lost stats. Similarly, "saving" a few frames at the beginning isn't worth having to wait 10+ seconds for the better stats.
That said, I only checked the first few wait frames (I've dubbed them "seeds"). It may be faster to wait longer on the title screen so the stats maximize faster.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
Spikestuff wrote:
Edit: Looking over your TAS it's fairly sub optimal.
For starters 'cause it's my favourite part, is that you're end input could be ended my sooner than where it does. (84374 not 84388)
The second notable part, is that you're doing input like this:
Down,*Lag Frame*,Down
Your input won't register since this happens. This occurs quite a lot in your TAS.
Even if the case is meant to be like:
Down,*Lag Frame*,Down,*Blank*,Down.
You still have the chance of that blank frame being the frame you want to hit down on instead.
This last one probably gets factored out due to RNG. Terrible startup to the game.
Please don't use the Autofire in your input.
Yes, in the future I shall end it sooner. I thought it would make for a better show, rather than worrying about the absolute shortest.
What on Earth is Autofire? I hand-crafted every script, and autofire wasn't one of them. What is the effect of it, and how can I avoid recreating it?
It sounds like you want a minimal input style, rather than a stream. That is, when I need to press "down" to change numbers, I should press it for exactly 1 frame and wait until the lag frame passes (3 frames, typically) before pressing again, rather than press-and-hold for 4 frames, or toggle on/off every other frame.
Is that correct?
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
Sure, I will add subtitles next time!
So to understand the oddities in my run, first recognize how I'm making it. I've generated a series of scripts that execute specific commands quickly. Thus, when I make the video, I merely execute the scripts in the right order.
When I initially made the GPR script, my goal was to improve wealth and loyalty, which is useful for gaining more money. By default, I had it give 25% of total rice. I can move the number down to 1, but it messed with other scripts down the line, and occasionally changed the RNG. Thus, changing the default rice value meant redoing the whole video. Since I plan on remapping it anyway, I went ahead and submitted it on the off-chance I run out of energy to do this TAS.
Stats: Drive is the only one that isn't necessary. Luck and IQ both play a part in assassination, but I can't seem to manipulate them as well as I can Char. I suppose a moderately high health score would be acceptable, as long as I didn't die from sickness. Also, changes to this area also affect the RNG "seed", which means remapping the entire game. I'm not inclined to change this because the savings is so small and I already know this seed.
I haven't figured out the random events or the next-season turn order. My only solution for them is to adjust my actions. I've found the choice of action, rather than intensity, makes the most difference. For example, choosing "Send" instead of "GPR" might change the outcome. I also noticed that the event happened anyway, it just changed where it landed. I don't think there is much savings to be had there; maybe a few seconds at best.
In a battle, unit movement always goes from 1 to 5, with 2 and 3 being the most powerful units. I typically send the 4 or 5 to get killed by the enemy's 2 or 3, and save my 2 and 3 to kill the enemy's leader. In this particular video, it didn't matter. However, I agree that changing the assignments before the battle will greatly enhance speed, even if it costs a whole season.
The biggest bottleneck I've found is not using grants. The granted fiefs deliver 1 Char 50% of the time, so if I granted them all, it would take twice as many seasons but 1/3rd the time, with an estimated savings of 7 minutes. The next biggest is the GPR values, which were used 81 times and could easily save 2 minutes. Making assignments before the last battle may save a few minutes (assignments are slow, and the whole season will have to go by). Changing the bids might save 15 seconds and being looser on the stats selection might save 8 seconds.
So I guess I predict a better video that is 13 minutes or less in length.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
The old one has been cancelled.
Yes, optimization for seasons would be much slower, as starting a war would save a few seasons, but each war takes about 25-40 moves, or 3-5 minutes.
The resulting video would showcase much more the war strategies, such as bait-and-conquer, ambush, or even gaining fiefs by killing the leader (some of these are seen in the first video).
Because of the drastic difference in presentation, I propose that the type of optimization be a separate branch.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
Well I did it. Not only did I get a TAS for Level 5, it's also faster. I managed to beat the game using all assassinations except for the last battle.
http://tasvideos.org/4785S.htmlhttp://tasvideos.org/forum/viewtopic.php?t=16989
The assassination-only victory doesn't have the same feel, so I don't consider it the same as this original run, even though it is significantly faster.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
So I'm having a LOT of trouble even getting started in level 5. Not only does the computer have a significant starting advantage, every action I take is only about 20% as effective. Even when attacked with equal armies, I cannot kill their leader, despite getting 4 attacks. (The enemy kills my leader the following turn)
I still have an idea of killing everyone by assassination, so I'm not ready to give up yet.
As an alternative to difficulty, I could map solutions based on starting fief. Fief 1, 9 and 12 are notoriously weak and are generally killed within the first few seasons. It would fulfill a childhood challenge for me to beat the game starting from fief 9, and I do intend to create a solution for that on level 1.
@mklip2001 I don't blame you for not watching all the way through. With the menus going by so quickly, it's difficult to tell what's going on, and I found the music causes a bit of lag, so I turned both off for the sake of speed.
If folks think I should slow it down to make it more interesting, I am willing to oblige, as I don't think either the wait or the music affect the RNG.
Experienced Forum User, Published Author, Player
(23)
Joined: 7/26/2015
Posts: 9
This was my first TAS, and I didn't know the harder difficulty was preferred over speed. I will map out a level 5 solution as well. However, I believe assassination is based on the leader's stats, so there will be fewer assassinations and thus a much longer video.
In this particular scenario, Imagawa and Hojo were much weaker and were taken out more easily. When choosing Oda, I found no assassinations available. However, with a level 5 difficulty, I will likely change my routing.
Taking out the leader in battle gives all the fiefs, but if they are assassinated, you have to bid on each one, and even then I wasn't able to bid in some cases.
Gaining fiefs by killing the leader is tricky: they will typically run before there is an opportunity for victory, but a solution may be available using save states, though I was unable to find one when attacking Uesugi at Fief 2 (before attacking Fief 4 to cut him off).
I believe the fastest way to victory is to bait enemies to attack and use TAS to kill leaders and gain men when not your turn. In the cases the leader leads the army, then you'll gain their land too.
I was also wondering, is wall-clock speed or season count preferred for this TAS? The strategies will be vastly different, as assassinations or even waiting for enemies to die of old age could be much faster than fighting battles. Also, putting grants on certain fiefs will speed things up, but sacrifice the gold and rice that could be sent, taking more seasons to attack.