Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Lars_Hendrick wrote:
CoolHandMike wrote:
Could any of those possible improvements lapogne mentioned be implemented without redoing massive sections of the tas?
I'm willing to take a day to go back on the final level at the request of the judge and see if I can't scrape a few seconds off with the reload/grenade hotswap but aside from that it's basically throwing away whatever work has been done after said trick is recorded.
Use this clear as a reference for the boss battle https://youtu.be/1yi267UzRMg?si=94uuSc34PDswiwg1
You may also consider redoing the end of The Hooded Man, but yeah anything more would not be realistic as a "small" update
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Don't worry I read the messages on the TS Discord, I just don't post much on it. I have no complaint about the entertainment, but I find the technical aspect to be borderline, hence why I made my post.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Since I have a lot of knowledge about this game (literally my username origin), I will try to give my comprehensive opinion about this TAS.
General observations
Movement-wize, it's not absolutely perfect but it should be good enough as far as 3D game TASing goes, however the category is weird. Hard difficulty is obvious, but no OOB and all bosses have never really be runned before, either for full game speedrun or individual levels (ILs), mostly because they are only relevant in a few places : 2 levels for the no OOB, and I believe 2 bosses that can be skipped. If anything, it makes comparison with ILs times harder. My main issues with this TAS have more to do with some questionable gameplay choices throughout the TAS, most notably the lack of damage boost from enemies, and sometimes straight up mistakes.
Time to Split (8.4s faster than IL WR)
Cortez took a small detour at 2:00 to get grenades for unknown reasons. Timesplitters should die with 2 headshots from the sniper so there are some extra shot that were done, probably didn't cost any time. At the end of the defense section with the turrets a path previously blocked opens, I would have expect Cortez to be closer to this new entrance when that happened (~5s lost). Same for the final section, once all the timesplitters are dead I would have expected him to be close to the final door (~4s lost).
Scotland the Brave (can't compare with IL WR)
ILs use some sort of new game + glitch that gives you KSMG Rocket ammo each time you pick said weapon, so the TAS is at a natural disadvantage here. Regardless, this glitch should save around 10~15s in total, and the TAS is 58.4s slower than the no OOB IL WR. It is due to various things, including :
Using the crane to lift the submarine despite it not being mandatory (~7s lost)
Not killing the turret guy at 6:56 faster (5~10s lost)
Not restarting from checkpoint at 8:10(~9s of IGT lost, few seconds less in real time)
Taking a detour at 8:30 to get grenades, instead of a much faster one at 6:13 (~6s lost)
The Russian Connection (7.8s faster than IL WR)
Weird first section where you must kill all the guards, but it's more or less the same time than the IL so maybe there is a reason why it's done this way. Very clean "protect Harry" section, maybe could have killed the last guy with a grenade from 11:52 but that's all. Not sure if the clip through the door with the jeep (is that not an OOB?) could be done faster or not, it saves ~8s over the IL which doesn't use it as it wasn't a known glitch at the time.
The Khallos Express (4.5s faster than IL WR)
Both helicopter fights should only take 3 rockets to end instead of 7 and 6. Everything else looks good.
Mansion of Madness (can't compare with IL WR)
Headhunter skip is a thing since 2007, I couldn't find what was the IL WR on hard difficulty before that, but it was between 6:00 and 6:28. Current IL WR with the skip is 94.7s faster than the TAS, and the skip saves ~85s. From what I saw, the rest of the time loss mostly came from a slow "garden barbecue", slow trigger of the second boss and not hitting the "belly" of the boss at 28:40.
What Lies Below (18.8s faster than IL WR)
Nothing noteworthy.
Breaking and Entering (45.8s slower than IL WR)
The "protect Amy" section is ~6s slower than the IL WR. Lost some time due to using the time grenade, which I believe slow Amy's movement. Waited a few seconds for nothing at the end once Amy finished her speech. Most of the time loss is due to not using the pause glitch to save IGT (does not affect real time).
You Genius, U-GENIX (33.4s faster than IL WR)
Lost ~4s at the beginning by killing the spiderbots slowly. Saved ~19s with the cutscene trigger skip, currently not used in the IL WR. Not so sure about that "one mutant does not explode when hit with the injector" since IL WR use the injector just fine. The Dispersion Gun is inaccurate but Cortez should still be able to one shot the ceiling turrets with it or time grenades. Also, you only need to kill the ceiling turrets in the first and second phase, not the spiderbots. Everything else is fine.
Machine Wars (can't compare with IL WR)
IL WR is done on 50 Hz since it significantly increase the tank sections speed, the full game TAS obviously can't do that. Movement was sometimes a little weird. I don't understand why Cortez didn't immediately start to shot the spaceships at the Goliath section, this section can be ended significantly faster than that.
Something to Crow About (can't compare with IL WR)
IL WR is 19.5s faster but with the boss at the end skipped, which save ~12s compared to this TAS fight. Lost ~9s compared to the IL WR in the energy sphere room alone due to not killing some of the enemies (including the Goliath) faster, used 4 grenades instead of 1 to destroy the sphere.
You Take the High Road (28.2s slower than IL WR)
Most of the time loss come from not using the pause glitch to save IGT. Losing some time in the 1:08:48~1:09:00 section by killing enemies who don't matter for opening the next door.
The Hooded Man (7.7s faster than IL WR)
Snipping is good, except at the final section where every timesplitters can be killed with 2 headshots. Some shots were done before the timesplitters lost their spawn invulnerability.
Future Perfect (4.8s faster than IL WR)
Lost ~4s to the boss fight, he could have been killed faster by switching between weapons instead of watching the KSMG Rocket reload animation.
Sorry if I may have sounded harsh, it's just that it felt like a decently executed TAS but with limited knowledge about the game or RTA runs.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Is there a reason why this TAS is 30 minutes slower than this one? https://youtu.be/Tzj5vnpGMwA
While the execution looks good, the strategies used are suboptimal compared to this other TAS, this is particularly noticeable in the late chapters.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Yeah it wouldn't be surprising that the coordinates are at a different address for each level. If you have no luck finding the coordinates with 2-byte values, you should check 4-byte values too. If you have found the Z-position address, then it is near guaranteed that the X and Y-position addresses are next to it in the RAM.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
The class change from Dracoknight to Paladin was mainly to have a higher maximum Speed stat (25 instead of 23), which is relevant in this difficulty for the later half of the game.
Enemy phases can be slightly manipulated if there is a level up happening, letting me choose between 4 different outcomes instead of 1, but yeah manipulating the randomness is limited in this game in general apart from the initial seed.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Was it technically well done? Yes.
Was it entertaining to watch? For the 4 first hours, I would say yes, but then it was only fedex quests with occasionally a boss fight cleared with an hard code 9999 damage attack, which was a very sad change of pace.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
After looking at the damage formula thread, it's an underflow issue where the devs didn't expect Ryu to stay at a very low level. In a similar fashion, if TargetDef / 2 is exactly equal to AttackerLevel * 20, the attack should always deal 0 damage.
For combos, I would assume either the "20" in the formula is changed to a lower value, or the TargetDef is increased, leading to easier to fulfill underflow. You may probably be able to get a better understanding of the combo formula by manipulating TargetDef and AttackerLevel in the RAM and see in which cases you get an underflow (or even better, 0 damage).
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Meerkov wrote:
Looks good.
One question: At 3:45, there is a long hallway and you go diagonal towards the wall before strafe-boosting. Would it be faster to go directly to the wall (left), instead of diagonal?
Nice run
Given the speed with and without the straightboosting, a 45° diagonal is always faster than going directly to the wall. Though I just did the math and, in theory, a ~30° diagonal should be the optimal angle.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
Finally managed to get my account back after filling an appeal with minimal information, apart from "please give me a human being" to maximize the probability of actually getting one. Still got the "Please be aware of the servers that you're in, and the content that they post" though.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
So far I am at 2 rejections from the bot, does someone who got unbanned as a contact from the Discord support because it's really annoying to deal with this bot every time...
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
I noticed that the RNG doesn't change often (i.e. same damage and item drop even if I delay the attack by several frames), so I noted at what frame the result was finally different, and I kept all addresses which didn't change before this frame but were different after. Then it's just the elimination process by freezing some addresses I mentioned in a previous message.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
I got some time to look at this game, and the RNG address is at 0xC4468
It follows the default PSX RNG formula:
NextRN = (CurrentRN * 0x41C64E6D + 0x3039) mod 0x100000000
If you want to know how many RNs were used since the beginning, you can use this script:
local RNG = {}
local i, j
local MAXLIM
local CUR_RN
local a, b, c
RNG = {0} --First RN value
local MAXLIM = 10000 --Number of RNs to calculate
for j=2, MAXLIM do
-- RNG[j] = (0x3039 + 0x41C64E6D*RNG[j-1]) % 0X100000000
-- Very convoluted way to calculate the next RN without any big number rounding issue
-- 0x41C64E6D = 1103515245
a = (0000005245*RNG[j-1]) % 0X100000000
b = (0003510000*RNG[j-1]) % 0X100000000
c = (0010000000*RNG[j-1]) % 0X100000000
c = c * 110
RNG[j] = (0x3039 + a + b + c) % 0X100000000
end
local BASE = RNG[1]
local INDEX = 1
while true do
CUR_RN = memory.read_u32_le(0xC4468)
if BASE~=CUR_RN then
for i=math.max(INDEX-100, 1), INDEX+100 do
if RNG[i]==CUR_RN then
INDEX = i
BASE = CUR_RN
break
end
end
end
if BASE~=CUR_RN then
for i=1, MAXLIM do
if RNG[i]==CUR_RN then
INDEX = i
BASE = CUR_RN
break
end
dummy = i
end
end
gui.text(0, 100, "RNG")
gui.text(0, 120, "#" .. string.format("%07d", INDEX))
emu.frameadvance()
end
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
If you have many addresses, you can freeze several of them at the same time and check if it had an impact in random events. If so, then the RNG address is probably among them.
If you freeze a lot of addresses at the same time, the game may crash, in that case select less addresses and retry.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
You should first find where the RNG is in the RAM. It is probably a value which change very often, and if you freeze it then you will always get the same random event.
Just in case you should check the address 0x9010 (4 byte) because that's the default RNG address for a lot of PSX games.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
AmaizumiUni wrote:
AmaizumiUni wrote:
The first three battles
Chimeras, demons, and wizards can take the first attack with adjustments.
Left＆circle button push flamenumber battle in
I checked and now I understand what you meant. For the first 3 battles, you are supposed to go second and the game doesn't even use the RNG to determine this, but for some reason if you press the left or right button at the beginning of the battle, the game will use the standard RNG check to determine if you go first or second.
However for the first battle, it turns out that the RNG makes you go second if you use this trick, and as far as I know you can't manipulate the RNG at any point before this battle, so you can only save time with it for the 2nd and 3rd battles (around 10s time save in total).
I will try to hex-edit my way through the game, though I will probably have trouble syncing the RNG.
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
TIL the name of this RNG algorithm (pretty common for RPGs), not sure why it's "bad" though.
For these sort of TAS, you should progress normally up to the point where you need a specific lucky event, check how many RNs you must burn in order to get this event, and hex-edit your movie file to burn the same number of RNs in the most time efficient way (different movement, waiting in a screen where RNs are burnt over time, ...).
Editor, Experienced Forum User, Published Author, Skilled player
(1282)
Joined: 1/31/2010
Posts: 330
Location: France
andypanther wrote:
I have a question regarding a potential 100% TAS of Star Wars Episode I: Racer:
When would the correct ending point of the movie be, according to the rules?
Let me explain the structure of the game, so you can understand the issue here. Defining 100% completion is not the problem: To unlock everything, you have to win every single race of the tournament mode. The tournament mode consists of three main circuits, Amateur, Semi-Pro and Galactic. On a new file, the first track of each of these circuits is available. The Galactic circuit displays the credits when completed, so it serves as the any% category of the game. Beating these three circuits will also unlock tracks on a fourth circuit, the Invitational circuit. This means that 100% completion can only be reached after the credits.
This is what happens when you win the last race:
- After crossing the finish line, you advance to the last character unlock cutscene with one input.
- The cutscene will play and fade into the last race result screen. With one more input, you return to the track selection, where it will now be visible that the last race has been won.
What the RTA community does is to skip the credits with a reset and end the run when winning the last race.
By the way, in case someone wanted to bring up vehicle upgrades as part of the 100% definition: The individual upgrades are not permanent, you can change them at any point. The only permanent aspect of the upgrades is which ones are available at Watto's shop. This depends on the number of completed races and is just achieved as a side effect of every 100% run.
