Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I use frame search.
Justification is that RNG only seems to vary by frame delay.
Here's what I usually do for manipulation (must not use fast text speed):
- Make a savestate one frame before first letter of next dialog appears.
- Turn on turbo speed.
- Hold A and unpause, and see if it works.
- If it doesn't work, keep A held down and press frame search. Keep bashing frame search until it works.
- If it works, end frame search and make a savestate.
I bash frame search so quickly that I often overshoot the frame where it works and have to press frame search previous.
Occasionally, I run out of text. Then I have to go back and wait a bit longer.
P.S. If you are using fast text speed, too bad.
marioguy wrote:
What I'm trying to say is that I don't see a point in posting a new Sapphire run every month. Just keep on practicing until you can't find any other ways to make it faster, then post it. I'm betting there will be another Sapphire run released in about a month.
The two submissions I posted were 3 months apart.
And I'm not working on this game anymore.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Give it your best shot, Tragedy.
Check the published MMX run for the latest tricks.
Also see: http://tasvideos.org/MegaManXTricks.html
That page needs some serious reform.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
LagDotCom, have you tried making the HP as low as possible? I think I had 19 once.
Also, if you have spare time, try a run using mid-speed text. It may (or may not) be better.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Basically, my first submission of this game was at a standard which was higher than that of kirbymuncher's standard for its submission, and I beat it by 12.5 minutes. I felt my run was close to perfect (I often have high standards for submissions). Yet I felt that I did not do enough. I made a new run with the goal that I would use a very different strategy (including using a different Pokemon for over half the game), and make my other run look bad. I even had the discipline to give up on a "final" run (which also took two weeks to make before I gave up near the end), something I lacked on my first submission for this game. Not all was lost, since the failed run taught me a new trick.
So a run that I once believed was close to perfect was beaten by 1.5 minutes by a run that I now believe is much closer to perfect. So, no, it is not a minimal improvement.
I believe perfection is no less than 1:33. Of course, I hope I'm wrong, but that does not change my belief.
Also, I hope people don't believe that the improvement stage starts after a run is published. In this case, the planning started long before it was published. There were reasons for me not cancelling the submission, one of which was because I was not sure whether the new strategy even worked. Eventually, someone was willing to publish the submission and I let it be published (many thanks, adelikat).
This is not an argument against "watching Pokemon is boring". If one thinks watching Pokemon is boring, then to that person, watching Pokemon is boring.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Done and submitted:
http://tasvideos.org/forum/viewtopic.php?t=5783
I was joking back then but I did beat 1:35:11, something I didn't really expect. I didn't even think it would beat 1:36.
On a less serious note, the last 20 posts in this thread have all been from Canadians. :)
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I mindlessly switched Taillow with Castform even though Taillow was not the lead back-up Pokemon, and didn't notice until I reached Liza and Tate. This sets me back a whopping half an hour of TASing. Excuse me while I laugh at myself.
Yes, by skip, I meant run away.
Edit: Liza and Tate now done. I saved another 4.5 seconds somewhere.
Edit 2: Wallace now done. Saved another 1.5 seconds.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Norman done. I caught Taillow on the way, so a straight-up comparison says that this run is 20s faster than the previous, which in turn is 8s faster than the practice run.
Checking Machop's IVs, I found out that it has near-max Attack stat, with the (+Atk, -SpDef) nature and 28 Attack IV. It does come with a cost, namely the worst SpDef possible. I'm not joking. Who cares though.
IVs
HP: 13
Atk: 28
Def: 23
Spe: 1
SpAtk: 7
SpDef: 0
The 0 IV SpDef and low IV SpAtk actually threw me off while trying to find the memory containing the IVs, because the high byte of the encoded IVs was the same as the magic XOR number, which it rarely is, and I look there first.
Of course, the biggest thing is Attack and I have a lot of it. Defense and HP are important only to stay off the Aerial Ace OHKO route (if they can't OHKO, you can make them do no damage) while using items. The other stats are absolutely meaningless; they could all have 0 IV for all that matters, though whether I could discern the IVs from other stats in memory is another question.
I do need to make sure later on that Castform doesn't have bad SpDef, because it needs to survive Solrock's Psychic.
Regarding fadeout, here are the possibilities:
- no bike: can't do anything
- (on bike, no bike music) music A -> music A: do as normal
- (bike music) normal music A -> music A: just before exit, bump bike, get off, run to exit
- (on bike, doorless entrance) music A -> music B: from farthest point possible, bump bike, get off and on, bike to exit before bike music comes
- (door entrance) music A -> music B: do as normal
I also found out that the delay caused by turning after entering a new area can be cancelled somewhat by turning and then using the bike. It is still faster to go forward if possible, though.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
The time wasted to stop and accelerate again is definitely less than 84 frames.
It depends on the situation. In fact, I intentionally bumped the bike just to get off and quickly hop back on. This immediately plays the regular area music and fades, trying to get to the bike music. As it fades, I exit the area so it doesn't wait for the music to fade. This is assuming I can ride the bike, and the regular area musics of the two areas are not the same.
Currently done Flannery. Estimated to be about 12-13 seconds faster than previous as a relative comparison. I still haven't caught the bird yet.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I didn't correct anything before Liza and Tate. Just the Seafloor Cavern and later.
I'm currently at Elite 4, but some time will pass before the submission occurs.
The fadeout property does not extend to earlier games. The music does fade out but the next area is clearly loaded.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
No, the Machoke version is faster. The AI being stupid helps.
I discovered something rather unsettling. If the music fades out from one area to another, the game waits for it to fade before loading the next area. If the music doesn't fade out, the game doesn't wait. There were times that I triggered the bike music and entered a new area. It would have been faster to dismount the bike to trigger the normal area music, then enter.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Done Liza and Tate. Last time I checked (after beating Winona), the run is 15 seconds faster than in my test run.
As long as I beat 20 seconds, I can beat the now-published run.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Kyrsimys wrote:
How about the front page?
DK64_MASTER wrote:
It seems like it would fit in nicely under "starting points," perhaps under the "List the recommended movies" bullet list?
Done.
I forgot about that page after seeing it so long ago. At that time, it seemed like a bad idea. Not anymore.
Zurreco wrote:
Has anyone actually complained about this before?
Yes.
Zurreco wrote:
It seems like the only people who would struggle with this would be 56k modem users
Actually, even users who have anything less than insane download speeds. Not really struggle, but just don't have the patience. I have mixed feelings about the thumbnails, though.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I'd be surprised if any v19- GB/SGB/GBC movie worked.
There is a link on sticky topic's first post to the v19 rerecording version of VBA, which should fix the problem.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Somehow, I found this run rather repetitive. It was only until the last level that some different "ranger" was used. The boss fights all seemed the same.
Voting no. Keep in mind that I'm a first-time watcher.
Yes, I know that this game has a run already published. I do not want that to influence my decision.
Good job anyway, 西坡.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Started shortly after finishing some other TAS. I'm now trying to get Machop.
Steven's Armaldo is in the back of my mind. Somehow, I was so close (in the test run) to getting a Cross Chop KO but I couldn't (maybe 1 or 2 HP away). When I caught Machop, it was L16 with 21 attack. Now in this run, I encountered a L15 Machop with 21 attack, which is better than L16 with 21 attack. Problem is, the missing 500 experience will result in not leveling up to L48 for Armaldo! So I need to KO Kecleon or something, or take a Slash hit from Armaldo.
I wasn't able to come up with L16 22+ attack Machop. Machop is not common, and the RNG is not versatile since it only uses time dependence.
Edit: I did come across a L16 21 attack Machop somewhere. As a last resort, I will check if its attack IV is higher than in the test run.Edit2: I'm sorry, I really didn't know what I was thinking. My test run caught a L15 $21 attack Machop. By the way, the attack numbers were in hexadecimal. :X
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
For some reason, if the filename is too long, it takes the first 25 characters and the last 5 characters and sticks them together. It's happened with some of my FileFront files.
Angerfist, you meant "upload", not "download".
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
People don't seem to understand that the published run is one-player only.
Two-player mode is that much better.
I don't know how this game is revolutionary. This game looks like a laughingstock. Perhaps it's because I never saw this game before three days ago. Perhaps it's the TAS. To me, it's a bad game (to play) that makes an entertaining TAS (i.e. entertaining to watch).
