Joined: 3/18/2006
Posts: 971
Location: Great Britain
The tool works very well. At the moment we are are working out whether it's better to have a battle before or after GS... or have no battles at all. Some things: |Need Cloud to have a semi-high limit break bar for the next boss |GS can't hit Cloud |Barret needs a limit break in GS battle
Active player (276)
Joined: 4/30/2009
Posts: 791
If my understanding of the random encounter avoidance process is correct, you cant avoid every single encounter, even with adjusting movement between walking and running, and the menu trick. Wouldnt it make more sense to make use of these unavoidable random encounters to build up Cloud's Limit Bar, assuming you can manipulate the enemies to get a surprise attack or two on Cloud for that purpose?
Joined: 3/18/2006
Posts: 971
Location: Great Britain
In the run you can probably avoid every random encounter if you wanted to. There are certainly no unavoidable encounters at this early stage. Although what you said makes sense for later on. I can finally do some more tests today. It may even be faster to get into a battle before GS and allow both Barret and Cloud to get hit... This is because it is nontrivial to make Barret get a limit break in the GS battle. 1. Manipulate starting ATBs 2. Manipulate critical hits 3. Manipulate GS into only attacking Barret [If GS attacks Cloud, quit, enter the battle a few frames later, and start again from 1] In total this wastes a lot of time indeed.
Banned User
Joined: 5/11/2004
Posts: 1049
Sounds like rom hacks and programs that take the AI algorithms and show you what they'll do on each frame would help a lot. Can extraneous user input during battle effect anything? Hope this run gets all the support it needs, it will be awesome to watch.
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Joined: 3/18/2006
Posts: 971
Location: Great Britain
I think only delaying an attack can manipulate the enemy.
Banned User
Joined: 5/11/2004
Posts: 1049
ahh cool. So what did you do, just keep delaying attacks until GS did what you wanted? I'd be really nice to actually have a look at the code to figure out if there's anything besides delaying attacks or entering battle at a different time/state can affect what the enemy does. Maybe changing menu options before battle could help or configuring your guys sub-optimally or trying to run away for a while, is there anything else you can do?
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Banned User
Joined: 5/11/2004
Posts: 1049
ignore double post
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Banned User
Joined: 5/11/2004
Posts: 1049
ignore triple post
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Joined: 3/18/2006
Posts: 971
Location: Great Britain
update Guard Scorpion is dead! I'll let the HEADS OF FF7 SPEED-RUNNING take a look before I continue. Many thanks for Apo123's updated Java bot. It was indispensable against Guard Scorpion.
Banned User
Joined: 5/11/2004
Posts: 1049
sweet, what was java bot able to do? This seems like a project where programming assistance will be necessary.
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Post subject: Re: Final Fantasy VII (guard scorpion killed!)
Joined: 1/26/2009
Posts: 558
Location: Canada - Québec
antd wrote:
I'm aiming for fastest in-game time. This means there may be soft-resets used later on.
soft-reset might be a bad idea if you don't want to put time in studying memory corruption while saving, etc... or maybe there a difference between "using memory corruption" and "reset in a safe aera" categorie Also, I simply hope that you won't going too far in optimization, since we're still somewhat limited about the actual tool and this game is a bit long :) anyway, let's keep thing rolling on and please.. post your pxm wip!!
Joined: 3/18/2006
Posts: 971
Location: Great Britain
Why would soft-reset, to return to the main-menu (start+select+shoulder buttons), cause memory corruption?
Joined: 1/26/2009
Posts: 558
Location: Canada - Québec
oops sorry, look like that I was confuse on this point. Alright then can't wait to see your run strategy with limit break manipulation....
Joined: 5/17/2008
Posts: 212
Location: Virginia
Spider-Waffle wrote:
sweet, what was java bot able to do? This seems like a project where programming assistance will be necessary.
All it did was manually iterate over a sequence of frames. In this case, wait x*4 frames, attack, then wait 15 frames, with x increasing by 1 every time a button in the program is pushed. Essentially, it's a very dumb bot (the sequence is currently hardcoded, although I'm going to change that very soon). I'm currently trying to make a user-friendly way to change the sequence (without requiring programming experience), change the key mappings, and save and load sequences.
adelikat wrote:
It started off fairly tame, but as more balls entered the picture it sure got a lot more entertaining.
Active player (276)
Joined: 4/30/2009
Posts: 791
Any chance of an encode of the GS fight? :) Great news, keep it up guys and gals.
Joined: 3/18/2006
Posts: 971
Location: Great Britain
New plans Midgar + GS (Part 1) http://www.youtube.com/watch?v=2o84iJwnRmU
Joined: 11/22/2004
Posts: 1468
Location: Rotterdam, The Netherlands
Glad to see you're working on this!
ventuz
He/Him
Player (123)
Joined: 10/4/2004
Posts: 940
I think making "Battle message" to FAST was unnecessary.
Banned User
Joined: 5/11/2004
Posts: 1049
does battle message not effect anything?
"Great spirits have always found violent opposition from mediocrities. The latter cannot understand it when a man does not thoughtlessly submit to hereditary prejudices but honestly and courageously uses his intelligence."
Mastania
He/Him
Joined: 5/30/2009
Posts: 135
Location: Top of the pops
Ah that reminds me of something. Has this testing revealed how limit break crits work? Is the chance really the same as a regular attack or is it lower?
Waddle Waggle Wattle
Joined: 3/18/2006
Posts: 971
Location: Great Britain
This part never synchronises
Active player (276)
Joined: 4/30/2009
Posts: 791
ventuz wrote:
I think making "Battle message" to FAST was unnecessary.
The battle messages can be skipped faster when the bar is set to fast. So yes, it was probably necessary, even though there are few battle messages throughout the entire game.
Joined: 3/7/2006
Posts: 720
Location: UK
I'm not sure about the battle message thing either. The only ones that come to mind which actually slow the battle down are: 1. Barret talking about Guard Scorpion (never appears in current TAS) 2. Hojo? 3. Bizarro Sephiroth's various part messages
Voted NO for NO reason
Joined: 3/18/2006
Posts: 971
Location: Great Britain
I cannot get 2 steps out of the reactor. It desyncs when I leave the elevator. At first it would desync on the first step, after I re-did it 5 times it now desyncs on the second step. (It is a very precise part which requires frame precision timing) Any idea how to fix this?
Joined: 10/3/2005
Posts: 1332
antd wrote:
I cannot get 2 steps out of the reactor. It desyncs when I leave the elevator. At first it would desync on the first step, after I re-did it 5 times it now desyncs on the second step. (It is a very precise part which requires frame precision timing) Any idea how to fix this?
I had an idea for debugging desyncs that I was going to use in my port of FBA. Haven't tried it, but I keep coming back to it, so maybe someone'll try it and see if it works. :D if someone were to mod the emulator so that it dumps the entire memory state, every frame, while going through the desyncing part, and then compare the active memory in the movie against the dump, they could probably pinpoint the desync to a specific frame where the memory states mismatch. From there, the debugger could probably add more clues about why it happened and how to fix it.