Posts for r57shell

1 2
11 12 13 14 15 16
Posted: 10/24/2014 7:55 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Forgive me, but I'll repeat again what I'm thinking about any%/100% tier and side goals tier. I don't know why people don't understand what I was talking before, there may be many reasons: 1) They "DIDN'T READ!!" haha. I mean, they don't like so long posts... 2) They simply don't accept other points of view 3) They ignore my words because I was laughing on this idea. type your option here But still, let's compare system "NO TIERS": 1) All movies in one list 2) There can be many tags: uses death to save time, pacifist, puzzle, platformer and so on. 3) Judge decision STILL depends on what category run belongs. (any%/100% or side goals). There is still active following rule: if run is not aimed for fastest speed, then it will be rejected without enough of support (entertainment) 4) Is tier affects judge's decisions? No there is no tiers. 5) Is tier selection affected by judge's decisions? No there is no tiers. 6) There is no tier appearance, because there is no tiers. Captian obvious helps a lot. with system "any%/100% tier and side goals tier": 1) All movies in two tiers (lists) 2) same as in "NO TIERS" 3) same as in "NO TIERS" 4) Is tier affects judge's decisions? No, because of (3) 5) Is tier selection affected by judge's decisions? No, because only fool can't understand is it aim for fastest completion (any%/100% tier) or is it something else (side goal tier) 6) There is tiers icons, along with publications. So, if you really accept that (2), (3), (4), (5) is same in this two given systems, then we have to deal with (1), and (6) only. 1) One list vs Separate two lists. Let's find out, where they separated? I know only two possibilities: a) in our minds, b) in search response. Take a look at (a) variant. I think, you don't need tier to separate in mind speedruns vs non-speedruns. Do we need tier for that? I think NO. If you need tier to compensate your foolness: go away :D Now, take a look at (b) variant. Do we need tier for searching speedrun TAS-es? It would help to find, or cut off search results. But there are tags already! They made exactly for that purpose: searching. and cut off unsuitable results. Deal with (6) 6) "No icons" vs "Icons". Icon purpose is: additional information. It's mark, it's showing some information. If "No icon" then no additional information. If there is icon, then it gives us some information. In this case, icon will give us information "that it is speedrun" or "that is side goal".It might be useful yes. Only fool can't distinguish speedruns among "side goal runs". But imagine, I want such icons: (speedrun/side goal). And I'm crazy about that! Then why not add tag "speedrun", and add tag "side goal" and make icons for them? Or... may be I really so excited, and I want icon for "pacifist" run? And for puzzles? And for "arbitrary code execution"! Make icons! I want them! It will be AWESOME feature! Conclusion: "NO TIERS" vs "any%/100% tier and side goals tier" 1) One game list vs Two game lists. 2) No icons vs Two icons. get "NO TIERS" system, add tags "any%/100%" and "side goals", and make icons for them. You'll get SAME result, as if you make two DIFFERENT tiers. What's a point to make them Tiers?! I don't get it Now, compare "NO TIERS" vs "stars, moons, valut" system. If you not retarded... then (2), (3), (4) are same! So we have to compare only (1), (5), (6) 1) One game list vs Three game lists. Question that I was asking for this case was: "where they separated?". Do you remember what my answers was? Haha! a) they separated in our minds. Is that true? May be so. Is tiers helps to make this separation? I don't know b) in search response. Here is not same situation as in "any%/100% and side goals" tiers. Vault is kinda recycle bin, but without cleaning option :D. And vault movies are hidden. (you have to go on other page, to look on all TAS including vault ones). So, for me, it's actually two lists: vissible by default, hidden. This is why I don't like vault tier as it is right now, because it's not equal to other tiers: it's hidden by default, and it has bad tone of name. Positive change would be: name it at least with neutral name, and DON'T HIDE IT as some shit. What if vault would be visible, and with at least neutral name? All games would be in three lists: stars, moons, vault(but with different name, and visible), then again... Where they are separated? In minds? You don't need tier to make decision yourself. In search list: yes, separated. 5) Is tier selection affected by judge's decisions? In "NO TIERS" there is no tiers, so nothing affected. In "stars, moons, vault" system, choice of tier is given to judge. He will select appropriate tier according to community feedback and support. Decisions of judges are subjective because there is no strict rules about making decisions. Hello borderline cases :). Also, "stars, moons, vault" tiers purpose is reflection of community feedback and support. I would call it raiting. So, in other words, it reflecting rating of TAS run. In "NO TIERS" system this information is ommited, same situation in "any%/100% and side goals" tiers system. But yes, you can always check raiting in "rating" section and in discussion topic. 6) "No icons" vs "Icons". As in previous comparison of systems "NO TIERS" vs "any%/100% and side goals" Icons may be achieved with adding tags, and bindings icons for them. Conclusion: Idea to change "stars, moons, vault" into "any%/100% and side goals" looks for me like removing feedback and support reflection with simple reason: borderline cases. And that's why I laugh so much about it, because you want to solve problem with simple deletion of part of system that has issue. It's like cut off hand instead of its treatment. And, I'll repeat agian, "NO TIERS" is same as "any%/100% and side goals" by its fuctions provided to people. (except icons, they can be added with tags) LOL they are so excited by cutting off hand (removing rating reflection because of borderline cases) A little about poll... Do you know so called supermajority or a qualified majority? It's what got more than or equal to 50% of all votes. What's a purpose of term? Lets say, you got "YES" result in poll with maximum percentange, and it's lower than 50%. You may think: maximum votes got YES option, so "YES" won. But as I said, if it's lower than 50% then all other votes in summary will give you more than 50%. And with selection of YES, you discard all other votes, and their count is actually more than count of those who choose YES option. That's why "supermajority" term exist. In other words... If "yes" win with <50% then it's not win, it's defeat, because all other options are "no" with different flavors. And, in the end... My vote in this poll is: "Keep the Vault tier but change its appearance (name/icon/etc)" What exactly? Remove vault icon, and name. Let vault movie looks like normal one without "moons" or "stars". I hope, it's my last post in this thread. Thanks for reading to those who read it all Hmm, actually, it would be nice if someone who agree with me, don't keep silent, and said that, because many people just don't wanna repeat what other people said, so they don't post anything. That's why I would like to see if anyone agree with me. If noone, that's sad for me :C
Posted: 10/20/2014 8:08 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
How rerecords changed in TASEditor in FCEUX? And, How rerecords changed in Tas Studio in BizHawk? Is it incremented for each key edit? Or only for running after edition if it was edited? I think if rerecords count incremented with each keys action in such tool it will generate high rerecords count. But, I don't mind, because it's hard to track rerecords. There is so many tools available. For example, what I should do when I'm mixing two different movies into one? :) Should I summarize them? (A+B)
Posted: 10/17/2014 9:50 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
feos wrote:
It's about how less if's != better.
Nope, your question was: "is it simple". I said "not". I didn't say anything about what is better. Anyway, if all other criteria are same then less if's = better.
feos wrote:
We tried to define criteria to publish "interesting" runs. It can't be defined.
Simple criteria: poll "should it be published". I don't remember already, did I say anything about polls in Workbench. In my viewing of things, there would be only one question in poll: "should it be published", and entertainment rating votes after it is published. Hmm, actually you can accept entertainment rating votes for any of submissions, even not publicated, there shouldn't be a problem to make it available.
Posted: 10/17/2014 8:58 PM
(Edited: 10/17/2014 9:40 PM)
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
You mixing up together interesting and entertainment. It's not the same things. Entertainment means = FUN, interesting means - you have interest, you wanna know something, or wanna watch something, because it's interesting, not because it's FUN or enjoyable. EDIT:
feos wrote:
r57shell wrote:
feos wrote:
Pretty simple, huh?
I don't think so. So many if and then.
- Sorry, your program contains too many if's. Cut a few and it will be perfect.
Is it argument or what? Anyway, with each additional "if" things become more complicated.
feos wrote:
We tried.
"We tried". Tried to make tier? Did I said that you suppose to add tier? Why not just be able to accept "exceptions"? If they really bad, yes, you should reject it, as you always do.
feos wrote:
r57shell wrote:
How could speed records be "ideal"?
It's what Stars were meaning for all these years: http://tasvideos.org/Stars.html
The TAS itself must be perceived as of highest quality.
"Highest", and "ideal" is different for me. Highest means: highest among something... "ideal" means - no imperfection. So if you mean star - highest quality (not ideal) - then that scheme looks fine.
Posted: 10/17/2014 7:53 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
feos wrote:
Pretty simple, huh?
I don't think so. So many if and then.
Warp wrote:
r57shell wrote:
Warp wrote:
The "all games deserve a TAS" principle is reserved for completion time only (with any% and 100% being the two possible goals).
Rly? Then it's really bad. Don't ask why. I don't know, it's complicated question. If run was not accepted only because it's not any% or 100%, and it's interesting (but not entertainment, because it's vault), and it was rejected ONLY because it's not any% or 100%, then it's silly.
If it's interesting, then it gets published into the Moon tier. If no alternative goal is interesting enough, the game can still have a completion-time TAS. What exactly is the problem? Do you really want to accept any arbitrary goal for every possible game, inundating the site?
It won't be published into moons, because it's not entertainment, and it's not speed run -> so it rejected. Yes I would accept any arbitrary goal for every possible game, if it's interesting, and high quality (very optimized... may be some more technical requirements). But, indeed, only if it does makes any sense. I don't care about inundating this site. If you look into not obsoleted runs - there are not so many of them.
Warp wrote:
Warp wrote:
Edit: Which gives me the idea: Change the tiers to "the any%/100% tier" and "the everything else tier". (I don't have good ideas for the actual names.)
lol
What's that supposed to mean?
It means, that it makes me laugh. For what it is intended? How it helps? For searching? I can open list and CTRL+F any% for example, or somehow better. If it is for searching, then best way to remove tiers at all, and add tags "any%", "100%", "not speed attack" something like this. When you have tag system, you don't care how many tags you may add, and there already enough of them: "takes damage to save time", "Pacifist version". Why tiers then? Your tiers system cause following result: tiers only for searching, and very easy to find speed runs, and very easy to find all other. Very easy to select right tier. What's a point? Looks like suggestions about new search tags. What Stars, Moons, (nothing)/coins tags (call it tiers if you want) are doing for us? It is some kind of rating (or rating itself). Rating and search tag - completely different. Sugguested tag "New" supposed for that period while rating is not clear.
feos wrote:
        (Stars)
     ideal anything
     /         \
(Coins)      (Moons)
 speed     entertaining
records     side goals
How could speed records be "ideal"? They could be all ideal from position: it's current record, no one beat it yet. They could be all not ideal from position: you can't prove that it's done in fastest way. So, either coins removed at all, or speed records can't be stars. If you allow speed records to be in both (stars/coins) you'll get another borderline cases branch :D. Everyone would ask: why my speed record is not ideal?
feos wrote:
Is this not fleshed out? It handles everything exactly as we do now, but removes all the hardship we're having. In other words, elaborate.
check reply to Warp in this post.
Posted: 10/17/2014 12:42 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
feos wrote:
First of all, the main problems always was and still is with borderline cases.
If you think, that main problem is borderline cases, then how you suppose to solve it? Do you suppose to find solution for that cases? I don't know such solution, and can't even imagine that. There are borderline cases only while there IS BORDERLINE. So one of solutions is: remove borderline Say strictly: Moon in this cases, and Star in this cases... Honestly, I don't mind is it star, or moon, but I don't agree with vault at all, because, but about it a little after...
feos wrote:
Those are what makes the Vault look bad - runs that don't look deadly bad get placed there.
Let's make then tier: trash, for thos "deadly bad" runs. But for me, any technical good run must be accepted, and it should not called trash nor vault.
feos wrote:
RATE ME makes more sense, because it can still have little to no ratings after a year, when it's in no way NEW anymore.
RATE ME makes more sense yes, but if you think that two months old run is really old, then I can't say anything. I don't think that there are so many people who every day checking new publicaitons. If you will keep such runs with NEW tag, people who look once in two month would be able to check and watch them.
feos wrote:
The only 2 problems that are possible here are: - ambiguity of what is ambiguous (how do we know it needs to go to the undefined tier?) and - should that tier be used to accept unvaultable runs of uncertain entertainment value (pretending they will be Moons, to justify acceptance)?
I don't like sophisticated rules, they looks ugly, and it's difficult to understand them and apply them right. For me, any new run must go into undefined tier in your terms. Honestly I don't understand second sentence so I can't comment that :D. But I say it once again, all technicaly good runs should be accepted, except those who does not satisfy other requirements such as "Complete Whole Game".
Warp wrote:
Why should the vault tier need to feel like just a dump to throw the "garbage TASes" into?
Because it's sounds like it. Look at this: crypt, tomb, grave, mausoleum, cavern... Garbage is not synonym, but it's some near by. For me vault has association "garbage". If you calling something "turd", would you claim that you didn't mean "shit"?
Warp wrote:
The "all games deserve a TAS" principle is reserved for completion time only (with any% and 100% being the two possible goals).
Rly? Then it's really bad. Don't ask why. I don't know, it's complicated question. If run was not accepted only because it's not any% or 100%, and it's interesting (but not entertainment, because it's vault), and it was rejected ONLY because it's not any% or 100%, then it's silly.
Warp wrote:
Edit: Which gives me the idea: Change the tiers to "the any%/100% tier" and "the everything else tier". (I don't have good ideas for the actual names.)
lol
Pokota wrote:
I guess this kinda goes along with r57shell said, though I would point out that Gruefood is the loser tier when not deliberately sought for (such as with the April Fools' submissions).
Yes I'm not againts good names, which means like it's worth something, that it's not trash. Also, there is coin in mario :)
Posted: 10/16/2014 4:52 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
This is what I want to post into discussion about #4452: dunnius's NES Hydlide in 05:11.62 Thread #15869: #4452: dunnius's NES Hydlide in 05:11.62
We need an appeal to a higher court! :D Actually, I think... First, we need to understand goals. Second, we need to locate issue, and find out where its stink comes from. Goals: as far as I understand, main goal is entertainment. Second goal is history. Star/Moon means that it's high entertainment. Vault category supposed to be as history of top records for particular game. Issue: I don't know exactly what issue you see, but I see this: If run going in Vault, it marked as "not entertainment", literally: "don't watch it, it's only waste of time", and as result: less votes about ratings, and so on. More than that, vault runs are hidden in list by default. With runs in Moons/Stars category, completely different situation, they marked "this run cool, watch it", and result: more votes about ratings, and so on. And they are visible in list by default. Where is issue here? Good runs forces its ratings to be good, and not entertainment runs forces its ratings to be bad. In other words, it's not good that rating is affected by tier.
Why I didn't post that? It actually doesn't represent actual situation. There isn't enough votes in ratings to make statistical conclusions. But now, I'm posting that, to reveal my way of thinking. We need more votes to be able to work with staticsics. It would be nice to force somehow people to vote. For example, me, I didn't ever vote for publicated movies, actually, until some moment, I didn't even know that it's possible. It's hidden somewhere in list of links, and it's not noticeable enough. Vote input must SCREAM! Vote me! Do it! Now! But instead of it, you have to find yourself where it is... then go on another page, with vote form. Also, why it's showing only entertainment rating by default? Why not show both? In Workbench situation completely different: you have encode, and you have poll. All in one page. May be that's why I vote so much in Workbench. For me, vault is like "black mark". All going bad after you got it. :D I can't vote in this poll, because there is no such option. Idea is simple: why I, myself, don't have tier: vault? Some people has rewards like "tas of 2013" and so on, why I don't have reward "He don't have reward, he looser." For me, moons, and star - must be like rewards, like tags. And all runs that is not moon or star, must be clean, without tier. It does not mean that it is bad, it means that it's not cool enough to be Moons. It's my opinion what needs to be "at least". Now, what I would like to see in best case: 1) No vault: as I said, just clean, without tiers. 2) Auto moons/star: based on rating. It must be not tier in my view of things. It must be like tag. Moons meaning will be "it's more than X by rating, but less than Y by rating". Star: "It's more than Y by rating" where Y > than X. And searching for "stars" will be equal to looking for publications with rating above some value. 3) Tag "New": means that it was publicated recently. Period can be chosen as month or two, or week, I don't know, depends on voting speed. Then, after selected period, tag would be selected automaticaly (moon or star). In this system judges duty would be to check technical part, accept for publication, select branch, in other words: manage submissions. But it won't work unless you have enough votes. Where is my "looser tier"? :D
Posted: 10/16/2014 3:39 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Programming knowledge is not required. Math knowledge is not required. RAM addresses is not required. But all this things can boost a lot! First is RAM addresses: very easy to understand how to find them. When you have RAM values, you can easily compare positions/speeds (if you found them). It's much easier to deal with exact values of position/speed instead of trying to find out it from actual image (screenshot). You always have to compare values, because there is no other way to find out which input is better. Second: a little of math... It helps, for example, if you understand how velocity affects position in time. I don't know how to describe... For example, in platformer, if some door opens after some timeout, you may accelerate before door, and enter in it right at that moment when it opens, instead of holding arrows in direction of door. Most of games in this case will clear your speed, and when door will opened, you'll start to gain speed, instead of running at top speed already. May be math - no so cool boost, but for me - it helps. Programming knowledge: for me, programming is developing of architecture for some program. Writing code for me is: coding :D. You may easily find out how to write Lua scripts looking in existed ones, and changing them. (as feos mentioned) Reverse Engeneering: it will help you find out algorithms how game works, in any sophisticated cases, such as AI decisions, RNG manipulations and so on. Here you need to know particular assembler, and programming skills. With using all mentioned before things, you can make very high quality TAS. But, even if you have great Lua scripts, and whole RAM map, but lack of attempts to improve that you already reach, then you'll end up with mid quality TAS. Also, if some kid will find better route, he can beat your TAS easily even without such skills and tools. But if he beat your TAS, and he did't make it on your level, then you may "easily" beat his TAS. Last thing: algorithms. In some cases, you may optimize some actions in total perfect way, for example subpixel positioning while jumping in some platformer. I wrote program to achieve that during one jump for game Donald In Maui Mallard.
ALAKTORN wrote:
He creates bots that make TASes, that’s TASing at the highest level.
If you rely on bot, your bot can't check all possibilities, so you have to watch after your bot, how he is doing things, and for all sub-optimal decisions, alter your bot to take into account that certain cases. So all that cases alteration looks like you was editing input by your hands. For some games it's possible to make bots without such tweaks: puzzles for example. And yes, for puzzles - best way is solving program (bot).
Posted: 10/14/2014 7:17 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Если бизхок это будущее, то у меня нет будущего :D
Archanfel wrote:
Просто в отдаленной перспективе поглядываю в сторону возможности улучшения этого замечательного мувика если никто другой так и не возьмется.
А есть куда улучшить? :O
Posted: 10/14/2014 2:49 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
особенность в том, что она будет. (тас студия) как бы смешно это не звучало ).
Posted: 10/14/2014 4:07 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
А я пока не доделаю дональда - ничё не начну. А вообще да, странно, не любят наверно его.
Posted: 10/12/2014 9:09 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
WST wrote:
Also, being only 43 frames behind an expert player is really good start, I think.
It's within first level only.
Posted: 10/11/2014 7:56 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
WST wrote:
Тем более, эта тема, посвящённая любым вопросам по Сонику, не была создана вчера, как ты, вероятно, подумал — обрати внимание на дату первого сообщения.
:O Это черная магия! Или ты некропостер! :D
Posted: 10/11/2014 4:43 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Вообще странно ты сделал, вынес один вопрос в целую тему, вместо ответа к примеру в личку. А сразу после того как вопрос по сути был закрыт (а он был закрыт у меня сразу после первого же твоего поста), говоришь что не хочешь что-то обсуждать. Я понял позицию, я доволен ответом, всё разумно, и отдельный бранч из-за 3-х зон импрува нет смысла открывать. Если когда-нибудь доберусь до соника, то обязательно сделаю TAS.
Posted: 10/11/2014 10:31 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Sonic 3 and Knuckless = Sonic 3 & Sonic Knuckless Sonic Knuckless - продолжение: дополнительный персонаж + уровни. Правда если играть в Sonic Knuckless без Sonic 3, то там нет Tails.
Posted: 10/11/2014 9:37 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
http://coub.com/view/3ju3h
Posted: 10/11/2014 6:38 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
TAS not need quality! Push by quantity! LOL
Posted: 10/11/2014 6:36 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
WST wrote:
Скажем, во втором Сонике в актах, в которых достигнуто время от 0:24 до 0:29, для скорейшего наступления следующего акта нужно просто подождать до наступления 0:30 и прикоснуться к флагу только после этого.-
То есть всё же есть случаи в которых можно ускорить время. А почему бы не прийти к компромиссному варианту? Для этого нужно просто придумать критерий оценки времени когда был закончен уровень не основываясь на флажке, тем самым ты сможешь оценить как быстро он был пройден, а потом можешь ждать до флажка. другое дело в том, что если все уровни будут пройдены быстрее чем за 23 секунды, такой подход уже не понадобится.
Posted: 10/11/2014 6:23 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
I haven't seen this game before, and I haven't seen previous run, but it looks entertainment for me, because: he complete game fast, he killing monsters like a crazy, he avoid stuff as super human if needs, he predict everything, and there is no idle moments, may be only killing stuff while awaiting something.
Posted: 10/10/2014 3:07 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Now I understand, somehow I didn't think that if it's not vault, then it moons, but it's not moons so it's vault.
Posted: 10/10/2014 1:47 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
btw, someone can explain why tier vault?
Posted: 10/10/2014 10:43 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Fix "Special Thanks" Zeropass author is Organic.
Posted: 10/7/2014 2:35 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
WST, это не лично мой сайт, это наш сайт, чей - в подписи в самом низу. Тебе в раздел HandMade. Вообще, подойдёт любая память с откликом меньше 150 us (нано секунд), только желательно стандарт под 5v, а то сейчас 3.3v (или сколько там?) в моде. Картридж сеговский - тупо кусок памяти. feos, тыкай полностью! :D А если натыкаешь? всё? Rejected? :D
Posted: 10/7/2014 8:15 AM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Ну, знаешь, тяп ляп пробежал 100 ререкордов за весь мувик, и тыкание мыкание во все углы втыкание, - тоже не дело. Надо всё равно иметь какую-то планку. У меня вот такая планка, если она достигнута - вот это тас, вот это качество, к примеру глянтьте WIP по Batman & Robin в секции по Sega Megadrive. А вообще да, можно запилить только роут, а потом чтобы оптимизатор пришел, и оптимизировал на максимум. Но, лично я придерживаюсь мнения, что оптимизатор или не придёт вообще, или ждать придётся долго, как с привидённым тобой мувиком. Там да, инструменты были другие, потому и можно улучшить. Но с другой стороны, где это улучшение? Остаётся только верить и надеяться что кто-то придёт и улучшит. Не смотря на популярность соника, этот забег всё ещё не побили. И ещё, ты так говоришь, будто надо чтобы улучшали, а по мне, если не улучшают - признак качества. Как ни крути, оптимальный пробег - не улучшишь, исходя из определения оптимальности.
Posted: 10/6/2014 11:19 PM
Quote
r57shell
Experienced Forum User, Published Author, Player (97)
Joined: 12/12/2013
Posts: 376
Location: Russia
Там в обсуждении написали очень много, и переводить/обсуждать я всё то не хочу. Вообще у меня есть очень грубая оценка TAS-ов: ререкордов должно быть минимум в два раза больше чем инпута. Но это грубо. Это не обязательно, просто смысл в том, что если ререкордов мало - значит мало пробовал. Сам же должен понимать что малое количество ререкордов только если ты несколько раз пробовал какой-то момент, например на slowmotion. Качественные TAS делаются на framestep. То есть ты взял остановился на каком-то кадре, сохранился, и ты пробуешь по разному какие-нибудь 60 кадров к примеру, по несколько раз, даже если ты их попробовал 10 раз (10 раз это мало, сам подумай) то это уже 1/6 плотность (1 ререкорд на 6 кадров), 2 ререкорда на 1 кадр это к примеру 20 попыток на 10 кадров - когда сидишь, тестируешь, не замечаешь как быстро делаешь 20+ попыток. Это по поводу очень грубой оценки. (я не смотрел сколько там у тебя ререкордов) Второе: роуты. Route - путь, в общем как ты проходишь. Надо тщательно продумывать путь по которому ты будешь проходить: планировать что из итемов тебе надо поднять, сравнивать несколько разных путей грубым TAS-ом - какой из них быстрее. Если же разница получается мелкой, увы, придётся TAS-ить тщательно, и снова сравнивать. К примеру по одному пути быстрее на 10 фреймов, но так как ты TAS-ил хламово, возможно ты просто где-то потерял 20 кадров в том который медленнее показался, а значит он может быть быстрее. Второй пример: первый раз пробежал собрал дополнительную аптечку - потом выяснилось что она не понадобится - пробегай заново без аптечки ). Как бы хорошо ни был оптимизирован TAS, если путь выбран коряво - любой школьник придёт и перебьёт твой TAS, и будешь ты obsoleted :D. Третье: утилиты. TAS без утилит это как-то не серьёзно, я так считаю. На одних slowmotion и save/load (даже с taseditor) не выжмешь максимум, получится "на глазок". Получается так: если видишь что ты ткнулся в стену - значит ткнулся, если не видно, то не ткнулся, наверно. Как "в потёмках". К примеру, первое, что я лично бы сделал, это нашел бы координаты героя в памяти игры. У тебя есть оффсет в RAM на позицию X,Y? Вот даже если ты её найдёшь, сразу будет проще и качественнее - ты сможешь понимать где стены (смотреть ограничения на x,y, правда это сложновато но можно), и оптимизировать на основе этого. В идеале, я бы порылся, и нашел: позицию, скорость, направление (куда смотришь), стены, и потом ещё до кучи карту бы вывел в виде: линии - стены, крестик - персонаж, а потом чтобы не было такого: И если ты приведёшь lua скрипт который будет выводить стены, и позицию персонажа (стены - линии, позиция - крестик), то не будет таких вопросов. Особенно если ты ещё и траекторию выведешь. По поводу оптимальных траекторий: если даже есть у персонажа радиус поворота, то зная стены, и имея модель этого поворота, можно оптимизировать. Вот к примеру как перейти от точки до точки с радиусом поворота: http://www.sid337.newmail.ru/1340.html Это я к чему? Если есть возможность двигаться по дугам - почему нет? Разогнался и вошел в поворот под нужным углом ). Опять же, если путь не оптимальный - нет смысла пыхтеть над координатами, и прочими техническими оптимизациями, но никто не знает оптимален ли выбранный путь, до того как его улучшат. Так что, ответь на несколько вопросов: 1) Проверял ли ты "все" пути прохождения? (все мыслимые) 2) Как именно ты делал попытки? Framestep от угла до угла? (от checkpoint до checkpoint?) Checkpoint - это место где ты что-то должен сделать если следовать выбранному пути. 3) Пользовался ли ты RAM Watch, Lua scripts, и прочими техническими хитростями?
1 2
11 12 13 14 15 16