Tool-assisted game movies
When human skills are just not enough

Feos / Ideas

<< Feos

Table of contents [expand all] [collapse all]


Everybody knows that when you have unlimited time to ponder solutions, you can end up acting like a super-human. But do we all keep in mind that being a human isn't anything bad by itself? The wise balance around this borderline wasn't quite found yet, but many people out there notice that sometimes it looks too superhuman. Which is distracting for them and in terms of enjoyment falls into sub-human spirit (like, some kind of autopsist broke the game apart with no respect to anything).

True, when the viewer is a good gamer, he has a certain kind of knowledge how the games work, and how they can be played by a well-skilled player in real time. And what's exciting for most gamers is when a real person acts above their knowledge (the deeper that knowledge, the more impressed the viewer, as long as what they see is still above their knowledge).

In my opinion, a good TAS must include both fashions: the superhuman precision and the real human touch. All actions must be distinguishable in real time and not look too overloaded. The ideal impression from the fashion must be exactly like from a real person playing, just on an unexpectedly high level. To put that touch into his run, one must make himself maximally familiar with all real-time strats the game allows, since examining the tricks on a TAS level is necessary by default already.

Basically, the ideas you get must come to you when you play in real time yourself, exploiting all kinds of strategies for every possible segment. Then these ideas, after being pulled out with tools, would look suitable for a real time view. We were trying to add a superhuman touch so hard we almost erased the human one (in the finest sense). We must abstract from restrictions real time puts onto our mind abilities, but not from the very soul inside it that encourages the viewer as well as the real-time player of the game we TAS.

That's why I'm sure good TASes must be done by huge teams. Not only to optimize the hell out of the engine, but also to compare the look with what regular players see and fix it when it looks too superhuman, which turns into inhuman and must be avoided.



Watch how many views your homepage and subpages got.


  • Simple scripted buttons to add wiki markup of common types by selection
  • Fast Basic menu, optional Advanced menu
  • Drop-down module list
    • "Add all/no parameters" option
    • Mark favorite modules to show before further expanding
    • Hide unmarked under "More..." unless none was marked

Subscribe on Wiki edits

  • Subscribe on Wiki edits for a certain page
  • Receive e-mail notifications, linking to the diff page, considering the edit time
  • Minor edits don't matter


Each page that has translations must store them as a special kind of subpages, so that once you put module:traslations (which doesn't exist yet), it gives links to translations in the same fashion module:listsubpages works. Translations must be officially encouraged and can be done on language forums. Hint for translators: View the page source and post translations of it.

Add a Translator role with probable score and some privileges.

Fixing links to moved pages

  • Once a page is moved, TasVideoAgent updates all links leading to the old page
  • If some page gets obsoleted by another one without moving, there must be an option to mass-edit links in the referrers (only for Vested Editors).


If a page uses all types of headers, it becomes hard to distinguish their types. Make them differ from each other more. 4th header must allow vertical indent as lower types do.

Bug: 4th header isn't recognized by TOC module

Table of contents

Include options to expand/collapse by default

Bug: If a page uses only 3rd and 1st headers, TOC module gets broken.

Add expandable spoilers

These are imitated by Hide/Show tabs, which is wrong. Also option to expand/collapse by default.

Links to tabs

Just like there are links to chapters, we need links to specific tabs that would open them for us.

File base

User files

  • Every user can host movies, address lists and scripts
    • Scripts and lists can be viewed and embedded
    • Good syntax highlighters (Lua, AviSynth, Asm (various types))
  • Common - mark to include into Subversion, makes your files editable by other users.
  • Trusted users can host avatars and related pictures (non-common)
  • Filebase is public and inspectable
  • One user's files can't be edited by other users, unless marked as common
  • Hosting inappropriate content results in restrictions

Subversion system

Every common file hosted on the site can be downloaded and maintained in a Subversion fashion.
  • Non-common files aren't liable, can only be observed on the site directly
  • Read-only access to publication files (separate checkout link)
  • Files are sorted by folders, which are virtual on the site side
    • Sorting system must be discussed
    • Probably allow different checkout links, changing the tree (by file type, by uploader, by game)


Vault/Normal tier gets decided before publishing according to the submission votes. After the publication gets SOME amount of rates, it gets retiered according to the entertainment ratings.

Moons mean the runs that WOULD become Stars if their currently starred neighbors didn't exist. The runs the people must see right after they get familiar with the Stars.

Add Demo tier.


  • Stars - Best for entertainment, with strong tech value .
  • Moons - Really good movies, entertainment and tech.
  • Middle tier - Average content, not notable.
  • Demo - What was considered a pure tech achievement, or a pure entertainment one, but has little to do with standard TAS concepts.
  • Vault - Technical value, speed records, little to no entertainment value.

Movie rating

Rating is not just a tribute. It must be a system to maintain your own list of favorites/outsiders, highlight them on your page, and help the site to determine better movies. And as you are getting more experienced in TAS, your old impressions may change, so adopt a habit to revise your own ratings once in a while.

Force updating of average rating every time rates change.


  • Change voting criteria (divide feelings and knowledge)
  • Reduce the amount of options (too much precision in inconcrete things is pointless)
  • Use words instead of numbers (no need to invent your own meaning for each number)
  • "How well do you know this game?" - decimals imitation, required:
    • See first time - 0.0
    • I saw it beaten - 0.3
    • I played it myself - 0.7
    • I beat it myself - 0.9

Joy (your personal enjoyment: game, run, style, etc)

  1. Can't watch
  2. Totally boring
  3. Average
  4. Quite somethig
  5. Awesome
  6. Beyond any words

Effort (approximate author's effort on: optimization, entertainment, investigation, etc)

  1. I see nothing
  2. Minimal
  3. Quite some work
  4. Really impressive
  5. Insane effort
  6. Beyond my understanding


  • Pop-up menu to rate a movie
  • 3- raters results are not displayed, but can be viewed
  • Link to personal ratings
  • Active raters are listed in the last tab


  • Only option is to hide your name
    • Hiding the rating and not the name results in personal questions
    • Hiding both results in no feedback for the author
  • Option to formulate your reasons (maybe obligatory for rating): greatly helps the author to improve (anonymous rating + concrete response = effective system). Visibility of this text must be discussed:
    • All viewable
    • Rater decides
    • Author decides


Rates for Joy and Effort get weighted separately.


  • Default - 0.5
  • Player - average Joy rating of his own current movies (or of his best movie)


  • Default - 0.5
  • Lurker - 0.2
  • Newbie - 0.3
  • Player - average Effort rating of his own current movies (or of his best movie)
  • Member - must depend on post ratio

Better rating guidelines

If you don't know the game, and are going to rate its TAS, it's highly advised to watch how it is beaten in real time (or try beating it yourself). TAS is superplay, and you need to have the idea of normal play to enjoy TAS.

While watching in order to rate, examine the following factors (viewer must be able to know, what factors exist at all; not having such a list, he remains unsteady in what to rely on).


  • How much is going on
  • How fast-paced the game is
  • How irksome, monotonous the gameplay is (kills long runs)
  • How cool the graphics/music is (the game's spirit, making it enjoyable)
  • How much work was put into creating entertainment (how much the author cared)
  • How much/deep glitches are used
  • WTF factor
  • How much players were used (how they interact)
  • How superhuman does it look overall
  • Is there something annoying, and how much


  • How many rerecords were used
  • How long/in-depth the submission text is
  • Subpixels matter
  • Is the game overtased
  • How deeply was the run planned (life, item, character management)
  • How much players were used (how they interact)
  • How good were the tools available/used (fceu .12 itself was hard to TAS on)
  • Was the code/RNG studied/exploited
  • How superhuman does it look overall

Player points

Must consider game diversity.


Publishing form calculates frame diff

Once a publisher picks the obsoleted movie, not just text and classes are copied.
  • Count the frame diff
  • Add a frame module
  • Fill in the FPS of the platform equal to what the site uses

Tier and obsoletion selected by a judge

  • What was chosen by a judge can't be edited by a publisher
  • Default list of movies to obsolete (same platform)
  • Option to add more movies to obsolete
  • Options to show all platforms
  • Tier list

Publication page improvement

  • Share buttons show on click only
  • Add links to other movies:
    • Of this exact game
      • All current branches
      • All publications
    • From that author
    • From that franchise
    • From that platform
  • Explanation tab is shown only to non-logged in, Lurkers and Newbies
  • YouTube views and likes get polled
    • Token builder allows sorting by that data


Keep as many different branches as possible. For different goals or time/entertainment difficulty, the audience determines the decision.


The goal must be represented in the publication branch, do not skip the its name for fastest branches (prevents misconception when user can't find the branch he wants because it's skipped).

Gameplay-based (showcase the ways to play):

  • Fastest completion
  • Maximum completion
    • Best ending
    • All levels/Warpless
    • All items
  • Side branches
    • By route
    • By character
    • By item
  • Playaround/Glitchfest
  • Various exceptions

Tech achievemnts:

  • Glitched (gameplay is deeply broken/skipped)
  • Botted (most of the run was brute-forced)
  • Demo (tool/method showcases)
    • Multi-TAS
    • Script showcase
    • Esoteric TASing method (like these two)
    • Other fun concepts


  • Same branch and platform
    • Faster
    • More entertaining
  • Different branch
    • Too similar to the existing run, which looks worse


  • Same branch and platform
    • Slower
    • Less entertaining
  • Cross-branch obseletion
    • Too similar to the existing run, which looks better


Verify and fix the game name as it is written at MobyGames or FameFAQs.

Fill in emulator field if empty.

Select tier and obsoletion.


Post voting

Each post can be voted for (+/-). Audience can be reliable on how sane someone's posts are. System example:
  • Lurkers and Newbies can't vote
  • Unvoted posts don't matter
  • Subtract total negative votes from positive. Divide the result by voted post amount.
  200 -40 = 160
  160 / 100 = 1.6 (positive ratio, posts are helpful)
40 -200 = -160 -160 / 100 = -1.6 (negative ratio, posts are annoying/stupid)
  • Post sanity ratio can be used to weight movie rates, calculate Member score, replace post number with post ratio, etc.

Combined RSS Feed
Feos/Ideas last edited by feos on 2018-04-08 08:44:30
Page info and history | Latest diff | List referrers | View Source