Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Daniel, Lord Bahamut wrote:
feos wrote:
The RGB version looks broken in my VLC. It has strange green color.
Same. VLC, MPC-HC, and Windows Media Player all had green stuff, while SMplayer and KM Player were staticy. What codec was used?
I guess I should have been explicit about this, but RGB in h264 is very poorly supported. It is implemented as yuv444 + a custom color matrix specified in the file, but if the player does not support color matrices the colors will look wrong, of course. And to be honest, I don't know of any players that do support this (I used a specially patched version of mplayer2).
The symptoms of not understanding the color matrix is that things look mostly green. If things look mostly gray, that is an indication that the player does not support yuv444 either.
I am uploading a more compatible yuv420-version here now. It should be done in about an hour.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
22 Samuses frolic around Zebes:
Link to video
This is a comparison of all the Super Metroid submissions at TASVideos (except one which was rightly rejected for poor techincal quality). It also includes the RBO wip by Kriole, Taco and Herooftheday. It it interesting to see just how big an improvement Super Metroid has had through the years. The newest run gains 200 frames or so per room vs. the oldest one, for example.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Here is a comparison video between this run and its predecessors:
Link to video
And here is a lossless RGB version of the same:
http://folk.uio.no/sigurdkn/saturn_any_ingame_rgb.mkv
This is a great new run, and there are sizable improvements in almost every room (though I have not yet investigated whether the improvements that are not due to new techniques are due to movement optimization or different timing (i.e. lag management by Taco&Kriole)). However, there are some cases where time is lost, too:
When exiting Phantoon's room, 6 frames are lost, but 5 are then regained during the next room due to higher speed exiting the room, for a total of 1 frame lost. Was this in order to gather more drops from Phantoon?
When crossing the lava seahorse room at 8:14, you do not roll through the lava like Kriole&Taco do. This seems to cost about 29 frames. Did the Hi-Jump boots make the lower route too slow?
In the first room in lower norfair you lose 5 frames by having lower speed than Taco&Kriole, who start by going left to build up speed. Was this also due to Hi-Jump boots?
Anyway, this run was good (as most here seem to agree), and faster than its predecessors, so I vote yes.
Many posters here seem to be in a moral dilemma here: They think the run is good and want it to be published, but they think that would be bad for the site because we would get too many categories per game. I don't think this is anything to be torn about. We should have many more categories per game than we have - there are many interesting ways left to run every game. Of course, the problem here isn't that we will run out of categories (though you might get that impression from the way some are arguing) but that people are afraid that the most interesting runs will be swamped by more specialist/niche runs, thus lowering the signal-to-noise ratio of the site.
My answer to this is that firstly, that is what we have the star and moon system for. We use these to pick out the most noteworthy runs. I have previously suggested that we change the role of the moon symbol (or introduce a new symbol) for the most noteworthy (or possibly runs) for each game. The games list would then list this run prominently in that game's entry, with less prominent links to the pages of the other runs for that game. That way the games list would not be swamped by extra categories, but they would still be available for those who are interested in that game.
In many ways, I think the "popular categories will be swamped by niche categories" is exactly the same problem as "popular games will be swamped by niche games", and the solution to both is the same: Stars for games, and moons for categories within games.
By rejecting good runs like these and suggesting that they "should be 'published' on youtube" because they would give us too many categories, we are doing ourselves a disservice. Not only are we turning runs we ourselves admit are good away from the site, we are also reducing the incentive to submit here and in general promoting direct-to-youtube publishing as an alternative to TASVideos. It would be a tragedy if this were to become the norm, as putting the run up on youtube does not automatically ensure that the input file becomes and stays available. Consider, for example, the tragic loss of Saturn's Super Metroid Redesign any% TAS input file, and all the excellent Japanese TASes which are available only as video encodes on NicoNico douga.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
At 00:57 you run into a door, lose all your speed, and then start running again when the door opens. Couldn't you back off a bit and then start running just in time to go through the door at full speed when it opens?
By missed shots I mean shots that do not serve any purpose. For example, at 00:11 you shoot three shots to open the door. The two first hit the wall, and the last one hits the door. What was the purpose of the first two shots? A TAS is supposed to have perfect precision, so it shouldn't look like you missed the door. It is ok to shoot extra shots to entertain, but then those shots should be entertaining - they shouldn't just look sloppy.
You didn't answer question number #4. Unless I have misunderstood totally, comparing with console speedruns is difficult with dolphin because of loading time inaccuracies. What would work is to count frames in each room from the time the door opens to the time the next door is reached, both for you and for the WR video. That way loading time differences could be removed from the comparison.
The aiming for the targets at the beginning looks slow. Do you have to keep them in your sight for that long before shooting? Especially the last one. Is it to set yourself up for a scan-dash?
You have lots of missed shots. What are they for? Except for the place where you outline the door with shots, it looks very sloppy.
When going down corridors, you seem to adjust your direction a lot. Is it not possible to go in the right direction from the beginning? An example of this is the first corridor in the station.
How do you estimate how far ahead of the world record you are? Dolphin doesn't emulate loading times properly, so you have to take that into account.
Are you sure you are unmorphing with the right timing? The door at 1:19 looks like it would have been faster if you had unmorphed earlier.
I am not sure about the lock-on mechanics here, but at 1:23 it looks like it the lock-on temporarily misses to the right of the target before locking on. Could this have been done faster by looking more to the left before trying to lock on?
In general, I get the feeling that this was not fully frame-precise. But I have not seen a TAS of this game before, it might just be the game that is that way.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
I think we are making it too much of a barrier for Japanese TASers to participate here. There are many active Japanese TASers, but they seem to self-publish by uploading their videos to nicovideos only, and I'm not sure that they provide the input file anywhere. To them, the English of this site is a big barrier, but even more than that, our requirement that they use an unfamiliar (and usually not the original) version of the game for submissions is a big turn-off. Right now, the only Japanese submissions TASvideos gets are those that were made with TASvideos in mind to begin with. We have rare cases where peope are actually willing to remake their TAS from scratch to be able to submit here (see Secret of Mana), but most people wouldn't be willing to go that far.
I suggest that we change the rules to say that the original version of a game is preferred, but that due to the large amount of English speakers here, the American version has an equal position. When an obsoletion would change versions of the game, a subjective judgement would have to be made whether the new version is faster when ignoring differences like text scrolling speed etc.. Requiring a subjective judgement like this is a disadvantage, but we do most of this already.
Another possiblity would be to publish the different language versions in different branches, but that would require some changes to the publication system. Right now, we have four levels: Gruefood, published, moon and star. I'm not sure the difference between the current moon and star is that useful, so I suggest that we change the meaning of moon to be like a star, but within a single game. That way you can have a huge number of categories for a single game, but you would only moon the most interesting ones. For example, for Super Metroid we would probably moon the unglitched any% and 100%, while glitched any%, unglitched low%, RBO etc. would also be published, but with less visibility. Each published game would have at least one mooned run, and the current browse games listing would be changed to show only the mooned runs prominently (i.e. with pictures and full description), with text-only links to the unmooned runs.
The point of a system like the one above is that it would allow us to keep a general movies list high quality by showing only the most interesting categories prominently, but still make available the more specialized versions that experienced players of the games appreciate. As well as allowing several language versions of a game to exist - the one currently judged the most interesting (probably the fastest one) would be mooned.
Of course, we don't have to take over the moon symbol to do this either. We could also introduce a new level (a 1-up mushroom for example) for this purpose.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
I heard the explanation for this a long time ago, but I don't remember much of the details now. However, I think the reason is flux pinning, where small magnetic channels form through the superconductor, much like a pincushion with needles through it.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
People usually don't come out and explicitly state what they mean by free will, but the idea seems to be that people have some sort of "spirit" that decides their actions, but is independent from reality. I guess people who believe in this think the brain is some sort of interface device for this thing (implying that brain matter would not follow the laws of physics, since they don't include a mechanism for interacting with a "spirit").
So free will fits natrually in with religion and somethingism, but it isn't really something you would think atheists would believe in. Which is why it is so strange that perhaps a majority of the atheists I have discussed this with seem to believe in it.
Typical arguments in favor of free will appeal to the inconvenience that would follow if it did not exist: "If there is no free will, people cannot be held responsible for their actions, and all punishment will be unjust!", for example. Or put a bit differently, they are saying that "I subscribe to a system of ethics that depends on free will. It would be inconvenient if I had to find another system, so therefore free will must be true." Not very convincing.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
We had a discussion about something similar some time ago, and there the input of codes like these were identified as one of the possible obstacles for windows game TASing, since distribution of these codes may be problematic. This is especially relevant if you are TASing the installation process of the game itself. But if you can assume that the game has been installed and activated already this issue would mostly disappear.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
feos wrote:
The second variant freezes milliseconds.
Sorry about that. If you really need milliseconds, I recommend this:
while true do
local mins = math.floor(movie.framecount()/3600)
local secs = movie.framecount()/60-mins*60
gui.text(10,160,string.format("%s:%05.2f",os.date("!%H:%M",mins*60),secs))
FCEU.frameadvance()
end
I haven't tested this, but the point here is that we do the seconds and their fraction through %f, and only do the hours and minutes through os.date.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
How about this?
while true do
local secs = math.floor(movie.framecount()/60)
local frames = movie.framecount()-secs*60
gui.text(10,160,string.format("%s:%02d",os.date("!%H:%M:%S",secs),frames))
FCEU.frameadvance()
end
If you don't like the :60 format for the frames within the second, you can easily change that by replacing the gui.text line with
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
The main point of public broadcasting without advertisements is not that you don't have to endure commercials (though that is valuable in itself), but its independence from commercial interests. It will, for example, be in the interest of a television or radio channel that gets most of its income from advertisements to have a level of self-censorship when it comes to material that would be unpopular with its advertisers. In theory, it is also possible for such a channel to run programs which are less popular or (more expensive per viewer) but more informative than commercial channels, though in practice, such channels tend to care a great deal about viewership after all. Publicly funded channels are also valuable as a counterweight to the consumerism-oriented viewpoint necessarily projected by channels with commercials.
However, I don't see why such channels should be funded through TV fees. Such channels are a general public service, just like education, roads, water, etc., and should be financed the same way: Through normal taxation. That would cut down on all the bureaucracy with TV licenses and inspectors. It would also allow progressive taxation, so that the poor would pay less (or nothing), while those who can afford it pay more. As Bisqwit points out, the current TV license fee hits the poor much harder than the rich, since it is a constant amount per year.
The Arixv is a very important resource for physics and astronomy. But it is not peer reviewed. That means that something being on the arixv is no guarantee for its quality, and it has its fair share of crackpot articles. After reading this one, I can confidently say that it is one of them.
Theoretically, it seems to be based on the old and discredited idea of shadow gravity, but this is only mentioned in passing - no actual attempt is made to explain anything. The bulk of the article deals with correlating various earthquakes with the alignment of Earth and various other bodies. I guess this is based on the idea that alignments cause gravitational shielding under shadow gravity (nevermind that this has been thoroughly looked for but never found).
The biggest problem is that the alignments are allowed to be pretty imprecise. So imprecise, in fact, that it seems like every day of the year may qualify as some kind of alignment: The article states that there are 30-40 alignments a year, and they are at least 3 days long each, but most are much longer (see table 3: typical lengths seem to be 10 days or so). No wonder he manages to find an alignment for every earthquake! This would have been obvious if he had attempted to estimate the significance of this, i.e. the likelihood that as good a match as the one he found could have happened through pure chance.
Additionally, the choice of Elenin here is not well motivated - there are countless other bodies of the same size category (3-4 km in diameter), and more are being discovered at a tremendous rate. Any one of those would probably produce just as good "matches" with earhquakes as Elenin.
Finally, let us do a quick back-of-the-envelope calculation of Elenin's ability to cause earthquakes using proper physics. For Elenin to disturb the Earth's interior, it is not enough that it pulls at the Earth - it must pull stronger at some parts than other. This is possible, since some parts of the Earth are closer to it than others. Such differential forces are called tidal forces. Unlike the direct pull from Elenin, which scales as 1/r^2, its tidal forces scale as 1/r^3. Specifically, the tidal acceleration from Elenin at the surface of the Earth is 2*Rearth*G*Melenin/d^3, where d is the distance to Elenin. At its closest approach (much closer than it is now) it will be 0.2338 AU away from the Earth. This gives an acceleration of 5.3*10^(-19) m/s^2, which is over 40 trillion times weaker than the tides from the Moon, which we experience all the time.
It should also be mentioned that while alignments are important in many superstitions like astrology, they have no important role at all in neither Newtonian gravity or general relativity. The only reason why an astronomer might be interested in an aligment is to study the atmosphere of a planet by having starlight shine through it. Note also that orbits in the solar system are not 2D, but 3D: They do not all occur in the same plane (though many objects almost do). Thus the 2D plots used in the article and first youtube video (I did not look at the others) are insufficient for determining alignments (not that they matter anyway).
To summarize: The theory is nonsense. The method is nonsense. The article is nonsense. And the video is nonsense.
It was entertaining, though, and it was a relief to see that few people here were taken in by it.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Wow, that was very good! I first thought that the any% wouldn't be able to stand up to the 100% in entertainment, but you really did a good job on that. I especially liked the music synchronization, which I think has become something of a rarity lately. Excellent entertainment choices overall! And the glitches didn't disappoint either. Keep up the good work!
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Titus Kwok wrote:
while true do
istop = memory.readbyte(0x7E0009)
dstop = memory.readbyte(0x7E05F5)
estop = memory.readbyte(0x7E0E18)
-- Items, Doors, Elevators
if istop+dstop+estop > 0 then
snes9x.speedmode("turbo")
else
if istop+dstop+estop < 1 then
snes9x.speedmode("normal")
end
end
snes9x.frameadvance()
end
Doesn't work properly with all hacks, though.
That should be snes9x.speedmode("maximum"), not "turbo". Turbo is like holding down the turbo button: It will skip most frames, but not all. Maximum skips every frame. This is especially important when dumping video and audio while using the script.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Regarding water flouridation not being "natural": The effect of flouride on tooth health was, if I remember correctly, discovered *because* it occurs naturally. Flouride concentration in water varies naturally by region, and it was discovered that people in regions with higher flouride concentration had significantly better tooth health than those in areas with lower concentrations. Follow-up tests then confirmed that flouride improves tooth health (unless the concentration gets too high). Water flouridation brings those benefits to everybody, so that you don't have to be lucky enough to live in a city with naturally high flouride levels.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
I don't really understand these "I don't like the game" or "This game makes for a boring TAS" arguments for voting no. Shouldn't we just compare the alternatives and pick the best one?
1. Publishing: We get a good (and the best so far) answer to the question "How quickly can this game be completed". People interested in the game may enjoy watching the whole of it, or using parts of it for reference.
2. Not publishing: We save some work for the publishers, and some disk space.
#1 should be the obvious choice here.
A TAS can serve several purposes, and I think the "It must entertain me personally or it's no good" aspect has come to dominate far too much here. Aside from entertainment, a TAS is useful as
A speed record
A compendium of demonstrations of useful tricks in the game
A reference for speed runners
A measurement of how long a certain path through the game takes, which is useful when considering new TASes under new conditions
Now, I still have 4 hours left to watch before voting myself, but so far, it looks impressive - it is amazing how quickly he becomes far too powerful for the bosses, to the extent that most battles end in a single attack. I have only played parts of the game a long time ago, though, so I can't watch the TAS with the critical eye it deserves.
Some things I noticed that looked non-optimal: When moving the garden across the world map, several small adjustments of the course were made: Couldn't the optimal course have been chosen from the beginning, or is the direction resolution too low? Also, at one point you are forced to walk halfway around the garden because you are placed on the wrong side of it when exiting. Could changing the direction of the garden before exiting improve this?
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
ElectroSpecter wrote:
Ah cool! Yeah it looks like there's nothing much of interest going on there.
Well, I think some interesting stuff i happening here. But the big unknown is the value of the x register, which plays an important role here. So a trace starting a bit earlier would have been nice.
This looks like it might be looking for a free spot
in the inventory (which is pretty small, just B places).
Yeah, the game only lets you carry 4 Candies, and if this trace is used for other items as well it makes sense for B to be a maximum because the number of items you can carry of each type varies, and I don't think it ever goes past 11.
Ok, that is a strong indication that the 2 upper bits of the inventory entry are used to store the number of items of that type. That also means that the limit of 4 candies would be a lot of work to remove if someone wanted to hack the game.
The $652E branch. Which is an alternative loop to the
earlier one. Different kind of item?
Hm, I'll look into this.
Could this be for the different inventory menus? I.e. weapons, consumables, magic, etc.?
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Ok, I had a go at this, but without a memory and rom viewer, I can't be sure of much. I have annotated your trace with my hypothesis of what is happening. It is pretty similar to what you said, so this might not be very useful.
http://pastebin.com/DAaj1kqu
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
ElectroSpecter wrote:
http://pastebin.com/3YdnNL2T
Here's a chunk of code that I only partially understand (mostly because of familiar memory addresses). From what I can understand, the first block is the RNG going through its cycle and the last block is putting a Candy in my inventory from the chest. In between there's gotta be some clue as to how the game chose to give me a Candy instead of the "rarer" drop, 4GP. Of course, the answer might be buried in some subroutine that isn't written out there...
It's too bad you only have the parts that actually were executed, and not disassembly of the rest. The branch that did not happen (I.e. the 4GP part) won't be a part of this, though the logic leading to not taking that part will be there. Do you have a disassembly of the relevant regions of the rom too?
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
If everything else fails, and if your emulator supports it, you could try to trace the processor execution. From a state where you know that an enemy is just about to be generated at a given address, start tracing the execution, step 1 frame or so forward, and stop tracing. You will now have a disassembly of everything the processor did during that frame. Search for the address in the dump, and then work your way meticulously backwards until you find out how the processor arrived at that location.
This way of doing things is a last resort, though, since it requires assembly knowledge and huge amounts of patience. I used it to find the loading zone triggers, treasure chest locations and event triggers in final fantasy VI, but it took many days, and I don't recommend it unless you are interested in the inner workings of games.
Experienced Forum User, Published Author, Former player
Joined: 2/19/2007
Posts: 424
Location: UK
Drewseph wrote:
you know what really depressing? During the Frigate escape, the WR escape video is 2 seconds ahead of me for the first 1/3 of the escape. then in the end I manage to pull ahead by a lousy 0.03 seconds.
Do you fall behind gradually, or does it happen at cutscenes or similar? If it happens at a few discrete events, then this could be an emulation inaccuracy instead of you doing anything wrong. A famous example of this type is opening the menu in zelda oot, which was much slower on the emulator than on the real console.
If the loss happens gradually while you move, it sounds more likely that you are moving suboptimally, but it could also be different emulation of lag.
As AngerFist says, and encode would be helpful here.
