Joined: 4/20/2005
Posts: 2161
Location: Norrköping, Sweden
adelikat wrote:
what version of FCEU are you using? I tried the game some time back and it kept crashing and corrupting my savestates.
I'm using 0.98.16. And yes, the game does crash a whole lot, making it very tedious to record a movie... In general, it crashes every 20 seconds or so, but if you stop movie-recording, and replay your movie, and then load the savestate again, it usually works.
Sometimes it crashes at the exact same frame no matter how many times you try to replay the movie. In that case, you have to replay the movie, then save a savestate a few frames before the game crashes, and continue recording from there.
To put things short, yes this game is a real pain to record, but on the other hand, it's a funny game. :) Though I can't say that any of my savestates have been corrupted so that they won't load or anything.
Joined: 4/20/2005
Posts: 2161
Location: Norrköping, Sweden
Thanks! :) I've been working quite a lot on this run these past few days, and as a teaser, let me say that I'm currently 33.5 seconds ahead of lithven's run, and I'm about halfway through the game (I just finished the cave-level). I'm guessing that the run could be done withing two weeks or so.
Suggestion: if someone would make a Wizards & Warriors run with normal route that would be faster than both of the famtasia runs, shouldn't it be enough? I'm having an impression that two identical runs of that game with only the final set of levels being different (in this case, it's not even different, since the last set of levels is technically the same as the first one) is more than enough for being redundant.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
I think we should definitely have 2 runs. We have to have a shortest completion but it's also nice to see all the levels played. I don't see what's wrong with having two runs. Not allowing the use of a certain glitch is completely against the TASvideos policy and would set a bad precedent. There are videos that use a lot heavier glitches that also have the sole purpose of being fast (ie. they have no entertainment value).
I didn't talk about disallowing anything. I just think that substituting the last set of levels with the first one and making an additional run just for that purpose is a bit over the edge. Especially since it would be 100% identical to the full version in all the common parts when played with full precision.
Come to think of it, are we choosing the faster movies only because they're faster, or there's more to that? Because if we choose to have two movies, the faster one won't bring anything new at all (however, it may have its place as an external link in the description text).
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
You are saying that we should only have the slower movie published and that if someone were to submit a faster one using the glitch, it wouldn't be published. Thus you are effectively disallowing the use of that glitch. As I said before, this would set a bad precedent and most probably lead to annoying arguments about whether a certain glitch should be allowed or not.
Even though the major goal of these movies is to be entertaining, most of them still aim for the absolute fastest completion without sacrificing time for entertainment (as with the ALttP and AoL runs). What can be done with the game is what intrests many people about these movies: they want to see to what lengths the game can be abused. When I first saw the glitch I was truly amazed, even though the rest of the run IS just repeating the first set of levels.
What about the idea of having the fastest movie with the level warp redone, and get rid of the full run instead?
I find the full version more entertaining.
Kyrsimys wrote:
You are saying that we should only have the slower movie published and that if someone were to submit a faster one using the glitch, it wouldn't be published. Thus you are effectively disallowing the use of that glitch. As I said before, this would set a bad precedent and most probably lead to annoying arguments about whether a certain glitch should be allowed or not.
It's not about the glitch, it's about its particular application.
To make this difference a bit more apparent to you, I'll provide an example. Imagine that someone finds a glitch that allows substituting every level of a game with a first (or just the shortest) one, in this or any other similar game. What would you do about that? Watch the first level for 5, 6, 10, 20 times? Even though it's only 10 minutes instead of, say, 25? You'd still stand before the same dilemma: "disallowing a glitch" or watching 10 minutes full of redundancy.
Another example: Murder Beam glitch in Super Metroid. If aiming for in-game time, one may choose to use it and win ~20 seconds, which is a lot for a TAS. Now, if you don't remember, I can remind you of at least two big debates spawned around using of this glitch.
Besides, if I follow your logic correctly, choosing not to do something may effectively count as "disallowing" in nearly any case, so I'd say fixing oneself on such an interpretation may be counterproductive.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Another example: Murder Beam glitch in Super Metroid. If aiming for in-game time, one may choose to use it and win ~20 seconds, which is a lot for a TAS.
20 seconds? I have a clear memory that it was approximately 1 second of in-game time that was saved. (And in movie length time using the murder beam was actually slower.)
20 seconds? I have a clear memory that it was approximately 1 second of in-game time that was saved. (And in movie length time using the murder beam was actually slower.)
Well, 20 was a somewhat wild guess (apparently wrong). Actually it's about ~6 seconds if dispatching the beam at the beginning of 2nd phase, and something like 3-4 if dispatching at its end (here's a demo; note that about 1/3 of my beams went through MB while it was being harmed by the murder beam effect). The difference lies in the fact that the beam "sticks" to the object it was shot at (or something like that), so if you dispatch it at the lying head of Mother Brain, it will stick to it, slowing down the 2nd phase but skipping the 3rd almost entirely.
In an unassisted run, the difference is even greater, since those 6 seconds is the time saved against the optimal 2nd phase, which requires a considerable amount of charged shots, which all require timing. So in realtime playing, it is about 10+ seconds.
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Some people may have got from my past rants the idea that I'm a hardcore frame optimization fanatic (meaning I want every movie to have as little frames as possible), but that's just not true. In this case even if avoiding the murder beam would make the movie some seconds longer (in frames) I would still much more prefer it because of the entertainment value. OTOH, I believe the movie is actually a bit faster without the murder beam?
Why would anyone care about the game's inner timer in this case? I certainly don't. Assuming that the pause screen stops this counter, it might be possible to save time by continuously pausing and unpausing (it probably doesn't, but let's assume for the sake of it that it does): It would be really annoying. The inner timer in this game serves no useful role, at least not for the casual viewer.
This may be completely different in games where the inner counter has a very prominent role, one good example being superpunchout. In that game it makes a lot of sense to try to beat the game's own clock instead of the number of frames of the previous submission.
Well, maybe it is worth splitting the thread to a separate topic (something like "Ingame clock vs. realtime clock"), but there's a lot of points to consider in this particular case, now that you've mentioned it.
First of all, the ingame counter of Super Metroid stops for:
• Ceres station exploding animation;
• door transitions;
• item acquisition message boxes;
• pause screens (but ONLY after complete fadeout; thus, it is impossible to save any time by pausing);
• certain part of an elevator ride (when Samus's sprite reaches an edge of the screen).
It runs during the rest of the elevator rides, certain other uncontrollable sequences (involving the metroid hatchling particularly), crystal flashes, etc.
What else is worth mentioning is lag. Three greatest sources of lag are:
• sprite overload (triggered by usage of the murder beam or such things as Plasma shot on Crocomire);
• Power Bomb explosion (creates a lag frame approximately once per every 3-8 frames depending on the CPU load);
• particle-happy beam shots (such as Plasma+Wave+Ice).
These circumstances leave us with four possible options.
I. Go for ingame clock exclusively (Terimakasih's initial choice).
Pro:
• fastest action sequences (less restraints on certain items usage, more items available);
• more items -> more variety;
• lowest possible ingame time;
• no concerns about using menu screen (that is, as much as needs be);
• no concerns about minimizing lag;
• easier to measure (since the game clock is a universal reference, and meaningful input is only accepted on the ingame-approved frames);
• easier to plan.
Contra:
• longest in realtime;
• more non-action sequences;
• more visually redundant scenes (equipping/deequipping items);
• more laggy scenes;
• more likely higher amount of backtracking.
II. Go for realtime clock exclusively.
Pro:
• lowest frame count;
• less non-action sequences;
• little to no laggy scenes;
• little to no menu action;
• more likely less backtracking.
Contra:
• slowest action sequences (more restraints on certain items usage, less items available);
• less items -> less variety;
• lag is a concern;
• pausing is a concern;
• optimizing screen scrolling is a concern (if your position at the moment of entering the door is shifted from the center of the screen, it will take much more time to catch up);
• harder to measure;
• harder to plan.
III. Speed/entertainment tradeoff geared towards ingame clock (Saturn's and mine choice).
Pro:
• faster action sequences;
• more variety;
• lower ingame time;
• some concerns about using menu screen;
• some concerns about minimizing lag;
• easier to measure.
Contra:
• slightly more backtracking;
• slightly more non-action sequences.
IV. Speed/entertainment tradeoff geared towards realtime clock (JXQ's and Hero of the day's choice).
Pro:
• lower frame count;
• (slightly) less non-action sequences;
• less menu action;
• less laggy scenes;
• slightly less backtracking.
Contra:
• (slightly) slower action;
• (slightly) less variety;
• harder to measure.
To sum up everything said above:
1. It is impossible to aim only for realtime or ingame clock without sacrificing entertainment. There will always be a compromise.
2. Aiming for realtime clock gives more consistent, but less variative and overall slower action.
3. Aiming for ingame clock gives more waiting, but faster, more variative and easier to measure action.
4. The only viable choices are III and IV, and both have merit to them, so it's up to runner to choose their goal.
I hope I've answered your question.
[Edit by Bisqwit: Removed font size changing markup, because it used absolute sizing (size=9), which might be larger or smaller than the post text depending on browser and desktop settings, and I couldn't figure out which way it should be. Use relative sizing (%) if you need to.]
Warp wrote:
Edit: I think I understand now: It's my avatar, isn't it? It makes me look angry.
Forgive me if this has already been suggested, but would it be possible to make an area on tasvideos where all videos are listed by submission/publication date? That would be helpful in finding old runs that could be improved.
Suggestion: if someone would make a Wizards & Warriors run with normal route that would be faster than both of the famtasia runs, shouldn't it be enough?
Joined: 6/11/2006
Posts: 818
Location: Arboga, Sweden
I doubt that the W&W Full Run can ever be faster than the re-use levels (Not to mention that the re-use levels can be done A LOT faster), but I have seriously considered improving W&W Re-Use levels, so leave it to me.
Warp wrote:
omg lol this is so fake!!!1 the nes cant produce music like this!
Including RCR. I'm not very far, but some test runs have improved the last boss fight by over 30 seconds compared to the published run.
I'm wondering if the RCR run is improvable. Vatchern's run was faster, but got grued because it wasn't as "entertaining". Stating that the objective is to beat the time isn't always enough, but if you can make a significant improvement (I think Vatchern got 2 minutes or something like that), it may prove to be acceptable.
