Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
The route was more or less trivial. I had several route ideas, including abusing death. So I find it a bit weird that you chose a obviously slower route.
My lack of optimizations come from limited knowledge of the mechanics, which means I could've saved some more time too, if I know more of the game. The usage of TAS Studio helped me a lot trying different stuff out, it makes seemingly difficult games easy, even if one hasn't TASed the game at all.
Where did I say I had problems with you? That's just a wrong interpretation of yours.
Edit: Editing the post, instead of replying.
To me it was trivial. I don't know the mental capability of these guys or why else they would failed that hard.
It's a quick realtime test to see if they disappear. Would like to join, but drawing analog input isn't done yet.
Friends? I don't have friends. Also not joining your sm-club.
TASing any game is NP-hard, what's so hard about it. I didn't talked specifically about that hack. The point with my statement was: it doesn't matter how complex the game gets, including number of players to control at the same time, non-linear TASing makes it a lot easier. But you don't even know what that is.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
Now that Sonicpacker has posted the results of his competition, in which he wanted me in. I made a run of it, not with mupen64 for some reasons already mentioned, but with BizHawk's TAS Studio, which worked perfectly, a bit tricky to use but that will change. For greater joy, I ended up being 0.133 seconds faster than Sonicpacker :D, I converted his m64 file into a bk2 file and ended timing on first visual frame Mario is on ground, I also converted MKDasher's file (winner of the competition).
For those who are interessted, here a quick encode of my run: https://www.youtube.com/watch?v=2RjRsOK3vck
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
These 8 frames can't be saved, a RNG cycle would be needed in the end of the 3rd room of 4-2 which ended up wasting more than 8 frames.
Here's another WIP up to 4-4, with overall 64 frames of improvement.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
mupen64 is not good in general. As I said it doesn't fulfy my needs of tools. BizHawk got decent Lua scripting for example, not the shitty MHS scripts. And I don't care about hardware requirements. Or if I can't move analog stick while savestating, since I would brute-force the angle (haven't worked on my script lately). But I do plan to move to 3D once TAS Studio is stable with analog input, and TAS Studio slows emu lilation speed heavily down, but that doesn't matter for me.
Speaking of multitasking, I don't know your your TASing looks like, but most of peoples TASing I watched use mouse for analog stick AND buttons, checking checkboxes requires mouse course move around more. It's clearly more efficient to use buttons with keyboard, if you have a good hotkey layout.
Mupen64 is clearly not designed for any decent 3D TASing for me at least, if it's good for you I don't complain, but I do have high expectations in tools. I want to be able to work with highest possible efficiency.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
THC98 wrote:
My idea isn't something to help while TASing (Dolphin already has a TAS Input that works very well). My idea is an input viewer for renders of TASes, pretty much what is seen at 19:06 in the video that got4n posted or in the example video I posted before.
TAS input and movie editor are two completely different things. :)
So you want an input display on screen? Encode the run, then overlay it with the input display and let buttons light up and move analog stick by also reading the dtm file.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
THC98 wrote:
It may work as a plugin inside the emulator or as an external app.
You probably mean something like that. It is the built in lsnes movie editor.
Here it is an example of what it should look like
Single frame input viewers are pretty pointless, they don't give any advantage, you can't do non-linear TASing with them well. The lsnes movie editor works with normal savestates, not rewind (saving each frame like FCEUXs TAS Editor or BizHawks TAS Studio). I'm not sure if rewind would work out for GC/Wii.
It should be possible to code a movie editor with savestates for Dolphin.
My idea is to replace the game screen to the GC controller responding to the inputs of a recorded movie (.dtm file) when the plugin is activated.
Why do you want to do that? It's not that hard to put a new window somewhere.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
I couldn't messed up anything, since I instatlled the program yersterday. The problem seemed to be the video codec, I used CamStudio for screen capture.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
Samsara wrote:
If text on-screen is all you want, Aegisub is pretty damn good for that.
I finally had time to do the lsnes tutorial and wanted to add subtitles to the video file. But somehow the subtitles are all disorted and impossible to read, I use a newly installed version of aegisub.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
Here's a WIP with an improvement of 35 frames over my old WIP. However I lost 8 frames in the second last room of 4-2, but I think it should be possible to get rid of the loss.
The improvements come only from better lag reduction, the first improvement was made in the 3rd room of 4-1, the biggest came from 3rd room of 4-2 with 31 frames.
Editor, Experienced Forum User, Published Author, Skilled player
(1508)
Joined: 7/9/2010
Posts: 1317
Masterjun wrote:
Things that won't work good with TAS Editor:
finding fastest route
It also won't work with traditional TASing. Basically you need to know the route in advance. I tried to create maps with auto stitcher, but failed at it.
The biggest advantage of non-linear TASing is that the user don't have to memorize input, for example the TASer wants to add a shot after doing some crazy precise platform jumping the TASer can simply add a B press and test it for every frame, without worrying about the other inputs.