Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
http://tasvideos.org/userfiles/info/55639385766082321
Please replace the movie file with this one.
I found a place where I could save one death abuse with player two (and save time since it was unnecessary) and a place where I could swap out player one to take a death abuse instead of player 2 without losing time.
In this way I was able to play on hard mode instead of easy without sacrificing any time.
Now both players end the game with zero lives and on hard mode. So if one wanted to end the game slightly earlier by killing off one player they could. I still don't like this though. So i didn't do it, at any rate it is trivial now.
pretty cool it magically worked out this way.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
With a huge thanks to MrWint, a new version of gambatte has been merged into BizHawk. This version incorporates the improvements made over the years particularly by the speedrun community, into a newer code base that is much more manageable then before.
The primary improvement in accuracy is in RTC emulation, with the hope that some gen 2 pokemon games will be verifiable on console now, pretty cool!
If you are working on a TAS using the 2.3.1 version of gambatte, I strongly recommend you update to the dev build, especially if you are not too far along.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
So it turns out that audio related variables were never being savestated in the core, oops! I refactored the audio code to allow easy savestating. Now sounds should play correctly when loading a state.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
I'd definitely accept a PR for an updated gambatte, it would be great to be more in line with gambatte-speedrun. Just make sure you are being extra careful with savestates, as in my experience savestate bugs can be hard to track down in gambatte.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
The Pokemon folks have a fixed RTC for their Gambatte branch, but their codebase is quite different from that in BizHawk so I deemed it to tedious to try to implement. Might want to contact them if you are interested.
Going forward GBHawk will have an accuracy advantage over Gambatte unless someone makes the effort to fix up Gambatte . You can already see the difference in sprite-scroll PPU timing and other tests. Gambatte will remains more or less as it is , which is acceptable for TASVideos standards but isn’t likely to be hardware accurate for many cases.
But yeah GBHawk is considerably slower then Gambatte , speed just isn’t a primary goal, plus C# is slow. Also the loading times are the same, frame count is different only because frames are ended at different points (I feel like I should make a copy-paste thing since I keep having to repeat this.)
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
MrWint wrote:
I'd love to also present such a version for this submission, but it's literally impossible due to BizHawk's limitations of only allowing one input per frame. My magic conversion tool tells me that on frame 32344 (EFL timing) this submission would require multiple inputs (a neutral and an A input), which is not representable in BizHawk's input format. A real console has no problems with this of course. I guess you could try to convert it to lsnes which doesn't have this restriction, but it's probably not worth the effort.
Eventually I'll make a SubGBHawk variant like I did for NES which will accept sub frame inputs. It just needs a little more thought given all the debouncing stuff and possibility to have button press interrupts.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
I tried to burn my lives but I only managed to get down to 7 in the time it would take to otherwise climb the rope.
It would save time on hard mode, where I'd only have 3-4 lives, but you would have to have one less death abuse with player 2 at some point in the run.
So I guess it's mainly a matter of taste, but saving time in this way is definitely possible.
EDIT: actually even on easy mode there are points where player 1 is waiting long enough for player 2 that I could burn some lives. I could also give player 2 a couple extra gold without losing time to keep player one from getting another life. So it might save time there as well. But at any rate I don't like the idea.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
That only works for TSAes made in GBA mode to my knowledge. If you are making a new TAS in GBA mode then by all means use that script and maybe the pokemon people who have the hardware setup can run it and see which one works on hardware (if either.) The script just converts your movie to a file that the hardware can use, it doesn't verify anything on its own.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
You can't directly compare lag frames between gambatte and GBHawk. They deal with time when the screen is turned off differently. If you looked at cycle counts for the loading times / screen transitions they will almost certainly be identical. It's an unfortunate consequence of the fact that the gameboy can turn off the LCD completely and therefore lose the conventional idea of a video frame.
The lag in stage 8 is more interesting though. You might want to look at some trace logs around those frames between the cores to see what exactly is going on. Also try GBHawk in the current dev build to see what you get, improvements have been made since 2.3.1.
Also that accuracy test table is very outdated, and is missing some very important tests.
With so many screen transitions it's probably preferrable to use gambatte since the encode will be more reflective of an actual gameboy, especially with audio, given that there is no RNG.
Once there is a method of console verification available for original GB, it can always be revisited.
Also as a bit of an aside, 'more accurate' doesn't really mean much, it's either right or its wrong.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Thank you for the encode FractalFusion. I didn't know about (or maybe forgot) there was a score screen at the end. I don't really consider it part of the run, but if it's necessary whatever.
I also didn't test how different endings would change the time if I burned one of the players lives, but that seems pointless to me anyway, as I consider the run over once the top of the tower is reached regardless of the ending cutscene.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Okay! After a very long delay I finally got around to getting Game Gear linking working.
I tested using Dr. Robotnik's Mean bean Machine and it seems to be working correctly.
The GG linking isn't as well documented as gameboy, so some things aren't 100% nailed down yet. The most important thing is that there needs to be some latency between sending a byte and raising an NMI on the other console to read the byte. I have it set to 256 cycles currently, 128 is definitely too few and causes the game to crash. So this is something that requires some more research and refinement.
So if you are interested in playing Game Gear linked games, please test things out and let me know what problems you run into.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
EZGames69 wrote:
Alyosha wrote:
You wouldn’t judge someone’s singing talent based on a digitally ‘fixed’ version of their singing
Not a good comparison. It’s more like being a good singer but the microphone you’re using has a loud static noise to it, and you judge the singing based on that.
No, it's not like that. The output a SNES provides for a SNES game is the correct output. Personally I would even consider the word 'fixed' a misnomer in this case, but whatever.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Yes that is a big weakness of the current setup that I have thought about how to fix but haven’t come up with a satisfying solution for.
I’ve had that happen to me too when downloading a dev build or something and forgetting to add firmware , and it is quite annoying.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Seems like a pretty justifiable criticism to me. The first encode is what an actual snes would output, why should someone base their opinion on what not-a-snes would output? You wouldn’t judge someone’s singing talent based on a digitally ‘fixed’ version of their singing.
But there are lots of aspects of the movie that someone may find entertaining , and perhaps the ‘fixed’ encode just removes the distracting parts, but the run itself is the first one, not the ‘fixed’ one.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
It's cool to see how robust console verification has become for pokemon games now. Do you guys plan to push this in any new directions with other games or other pokemon gens TiKevin83?
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Even watching the encode that gets rid of the long loading screens, its just a car driving around occasionally punctuated by rather irritating 'execution' screens. I'm not familiar with the game, it just looks rather bland to me.
It's not just this game though, the star tier in general seems to have lost it's luster. It seems like something we do just out of inertia / habit. I don't know, maybe it's just me.
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
Hard to say, I won’t have much time in the next two weeks so only a little progress might get made during that time . I’d say there is a good chance of it being more or less done by June though
Editor, Experienced Forum User, Published Author, Expert player
(3536)
Joined: 11/30/2014
Posts: 2733
Location: US
I haven't had all tat much time to work on things lately, but I thought I'd mention what Vectrex progress has been made. the MC6809 is up and running and I can currently run through much of the vectrex BIOS code. It doesn't display anything or have any input yet, but I'll be incrementally working on that.
Also I haven't forgotten about Game Gear linking, just haven't gotten around to finishing it up yet.
