Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
So I was building out the virtualpad for this core and noticed that the image of the keyboard shows that it has a standard QWERTY keyboard and Yes/No buttons.
I noticed the core only has Yes/No and no Y and N keys. Does the Yes/No actually map to Y and N on real hardware? or is this an oversight of the core?
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
So, I did some testing and found that the keyboard seems to not be mapped correctly.
The numbers aren't quite right. 6 is pressing 1, 7 is 2, etc. And 1 is 6. This is apparent if you try Math-A-Magic, where you type answers to math problems.
Notice the screenshot, the input display shows I pressed 6, but the output is 1
Screenshot
I have some other issues I found but it is hard to know what is the emulator or the actual hardware, so I will do some more research first. Just wanted to point this one out as it is a pretty big one that makes testing hard if you don't know about it.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
Please don't change that, without a discussion. We will have to add support in the tasvideos submission parser if we change the systemId to "GBC".
Currently it is "GB", with CGBMode flag set to 1.
I would recommend fix the importer to do this, rather than change the systme id
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
Alyosha wrote:
adelikat wrote:
as of 2.4, I added sync settings to the importer logic. To do that means it picks the core. I picked it to be GBHawk. My understanding is that is the more accurate and therefore preferred core.
We have no support currently to add sync settings ANd let the user choose the core
it does? I tested just now and when gambatte is loaded the input roll still is imported as gambatte, so this at least needs to be fixed.
Here's the importer logic:
Language: C
if (isGBA)
{
Global.Config.GbaUsemGba = true;
var ss = new MGBAHawk.SyncSettings { SkipBios = true };
Result.Movie.SyncSettingsJson = ConfigService.SaveWithType(ss);
}
else
{
Global.Config.GbUseGbHawk = true;
Result.Movie.SyncSettingsJson = ConfigService.SaveWithType(new
GBHawk.GBSyncSettings());
}
As you see, it's hardcoded that GBA = mGBA, and GB/GBC = GBHawk (SGB not support in the importer yet)
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
as of 2.4, I added sync settings to the importer logic. To do that means it picks the core. I picked it to be GBHawk. My understanding is that is the more accurate and therefore preferred core.
We have no support currently to add sync settings ANd let the user choose the core
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
Are you sure you are ready for testing? There's a lot of games that don't work but because they need mappers I'm pretty sure.
But here's my preliminary findings:
Doesn't work at all
Super Cobra (EU)
Buraco Negro (Brazil)
Spiderman
The blobbers
Other issues
Frogger (EU) - game play fails (title screen works)
Chess - no pieces
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
BizHawk 2.4 Released
This is a hotfix release. Oddly it's also a major version update!
This officially releases the vetrex core (mistakenly not officially released last release).
In addition some major lua and tastudio bugs were fixed.
2.3.3 should be abandoned promptly.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
BizHawk 2.3.3 Released
This is a pretty major update to TASing tools. TAStudio is now 2-3x faster! As well as numerous bug fixes and improvements. Ram Watch and Ram Search have been sped up a great deal too.
And we have a new mGBA, and a bazillion fixes. Enjoy!
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
GJTASer2018 wrote:
The problem with that idea is that FractalFusion's run is from mid-2006 using a very old version of FCEUX (0.98.16). I highly doubt any judge is going to seriously consider any submission made on an emulator version from 13-14 years ago
This is nonsense. First of all, FCEU isn't notably less accurate than FCEUX (they are both bad). Secondly, the movie likely would sync on FCEUX and even BizHawk. If not, we could resync it. We have options here, and if there's a publishable movie hanging around in the queue, we could absolutely resurrect it.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
I'm going to throw a very strong vote for completion time over input time. Very strong.
FractalFusion's movie is a slower movie than any I listed in the submission.
It's neat when someone cleverly lowers input time while still completing the movie as fast as possible. It adds a level of strategy and complexity. Shortening input but drastically increasing the actual completion time is not so cool. I am now watching a longer movie because you want to optimize the tasvideos submission parser's reported time instead of the real time.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
the new release will have "invisible emulation" as an option. Current release does not unfortunately.
However, I wouldn't expect it to give you any meaningful speedups. Almost all of the speed cost in cores is the emulating of the hardware, not the displaying it on the screen.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
Bloopiero wrote:
There are 'Undo Load State' hotkeys, which seems like a much more efficient solution
Yeah, I coded those into BizHawk and FCEUX many years ago for exactly that scenario.
And to add to the stories:
Spent all day working on a difficult section and finally got to the end. Found out I wasn’t even recording.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
I just wanted to say, sorry I jumped the gun on any claims that the movie had copied input based on a few seconds of video.
I'll retract that and say I don't know either way. And I'm not going to investigate more than I have. I have better things to do.
I'll instead say, that I would have made xipo a co-author in this case, and I'm disappointed in TASeditor's choice on this matter.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
Xipo, thank you for bringing this to our attention.
The taseditor (the tool in FCEUX) view you provided is particularly interesting. I think even by taseditor's (the person) definition, you deserve to be an author of this movie.
I think his definition is too strict too. I think the community would agree that you have contributed meaningful work to this movie and should be an author even if technically none of your inputs were actually in the movie. (Again, I think they are and there is clear evidence).
My recommendation is that we add xipo as an author. That's easily done for the submission and publication pages. Someone will need to re-encode this though for the youtube subtitles.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
I just wanted to say that Radiant is very likely correct.
Also, Warp is correct. I think what technical was supposed to be isn't generally the way people use it.
I did some statistics and I have the 2nd most ratings next to Arc (I'd be curious what you think here, Arc).
What I can tell you with my 4930 ratings is that technical is not valuable. I see two things, either they are linear like Radiant suggests, or they are the "consolation prize" for rating something low in entertainment. Either way, they add complexity without value.
My opinion is that we should should WIP technical from our rating system. What we can do is calculate the average as (Ent * 2) + Tech / 3 and then replace entertainment ratings with that value, then drop the technical values.
Then the UI is a single value, which I think more people will participate in.
Also, I propose that we limit the decimals to .5 on the movie module, and let users go to the main rating page if they want to add nuance to their rating.
I'd be prepared to take on this work in the near future, if people agree with this idea.
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
"As Acmlm noted, we could get a more optimal solution if the movie input wasn't tied to frames and we could unpause after skipping a desired number of random numbers"
You could use BizHawk, and the "SubNesHawk" core, that is tied to input polls rather than frames. Would that help get better rng with fewer delays?
Also, I loved this TAS. Not getting Erdrick's Armor is surprising and even more entertaining. Thank you for submitting this!
Emulator Coder, Published Author, Site Developer, Site Owner, Expert player
(3602)
Joined: 11/3/2004
Posts: 4739
Location: Tennessee
I just wanted to post to say that I think this movie should be published.
I have a lot of opinions why. And it may take me awhile to articulate them. In the meantime, just wanted to say I strongly agree with publishing this, and to reject it over it's initial ram state choice is a mistake
