I'm ready to TAS the game, but I noticed the sound is wrong when transitioning to a new room or when a new room is entered. I haven't been able to fix it.
I will try again in the next 1-2 days and if I can't get it to work, I will ask in the DOSBox-x github.Fingers crossed.
Did more research on v1.2.
I have not managed to find addresses that correspond to the tiles in a room, maybe it is stored in bits instead of bytes.
But I found a few other interesting addresses that will make TASing this game easier.
I will soon study the 2nd part of the game (the 2nd maze).
Once the message starts after receiving the 10,000 pt and clicking away the "extra man awarded" message, the game decides on the room where you have to enter any hole to receive the key. It isn't a specific hole you have to enter.
You can edit your current room id and leave the room to travel directly to the room you want, and to influence the game deciding on the room with the hole, the room with the keyhole, and the corner room.
The room with the hole, the room with the keyhole, and the corner room can be any room that isn't adjacent to your current room.
You can luck-manipulate those rooms by delaying a few frames before the message pops up that sets up these rooms.
Map for the 2nd maze:
Looks close to my map in the opening post and identical to this longplay, so perhaps the rooms are hard-set like this.
Note that there are one-way doors. Entering a hole in the 2nd maze will sometimes take you back to the beginning.
The save/load feature is disabled completely once you enter the underground passageway in the 1st maze.
The TAS is dead. :(
Video playback in this game is messed up after all.
I assumed the game was behaving fine and video playback is functioning, since the intro and the chimney ghost video were playing.
But turns out, after talking to the attic ghost and then trying to enter the mirror right before the end of the game, the game softlocks as it doesn't play the mirror spinning video.
That's a Sunday wasted. The TAS would have been about 15 seconds long.
I could finish by using savegames but it's not something I will want in a submission-worthy TAS.
The game should probably be played in Windows 95 instead.
I've been running into a myriad of problems trying to TAS this game...
Turns out the 16 bit version that comes with the CD-ROM actually doesn't run as is on Windows 3.1 and win32s is required to make it run.
The vfw (video for windows) install that also comes with the same CD-ROM doesn't like any win32s version I tried so there is a "msvfw32.dll is missing" error when starting the game. I tried a lot of things but couldn't fix it. It could be possible the developer hadn't tested compatibility of their software with windows 3.1.
The game requires the sb16 drivers so I can at least get adlib MIDI. Unfortunately it wasn't possible to load the game .cue / .img pair along with the win32s .img disks and the sb16 .ima floppy file so I had to make two back-to-back verification movies.
Aside from the dll error, the game runs but it is nearly impossible to keep track of what input is doing what action. I'm basicly adding a click or key press and wait 10 seconds for it to play and see if it does the desired action, rinse and repeat thousand times. Mouse clicks don't line up with key presses. It is very easy to get an item, then add more input and then suddenly "not" get the item. It's frustrating to notice such things later in the game.
Then there are inexplicable "InvalidOperationException" errors suddenly happening. First it happened when I was just trying to copy the game files over to the .hdd during the verification movie, even though there was enough space. So I opted for running the game from the CD drive. It doesn't cost any time but I still would have prefered the other way... The error happens sometimes in very high density during the actual gameplay part, but I'm not certain if it is related to TAStudio or the game.
I've never had this error happen in other games I ran.
I hope to get the game finished within the next few days.
Bizhawk not recognizing files through .XML
I might have come across a problem where Bizhawk is supposed to mount a floppy disk file to A:, but doesn't.
I'm trying to load a few .img floppy disk files along with a pair of .cue / .img files as an .xml.
Bizhawk does provide "previous/next/swap floppy disk" inputs, but only the .cue/.img is mounted to D:
The floppy disk files are recognized and can be changed/swapped, but drive A: is not set up and files are not mounted to it.
This happens regardless if I list them before or after the .cue/.img pair in the .xml.
This means I will have to work with two verification movie files.
Bizhawk creates branches while I'm entering a number
In TAStudio when I enter a mouse coordinate/speed, Bizhawk sometimes interprets it as "save branch" even though TAStudio is in focus.
Make it possible to load Sync settings
Make it possible to load sync settings from a movie file or a savestate, to avoid problems like this.
Congrats on the publication!
I noticed the category ended up being called "all sub-games". That's a bit misleading because technically the game contains another subgame, Boss Endurance, which unlocks after 100%. Not a big problem but I needed to point this out. Sorry I'm late about this.
That's not to say Boss Endurance is missing from this run, since I think it should be its own category anyway.
For the record, about the problem with MIDI not playing correctly in Warhead, it's working perfectly fine in standalone DOSBOX-x (the January version).
I tried copying some midi, mixer and sblaster related settings from dosbox-x.conf for testing in Bizhawk, but...
I noticed adlib is listed in MIDI Mapper (if you try to Edit and try to change Port Name) during the w311 verification movie but it's no longer listed after a reboot.
So I was under the impression it might work before that reboot happens and so I tried playing the game before the reboot, but...
Did another 2 hours trying both sb16 and sb16vibra, uninstalling and re-installing Creative Labs Sound Blaster 1.5, but...
Still no MIDI sound in Bizhawk.
Could it still be some .conf setting difference between the standalone version and the Bizhawk version? Or it's a bug with Bizhawk?
I have not asked in their github yet. Since it works in standalone DOSBOX-x, there is not really a point.
In standalone DOSBOX-x, I can uninstall and reinstall Creative Labs Sound Blaster 1.5 (selecting the current driver) and MIDI still works fine with Warhead.
EDIT:
I copied the entire dosbox-x.conf to load in Bizhawk and it's not working. At this point, I'm thinking it's a bug on Bizhawk's end.
Leave screen instantly
When fading from inventory or sign back to the current screen, if you buffer a click to an exit, you will leave the screen immediately and the walking cutscene is skipped.
It seems you cannot buffer a click to an object or item, however.
Nevermind, I found that you can! This opens up even more bugs you can do!
I found that you can click on puzzles before they are even unlocked! And it skips the startup animation! You can also do puzzles again, which is normally not possible.
You can pick up items that have not been dropped yet!
I found these things in the room FOREST MOUNTAINSIDE.
It seems you cannot buffer a click on every thing on the screen, however. It will need research and I will put the results in the guide above eventually.
Misc notes
In room DWARF ROOM 4 - ECHSENHÖHLE, if you do the lizard puzzle again, while the witch is still behind the gate, the witch will "fly in again" and try to pull the lever to retrieve your items again.
There are none and the game bugs out with an error:
Script error: Handler not defined
#getAt
[Stop] [Continue]
If you stop, the game closes. If you continue, the screen graphically bugs and the error happens again every frame.
Eventually it looks like this...
I have tested the bug again with the witch not yet inside the cave and using only one item on the plate, the bug did not happen.
If you buffer click the egg in FOREST MOUNTAINSIDE and the witch asks what to use it with, you can click on other things early which will glitch out graphics.
If you click on the FIVE ELFSTONES and on the left exit, both actions will be performed simultaneously!
Next to the lake, select the plate from the inventory and buffer a click on the gap in the pier. This skips the witch talking. You can then click on something else to cause another action happen simultaneously. If you click on the pier, the cutscene with Marigor draining the lake will happen sooner since the animation ends sooner.
Select fishing rod and buffer click on something that doesn't work, then click on the pier. This will cause the cutscene to play even sooner.
In CAVE MINING CART, you can buffer click on the shovel, click on the heap of coal and then still click on something else for a simultaneous action. Need to test similar things in other rooms.
In CAVE DWARF ENTRANCE, after clicking on the riddle and positioning yourself in front of it, if you quicktravel somewhere and then quicktravel back, you will still be positioned in front of the riddle. The game will remember alternative positionings.
Yes, I used the same game files in ScummVM and it doesn't crash at all.
According to hyperformancespeedruns on his 7:48 WR video:
when you are in the glitched state inside the shipwreck: use/pick up the pin > ESC the cutscene > inv > use the pin (or skeleton arm) with voodoo doll > inv > use skeleton arm with lantern.
So I did, but it just brings me back to the shipwreck interior. When I use the glue on the skeleton arm, it works! Ah, I guess the 7:48 run used easy difficulty.
I have not been able to run ScummVM (the 2026.1.0 Windows 95+ zipfile version) in Windows 95. It just crashes on launch.
RTA runners have been using a new glitch to go to Blood Island early.
According to Youtube comments, the original bug was discovered by Totem, and was expanded by Zelop.
Explanation of the original bug:
Link to video
In the settings, Text needs to be checked.
Aside from the things you can do that are explained in the video, it looks like runners are basicly doing this:
Go to the shipwreck and click on the path beyond Murray.
Key "U" on the ship a few times - When Guybrush is past Murray, open inventory.
Key "E" on the pirate literature.
"ESC" out of the dialogue and double-click into the shipwreck.
The game runs fine on Windows 95 in DOSBox-X and I've been testing the bug.
Unfortunately, the game keeps closing on an error when performing the bug.
Not sure if I've been doing it wrong or everyone else is playing on ScummVM?
I went and started a TAS using the 1990 preset. Although animations are running fine, enemy movement is still way fast, but possible and easy to deal with.
Now, what's not so possible and easy is the behavior of boulders falling down. If a boulder is right above you, you can move down and to the side with frame-perfect input and dodge the boulder. But if the boulder is located on a ledge and it falls to the side onto you, it is impossible to dodge.
Unfortunately, it looks like level 12's level design expects you to move down through such a gap to reach the exit.
I don't see a way to complete this level so the TAS is dead for now. More research needs to be done to decide whether a faster or a lower machine setting needs to be used.
EDIT: Looks like it is the game's difficulty setting. I'm using hard difficulty. I thought it speeds up the timer only, but it seems it also speeds up the framerate in general, making enemies and boulders move faster. Using easy difficulty actually makes enemies move appropriately fast.
Can't believe the game's own difficulty setting renders it unwinnable... The developer apparently never tested this level on hard.
Added file hashes to the submission text.
Please replace the submitted bk2 with this one which changes the SHA1 hash in the Header:
I removed the file Pako 2 (1995).ba1 and created a new ISO.
The file isn't part of the original game files and not required for running the game.
The movie still syncs as before.
This is the kind of game that draws my interest. Love those fake 3D graphics. I would like to see the collapses compilation as a video.
Would it be possible to move the mouse away after the final click? The tooltip makes some descriptions unreadable. Is there not a hotkey to start the test?
The game runs unnaturally fast on the 1991 and 1993 presets.
Nevertheless I'm still interested in seeing answers:
1) What if I were to use the 1991 preset regardless of the game running way fast?
2) What if I make a TAS on the 1990 preset and someone in the future tries to beat my TAS by using the faster 1991 preset?
Dimon12321's guide suggests to change the cycles setting as one sees fit.
3) Is that allowed? Isn't that arbitrary?
4) What if someone in the future tries to beat my TAS by using a higher cycles setting than I did?
EDIT: Thank you @ feos
You must select one of these presets to run your game. The choice needs to be made considering the publication year of the game. That is, the machine should have been one produced at or around the game publication as reasonable. Choosing a much more modern preset may result in the game running too fast, whereas an older machine may be too slow or incompatible. In case of doubt, you can ask the Judging staff.
If I were to TAS VGA-BoulderStar in Bizhawk 2.11, what machine preset and what settings am I permitted to use?
The game was released in 1991. Animations and enemy movement becomes extremely fast, the higher I set cycles. The shareware version's nag screen says something along the lines of "On the full version, you can run the game even on a 486".
A Boulder Dash-like game developed by Standsoft in 1991. This is one of my childhood games and I only rediscovered it yesterday.
My memories of it were very vague. It's funny that I still remember the final password...
I can tell it looks a lot like Diamond (DOS) but I've always felt that wasn't the game that I had played...
The goal is to collect a certain number of diamonds and reach the exit before time runs out.
Game version
There is a shareware and a registered version.
The registered version has been distributed in another forum, along with the developer's quoted statement redistribution is allowed. It contains 100 levels and an ending screen after you pass the last level.
Passwords
I found that if a given password will allow you to access and reveal passwords for levels beneath it.
So I will only provide the final level's password:
Level 100 - BYE
Plucking
Most of the game mechanics are explained in-game, except one thing that seems to be explained only in a separate help file:
By holding SHIFT and pressing a direction, you can "pluck" grass tiles and collect diamonds without going into that direction!
Lava
Lava is a tile that spreads itself across the entire level unless enclosed by rocks, diamonds or the player.
It can transform into diamonds or rocks after a certain time.
I noticed some strange behaviors with this, since lava can sometimes stop spreading altogether and sometimes doesn't transform at all.
Its movement and whether it changes to diamonds or rocks seems arbitrary and random.
Game's speed
If cycles are set very high, enemies will move extremely fast and some graphical aspects are running too fast. The in-game time limit seems to run normally though.
If cycles are set very low, enemies will move very slowly, the time limit will move much slower. At this speed, it is possible to go next to an enemy and then move away again without getting caught. However, it seems - generally speaking - your movement speed will be slower as well.
If cycles are set to make the game run at normal speed, it seems to be impossible to buffer input such that you can move next to an enemy and away without getting caught...
I'm not sure what "normal speed" constitutes. The help file says the game is supposed to run fine on a 486. I will need more information on what settings to use to make the game run at a speed that is considered "normal". And testing needs to be done to find the point at which the "move next to an enemy and away" trick becomes impossible.
The internet has no idea what is "WAR_BIG.DLL". I assume, this might be a game-specific dll
It's a .dll file specific to the game.
I wrote:
When I use the 1994 preset, the drivers can be installed just fine, but the game complains that it can't find WAR-BIG.DLL.
The file gets displayed as WAR_BIG.DLL so maybe the OS confuses - with _.
It seems, depending on the cputype, files get interpreted as one or the other:
Since the game looks for "WAR-BIG.DLL" specifically, it fails on any cputype that interprets it as "WAR_BIG.DLL".
Take a look at Gravis Ultrasound, particularly gustype=max.
It's intended for MIDI playback.
I tried adding a .conf file with
[gus]
gus = true
gustype = max
But I can't find a difference. How am I supposed to use this?
I also tried installing Gravis Ultrasound drivers but didn't succeed because it didn't recognize one of the installation disks. The other driver I tried made Windows unbootable...
The only difference I see is that 1994 preset uses sbtype=sb16 while 1997 and 1999 presets use sbtype=sb16vibra.
I found that if I use sb16, I can install the drivers just fine, but when I use sb16vibra, it asks me to install the configuration software.
With that knowledge, I used a cputype that interprets the filename as "WAR-BIG.DLL" (don't remember which one that was), and made sure to use sb16.
The MIDI file doesn't get played.
I managed to get it to work eventually:
1) Installed the sound blaster drivers, before doing anything else. IRQ setting to 7.
2) I then copied the game files over to C:, started the game, confirmed to not show midi errors in the future.
3) opened MIDI Mapper.
Trying to set to AdLib would tell me, it is trying to reference to a thing that is not installed.
I could set it to General MIDI though, so I did.
When I try to edit General MIDI, only Creative Labs Sound Blaster 1.5 shows up in the list.
Using this produces no MIDI sound in the game.
4) Rebooted Windows twice.
5) Somehow, this makes AdLib appear on said list and choosing AdLib in the MIDI Mapper is now possible, so I selected it.
This made MIDI play in-game. But it sounds inferiror, nothing close to what it is supposed to sound like.
I made a rough map while playing in Shareware v1.2.
You can see what the game actually changes with each new plot advancement.
During testing, the keyhole room could be different depending if I waited before entering the hole, perhaps due to the cats' movement advancing the RNG. This seems to match the information I wrote in my previous post.
I have not been able to get a different corner room in my testing so far.
I thought the keyhole room will be exempt from getting changed, but turns out I was wrong, although it looks really odd. Like a part of the room got shifted one tile to the left. And the cats are gone.
(On the left, how it was while I was still in the room. On the right, after re-entering it.)
I loaded an ISO file in Bizhawk 2.11, played around, made savestates.
Closed the emulator for the day, started it on the next day and played some other game and did some other stuff.
Now I wanted to go back to the game I had played before but my savestates won't load.
It says, the sync settings don't match. I tried a lot of possible combinations for the settings but I can't get it to work anymore.
Isn't there a way to retrieve the original settings to make the savestates work again?
EDIT: Thanks, CasualPokePlayer!
The MIDI problems make me sad...
I spent like 3 hours trying to create a verification file for Warhead (the 1995 rerelease shareware 2.02).
In standalone DOSBox-x with Win3.1, I managed to make Tyler sound correct simply by changing MIDI Mapper from "Adlib" to "General MIDI". I think I installed those sound blaster drivers but I also did a bunch of other things so I'm not sure what the actual fix was.
Anyway, installing the drivers in Bizhawk and changing to "General MIDI" didn't work in the end.
Now, about Warhead... I need to install the drivers for that game, too, since I need to get its MIDI background music to play.
Got it to work in standalone DOSBox-x, but in Bizhawk? No luck.
When I use the 1994 preset, the drivers can be installed just fine, but the game complains that it can't find WAR-BIG.DLL.
The file gets displayed as WAR_BIG.DLL so maybe the OS confuses - with _.
In the 1997 preset this problem doesn't occur and the game starts, but the drivers can't be installed since for some reason it wants me to install some configuration tool instead.
I don't know why the driver installation is different under this preset...
Tl;dr: Much frustration has been had and Warhead is not TASable at the moment.
Possible audio crash fix
After trying a lot of things, I stumbled upon a possible solution to the audio crash problem.
Start menu → Run... → dxdiag → tab "Sound" → turn off hardware acceleration completely
This makes sound stuttery, although that may be FPS-based.
But it seems to fix some games that were previously unplayable, including Tony & Friends New Adventures, 3D Ultra RC Racers and Snake Arena.
After turning off hardware acceleration, I haven't had a crash in those games yet.
For some reason, Hot Wheels: Crash! became completely unplayable though, like, even if I reset everything I've done, set back hardware acceleration to full and re-install it...
It always gets stuck on that first screen now.
EDIT: It looks like core=normal caused this, a setting I forgot to reset back. After commenting it out Crash! began working again.
For some reason, when I removed my .hdd and copied in a backup .hdd, the stutters disappeared even with hardware acceleration turned off.
Use sb16 rather than sb16vibra
I learned that you can apparently use sbtype=sb16 after all.
Control Panel → Add New Hardware... → (confirm everything and reboot twice)
Note: For everything above, I played on Windows 95.
The "good ending" run is currently submitted and sitting on the workbench. Now made a bad ending run:
Link to video
Well, I just took the "good ending" run and changed it to head to the exit right away. Nothing special.
Only having two encounters is pretty great even in this short version.
I will submit this only if requested.
The game does run on DOSBox-X, but there is a bug with the score displayed at the side.
It needs to be fixed before a TAS should be started.
https://github.com/joncampbell123/dosbox-x/issues/6201EDIT: As has been pointed out, the configuration
[cpu]
fpu=false
has to be added to get rid of the floating point mistake.
Turns out, Bizhawk already does this in its base configuration.
Looks close to my map in the opening post and identical to this longplay, so perhaps the rooms are hard-set like this.
Note that there are one-way doors. Entering a hole in the 2nd maze will sometimes take you back to the beginning.
The save/load feature is disabled completely once you enter the underground passageway in the 1st maze.I have tested the bug again with the witch not yet inside the cave and using only one item on the plate, the bug did not happen.
I don't see a way to complete this level so the TAS is dead for now. More research needs to be done to decide whether a faster or a lower machine setting needs to be used.
EDIT: Looks like it is the game's difficulty setting. I'm using hard difficulty. I thought it speeds up the timer only, but it seems it also speeds up the framerate in general, making enemies and boulders move faster. Using easy difficulty actually makes enemies move appropriately fast.
Can't believe the game's own difficulty setting renders it unwinnable... The developer apparently never tested this level on hard.
You can see what the game actually changes with each new plot advancement.
During testing, the keyhole room could be different depending if I waited before entering the hole, perhaps due to the cats' movement advancing the RNG. This seems to match the information I wrote in my previous post.
I have not been able to get a different corner room in my testing so far.
I thought the keyhole room will be exempt from getting changed, but turns out I was wrong, although it looks really odd. Like a part of the room got shifted one tile to the left. And the cats are gone.
(On the left, how it was while I was still in the room. On the right, after re-entering it.)