Posts for Dacicus


Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
DJ Incendration wrote:
first IF TAS.
I think that was H2G2, actually, as linked in the submission. This probably would be the first Flash publication, though.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Sand wrote:
The version I know is KNUT0350.
I was originally planning to do one of the DOS versions but then found HOLL0350 via the family tree. One benefit of HOLL0350 is that the title screens take less time than booting to DOS in PCem. I might try a LETW0350 run at some point, since PCem seems to emulate the original IBM PC well, and that version is a PC booter that does not even require an OS (still takes 60-90 seconds to get in game, though). Already found some useful RAM addresses for that version, actually.
Sand wrote:
However, I tried the SWF at https://ruffle.rs/demo/ and also encountered a problem with text input: I can type, but cannot complete a command by pressing Enter.
Sounds like the same problem as Gnash and Lightspark.
Sand wrote:
Is is possible to post a text transcript of commands entered?
The commands are nearly the same as the walkthrough, but here you go (using -> as a separator in order to avoid excessive lines): n // Do not need instructions -> e -> get lamp -> xyzzy -> e -> get cage -> pit -> e -> get bird -> on -> w -> d -> s -> get gold -> n -> n -> free bird -> drop cage -> s -> get jewel -> n -> w -> get coins -> get axe -> e -> n -> get silver -> n -> plover -> ne -> get pyramid -> s -> plover -> plugh -> drop pyramid -> drop coins -> drop jewel -> drop silver -> drop gold -> get bottle -> get food -> get keys -> plugh -> s -> d -> bedquilt -> slab -> s -> d -> pour -> u -> w -> u -> reservoir -> fill -> s -> s -> d -> s -> d -> pour -> u -> e -> d -> fill -> u -> w -> d -> climb -> w -> get eggs -> n -> throw axe -> get axe -> pour -> n -> get trident -> w -> d -> drop bottle -> sw -> u -> toss eggs -> cross -> ne -> barren -> e -> feed -> unlock -> get chain -> get bear -> w -> fork -> ne -> e -> get spice -> fork -> w -> throw axe -> get axe -> w -> free bear -> cross -> sw -> throw axe -> get axe -> d -> drop keys -> bedquilt -> throw axe -> get axe -> e -> n -> open clam -> d -> d -> get pearl -> u -> u -> s -> u -> e -> u -> n -> plugh -> drop chain -> drop spice -> drop trident -> drop pearl -> plugh -> s -> d -> bedquilt -> w -> oriental -> n -> w -> drop axe -> drop lamp -> e -> get emerald -> w -> get axe -> get lamp -> nw -> s -> get vase -> se -> e -> get pillow -> w -> w -> w -> d -> climb -> w -> fee -> fie -> foe -> foo -> get eggs -> s -> d -> u -> w -> u -> s -> kill -> y -> get rug -> e -> e -> n -> n -> plugh -> drop rug -> drop pillow -> drop vase -> drop emerald -> drop eggs -> xyzzy -> get rod -> pit -> d -> w -> wave rod -> w -> get diamond -> w -> s -> e -> s -> s -> s -> n -> e -> n -> e -> off -> dig -> on -> rub lamp -> nw -> get chest -> get diamond -> se -> n -> d -> debris -> xyzzy -> drop rod -> drop chest -> drop diamond -> plugh -> s -> d -> bedquilt -> e -> e -> get magazine -> e -> drop magazine -> back -> u -> u -> e -> e -> d -> u -> w -> u -> s -> w -> w -> n -> s -> w -> w -> s // In maze after this command -> n -> s -> s -> s -> s -> s -> s // At vending machine after this command -> n -> w -> s -> s -> n -> w -> d -> n -> sw -> get rod -> ne -> drop rod -> sw -> blast Everything after drop magazine and before sw can be changed, since it is just waiting out a timer. But you do need something like 10 more points from finding new locations in order to get Master Adventurer Class A.
Sand wrote:
After toss eggs, does it save time to cross the chasm using ne instead of cross? (Similarly with sw after free bear.)
Yes, looks like both of those do work instead of cross.
Sand wrote:
Does the black rod count as a treasure in this version?
It turns out that the black rod is not a treasure. Oops. I also could have avoided the last get axe, but one dwarf remains somewhere in the cave.
Nicos wrote:
voting no because it don't have 1234 as rerecord count
I, too, am disappointed about that.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
In case this helps with other distros, I was able to build sdl-gnash on Tiny Core Linux 11.1 x64. Initial build with the Cairo renderer seemed successful but only showed black in the game window, so I did it again with Anti-Grain Geometry (AGG). Also had to build AGG first, since it wasn't available. Here's my process: Building AGG Source was agg-svn-r138-agg-2.4.zip from here. Dependencies are:
automake.tcz
compiletc.tcz
freetype-dev.tcz
libX11-dev.tcz
libtool-dev.tcz
sdl-dev.tcz
squashfs-tools.tcz (only needed for packaging)
Also had to download libglib-2.0.la and put it in /usr/local/lib. Then the process was straightforward:
./autogen.sh
./configure --enable-shared --disable-examples --disable-ctrl --disable-platform --disable-sdltest
make
sudo make install
Building Gnash Source was what greysondn linked earlier. Apply the sensible, boost, and agg patches from debian/patches. I think the download already has the tinatos1 patches applied (if not, oops). Edit libmedia/ffmpeg/MediaHandlerFfmpeg.cpp and libmedia/ffmpeg/MediaParserFfmpeg.cpp in order to update some variable names due to changes in FFmpeg: Replace FF_INPUT_BUFFER_PADDING_SIZE with AV_INPUT_BUFFER_PADDING_SIZE (1 occurrence in Handler and 3 in Parser). Dependencies are:
automake.tcz
compiletc.tcz
squashfs-tools.tcz (only needed for packaging)
boost-1.74-dev.tcz
curl-dev.tcz
ffmpeg4-dev.tcz
sdl-dev.tcz
libagg (built above)
libglib-2.0.la (download and place as above)
Run autogen.sh. Then edit lines 52231 and 52250 in configure to remove an extra double quote in each error message about SSH and SSL. Then
./configure --enable-gui=sdl --enable-sound=sdl --enable-media=ffmpeg --enable-renderer=agg --without-gconf --disable-jemalloc
make
sudo make install
Only did a few tests so far, but both sound and graphics work. There seems to be some sound lag, however, and the console shows some message about ALSA pcm underrun. Also, there are a bunch of messages about unimplemented things. Notes
  • I used FFmpeg for media handling rather than GStreamer because libTAS also depends on FFmpeg. "Two birds with one stone" and all that.
  • I decided against using gconf and jemalloc because Gnash allowed me to build without those. Someone more knowledgeable can comment if those might be preferred.
EDIT: Oh, the dependencies for actually running Gnash in TCL are:
boost-1.74.tcz
curl.tcz
ffmpeg4.tcz
giflib7.tcz
libltdl.tcz
sdl.tcz
speexdsp.tcz
libagg
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Is the latest commit from here still the preferred one for testing in Linux? Building fails for me in Tiny Core with a bunch of "undefined reference" messages, inluding:
  • undefined reference to ddc_i2c_change in vid_s3_virge.c
  • undefined reference to voodoo_flush in vid_voodoo.c
  • undefined reference to mvhd_close in wx-config.c
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Guernsey wrote:
Also why is AVI Splitting a thing in some emulators?
Probably because of the limits related to file size and VFW. There's also the matter of resolution changes in some games.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
feos wrote:
would "previous CD" ever be needed?
Yes. For example, The Journeyman Project 3: Legacy of Time has you switching between CDs to go to different locations, and you can change locations at basically any time. IDK how much backtracking is necessary in an optimal run, but there could be some. The hotkey idea with a "next" option sounds good to me.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
After watching the run, I'm having a hard time deciding between a Meh or a No vote. The game has some unique enemies--at least, nothing I recall from Mario games--but the levels look too much like SMW, and the character's speed is at best comparable to Mario walking.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Nice improvements throughout the run. Yes vote.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
What individual users do for their personal encodes could be variable. Most of the encodes for publications, however, dump the video and resize it using AviSynth.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
During the contest, there were a few levels that I thought would be mostly completed by one player while the other did little, but you all proved me wrong with the routing. Nice work. Yes vote.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
CIRAM holds the nametables, which is graphical data. You will not be able to change the actual score by modifying those values. Instead, I would start by searching RAM to find the score while playing. You might need to use the trace logger or debugger to find out where the score eventually gets saved.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Technically impressive, but it was too glitched out to really be entertaining for me. Meh vote.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
It seems like you spend as much time on the transition screen between levels as on the levels themselves, which reduces the entertainment. The ending music is OK, though. Meh vote. I notice that it said "Ending A" at the end. What's the difference between the endings?
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Just wondering: Did any of the teams look at the game's source code to help with TASing it? If you did, was it helpful?
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Here's the Lua script I used to get level times:
Language: Lua

memory.usememorydomain("SRAM") local addr_level = 0x0FEE local level_curr, level_prev = 0, 0 local frame_curr, frame_prev = 0, 0 while level_curr < 21 do frame_curr = emu.framecount() level_prev = level_curr level_curr = memory.read_u8(addr_level) if level_curr ~= level_prev then console.write(frame_curr - frame_prev .. ",") frame_prev = frame_curr end emu.frameadvance() end frame_curr = movie.length() console.writeline(frame_curr - frame_prev)
Since the forum doesn't have a table tag, here's the raw data I got: Team 1: 62,584,390,640,613,447,579,602,436,467,566,327,309,681,502,371,498,494,496,630,417,368 Team 2: 62,567,379,637,626,499,596,498,440,453,563,326,306,681,500,364,503,521,549,674,381,403 Team 3: 62,563,369,642,594,434,529,524,390,453,559,317,305,678,496,346,501,448,455,603,374,374 Team 4: 62,592,386,707,592,532,625,507,451,488,596,368,307,710,525,396,509,555,574,680,429,408 Team 5: 62,566,366,614,582,440,589,500,425,425,501,315,306,676,495,348,489,501,453,601,415,373 Team 6: 62,582,412,701,623,434,559,492,392,556,563,328,307,690,499,354,548,515,556,671,388,420
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
The result The contest is over. Here are the rankings: #6: Team 4 in 3:03.316 - movie file - YouTube encode #5: Team 6 "Team 5" in 2:57.533 - movie file - YouTube encode #4: Team 2 "LAZY TOWN" in 2:55.466 - movie file - YouTube encode #3: Team 1 "The Sisterhood of the Traveling Frames" in 2:54.650 - movie file - YouTube encode #2: Team 5 in 2:47.366 - movie file - YouTube encode #1: Team 3 "🐻🐦" in 2:46.933 - movie file - YouTube encode Thank you to everyone that participated and a special congratulations to Team 3 "🐻🐦" for winning!
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
I've made a Lua script that you can use to check for blank input frames at the end of your submissions, if you want. Get it here.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Yes.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
This is a reminder that the contest will end in slightly less than one week. The deadline is Saturday 2020/10/03 at 00:00 UTC for any submissions or updates to your submissions.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Yeah, the wiki description is vague: Wiki: http://tasvideos.org/PublisherGuidelines.html#Flags.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Haven't played this myself, but I found the run entertaining. The pacing is nice, autoscroller excepted. The stages don't seem repetitive compared to other games. It seems like some player projectiles are wasted, but I'm assuming that those couldn't be controlled. Yes vote.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
Here's an example I thought up. Say you want to find a counter/timer. You make some assumptions:
  • The value changes every frame.
  • It consistently increases.
You can start a new search, set Define Previous Value to Previous Frame, set Compare To / By to Previous Value, set Comparison Operator to Greater Than, and toggle the Automatically search every frame option. Then you just repeatedly frame advance or unpause the emulator to quickly (hopefully) reduce the options. One potential limitation to this method is that you may not know the starting value and when it resets, but it could save some clicks and time. Other things that you might be able to find using this method include camera position addresses and RNG addresses. You do have to make some correct guesses/assumptions about how the game behaves, though.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
I don't have concrete examples about when to use them, since I generally stick to the default options. Here's a basic explanation about what they do, though (sorry if it's obvious): 1. This option copies the current value in the "Value" column to the "Previous" column for the addresses in the search. I imagine it would be useful to manually update the "Previous" column at times when the next option might not update it otherwise. 2. This option determines when the value in the "Previous" column is updated. "Last Search" updates it when you press the Search button. "Previous Frame" updates it every frame to the value that the address had in the previous frame. "Original" keeps it at the same value as when you started/restarted the search by pressing "New." "Last Change" sets it to the value that the address had when it changed last.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
After bringing up the question about the level skip button combination with the site judges, it is no longer allowed; see here.
Memory wrote:
Dacicus wrote:
In the event that two or more runs tie for first place based on time, the combined final score (P1 score + P2 score on the trophy screen) will be used as a secondary determinant of rank; the higher score will have a higher ranking.
I really don't like this rule tbh. It means in the event of two equally as fast strategies, I have to go with whatever gets the most score rather than whatever is the most entertaining. Same with downtime.
"Most entertaining" is dependent on the viewer's opinion, as evidenced by the numerous arguments that have happened on the Workbench regarding how entertaining one submission was compared to another, whereas score is objective. WarHippy has already stated the reasoning I used to make that rule, so I do not have anything else to add. The rule stays.
Invariel wrote:
Pending approval from CoolHandMike, we are going with this team name and logo:
I'm sorry, but I believe that having a team number 5 and a separate team named "Team 5" would be confusing, especially since team number 5 has not submitted a name. RetroEdit: I have updated the rules to clarify that you can use any version of BizHawk to create the run, as long as it syncs on 2.4.0.
Current Projects: TAS: Wizards & Warriors III.
Editor, Experienced Forum User, Published Author, Player (67)
Joined: 6/22/2005
Posts: 1041
It has been brought to my attention that there are some limitations regarding editing of individual player input in TAStudio. For example, you cannot delete some frames of input from only one player while leaving the other player's input intact. I made a Lua script to assist with that specific task. Please read the comments at the top of the script, since there are some quirks to how it works. I'll address the issues in recent posts later.
Current Projects: TAS: Wizards & Warriors III.