Maze Master: Won!

Eat it, you basement-dwelling, spandex-wearing, ant-headed freak!

 

It all ends on level 5. You can't magically teleport there, but you can teleport directly to the stairs on level 4 that lead down to it, and the balrog's lair is just a few secret doors away. I found him almost immediately, and he wiped the floor with us. There's only one of him, but he strikes first, always hits, and instantly kills whoever he hits.

But let's back up for a second. Each of the upper floors has a clue hidden somewhere, which I have shown in the maps but not mentioned yet. Said clues are:

1. I HAVE LAID 3 CLUES TO BRING YOU TO ME...
2. I AM DESTINY PERSONIFIED
3. TOWARD THE NE WARE THE STOP ELSE MAGIC TRIES TO MAKE YOU DROP
4. MY NAME IS IN THE LOWER 4 MAPS

The walls of maps 2 through 4 contain one identifiable letter each. F-A-T. I don't think it takes a riddle master to figure out what comes next.

 

To beat the balrog, I assembled a team of three Freds. Magic is good for zipping around and eliminating crowds, but the balrog fights alone. I need concentrated damage, I know protection magic won't help, and I can get three good warriors by dismissing my wizards and restoring Fred's character code twice.

This means no teleportation, so we walk to level 5. No big deal. I have maps, and the path is clear. There's just one mandatory fight along the way, a square of infinite monsters on level 4, but each Fred can hit up to four times now, and you can pass through the square after winning the fight - it just doesn't "clear" for the future the way others do.

On reaching the hidden lair, we are presented with a riddle to enter.

FATE. You are FATE.

We enter and fight to the last Fred.

After winning you are trapped in a 2x2 room with no way out.

 

And, well, that's it. I didn't even need to explore level 5 or train further. I just had to slightly abuse a questionable character management system.

Maze Master defeated CRPG Addict, who wrote in 2014 that it would have been "fun" to be the only person online to beat the game, and later did with hacked supercharacters, save states, and spoilers. I may now be the only person online to beat it without doing any of that, and it wasn't that much fun.

GAB rating: Below average. Borderline bad, really. There's nothing dysfunctional about Maze Master, but it is emphatically not fun or interesting to play. Wizardry was good in spite of its bare-bones look and feel because of its deceptively deep combat engine, but there's no strategic depth to this imitator whatsoever, nothing of interest at all to find in these mazes, and once you gain enough power to do any serious dungeon exploration the combat becomes and remains pathetically easy, which is admittedly better than the more common alternative.

 

I found some spoilers regarding the magic items, and all of them except the Ring of Accuracy are useless.

• Staff of Light - Useless. Its just a permanent Cat's Eye spell, which should be no problem for you to cast as needed long before you can afford this staff's $5000 pricetag.
• Ring of Accuracy - A +25% chance to hit. Mainly useful for the fighter, but not a bad effect. The Accuracy spell does the same thing, but has to be cast during combat, so having this as a permanent buff is convenient.
• Amulet of Healing - Nearly useless. Heals one HP every ten seconds. Who wants to wait several minutes between combats to heal up? Especially since wandering monsters will certainly interrupt your rest. Just teleport to the surface to heal.
• Hawk Blazon - Useless and outrageously expensive. It sets your AC to -10, which is as low as it can get. You'd have the same effect for less money by buying the best armor and shield and casting Shadow Shield which lasts the entire trip.

Last call for inclusion in Cranford's follow up, Shadow Snare The Bard's Tale! The guild roster of Skara Brae's chapter includes so far Scribe as a magician, P-Tux as a hobbit rogue, and Viila as a to-be-determined class with a death wish. To be clear, you won't be actively participating; you'll just be represented ingame, at least for as long as your pragmatic and pitiless patrol captain deems your person useful.

Maze Master: Diminishing returns

Level 4 has a lot of empty space in it.

 

Level 3 is mapped out, and as I've cleared it of weaker encounters, my experience level has steadily risen. Monsters that used to be total party kills remained challenging, and towards the end I still could only endure one or two fights before being forced to retreat to the surface, but it's a noticeable improvement from when I started this level.

Gold XP Base A Wolves 34 12 B Orcs 36 12 C Lurkers 38 12 D Ogres 40 16 E Werewolves 48 16 F Maze Shadows 50 16 G Trolls 44 20 H Ogre Magi 54 20 I Mercenaries 58 20 J Stone Giants 56 24 K Wights 62 24 L Fire Dragons 60 28 M Spectres 66 28 N Green Dragons 72 32 O Gorgons 78 32 P Fire Giants 74 36 Q Clue R Stairs up S Stairs down

The XP rewards have gotten better as well, letting me level up every two or three sessions. My characters' core stats are maxed out, except for Fred's intelligence which he wasn't using anyway. The wizards' spell points have gotten high enough that I can finally afford to cast the AC-boosting Shadow Shield magic before combats, though combat boost spells still seem not worth it when group-targeting damage spells almost always kill half the mob or more.

 

Next, as I map out level 4, with 42 spell points between my wizards, the loop is a little different.

1. Enter the maze and cast Shadow Shield (5 SP).
2. Teleport to an unexplored portion of level 4 (5 SP).
3. Cast Cat's Eye (2 SP).
4. After any fight, teleport back to the surface. Be quick about entering coordinates - taking too long can trigger the infinite monsters bug. If I haven't got enough SP to teleport back, then just explore the level suicidally.
5. Restore any dead characters.
6. Update the codes of the survivors.

 

This is 17 points for utility magic, which leaves 25 for combat. That's enough for four firestorm spells, and any combat encounter that can't be cleared with four firestorms isn't going to be winnable with any other tactics. There's no reason not to return to surface after each combat either; a return is a free heal and SP recharge. Continuing just means your next encounter might be a total party kill that erases your gains.  If I can win a fight with just two firestorms, then I might press my luck and keep exploring until I fight another, but otherwise, it's one fight per session.

Survivable combats get me an average of 250XP/character, so I'm leveling up basically every four of these microsessions.

I've also bought everyone a $1500 Ring of Accuracy. Not the most useful upgrade, as magic is my main damage dealer, but meaningful equipment upgrades past that start at $4000 now, which will take some time to earn; combats down here are earning me $60/character on average.

Other items remaining include:

• $4000 - Wrathblade. The ultimate melee weapon.
• $5000 - Staff of Light. No description, but sounds useless if it doesn't do anything that Cat's Eye doesn't do.
• $5000 - Amulet of Healing. Also no description, could be great, could be useless, depending on how much healing it does.
• $6000 - Mithril Coat. The ultimate armor.
• $10000 - Hawk Blazon. No description.

 

You can't pool or transfer gold, so every character has to save up on their own.

 

Even at the end of the level 4 expedition, with over 28,000 XP accumulated and only the floor's strongest monsters remaining, my success or failure against them depends on luck. If I get to go first, then I thin their ranks with a pair of firestorm spells and likely finish them off unscathed the next turn. If they go first, then whether I survive the first turn or not is entirely up to the RNG deciding how many of them hit and how hard.

Level 4 is now completely mapped, and each character has over 30,000 XP accumulated.

One unique floor feature here is that the lower-left corner contains unending monster encounters, but this is not a bug.

Gold XP Base A Maze Shadows 50 16 B Ogre Magi 54 20 C Mercenaries 58 20 D Stone Giants 56 24 E Wights 62 24 F Fire Dragons 60 28 G Spectres 66 28 H Green Dragons 72 32 I Gorgons 78 32 J Fire Giants 74 36 K Ogre Kings 80 36 L Minotaurs 84 40 M Black Dragons 100 40 N Black Wolves 92 44 O Samurai 112 44 P Ice Giants 108 52 Q Clue R Stairs up S Stairs down T Infinimonsters

 

One floor left. The Balrog awaits. Hey, at least it's only five levels and not fifty.

Maze Master: Destiny personified

Even with over 6000 XP, this still happens.

I've settled into a gameplay loop.

1. Enter the maze and cast Cat's Eye.
2. Explore and map an unexplored portion of it.
3. Return when I run out of mana or if anyone dies - often after just one fight.
4. Restore any dead characters.
5. Update the codes of the survivors.

 

Leveling up is strange. There are no 'levels' per se, but for every 1024XP earned, you improve a little bit. One of your core stats (Str/Int/Dex) is randomly picked to go up a point, but if it's already at 18, you're out of luck. Constitution, which functions as your maximum HP here, always go up, usually by multiple points, and maxes out at 255. Fighters get a bit better at combat. Wizards never get any better at combat, but gain one additional spell point per 'level'. Right now, my wizards have 7 SP each, which lets them cast Cat's Eye, one moderate damage spell, and one major damage spell per trip between the two of them. Just another 2048XP, and they'll be able to cast Cat's Eye, one moderate damage spell, and two major damage spells per trip!

The result is kind of an inverse power curve. You improve at a linear rate, but the XP requirements are flat. Deeper dungeons have tougher monsters and bigger groups of them, so as you gain enough power to take them on, the number of fights it takes to level up gets smaller, not bigger. On dungeon level 2, I'm earning an average of 80XP per character per fight, compared to 20XP when I started exploring level 1.

Bigger groups also mean more chances of someone getting killed in the first round. If a group of seven pirates - not even an especially tough monster type - gets to go first, there's a chance that someone will get hit twice and die, or get hit with a critical blow and die. And there's nothing you can do to mitigate this except to keep your sessions brief (which I am) or to cast AC-reducing spells ahead of time, which seem prohibitively expensive for how little they do. One such spell effectively makes the wizard casting it useless for the rest of the session.

Another reason to keep sessions brief is that the longer you stay in the dungeon, the more you risk encountering a wandering monster, which is bad because it can trigger a bug that deletes the stairs to return home. At least this doesn't happen much if you always keep moving.

If someone dies, this isn't a big deal, unlike Wizardry. "Restoring" characters is free; you just create a new one and enter their last code (which is why you always note their updated code when returning to town). It's also more efficient to farm XP this way; when there are two survivors, XP per-character is doubled, and when there is only one survivor, the XP reward is quadrupled. One time, a group of six trolls flattened Fred and Jin instantly, but Houdini countered with a firestorm spell, earning him 560XP.

And if you suffer a total party kill, just restore all of your characters and avoid that part of the map until you're stronger.

In combat, Maze Master provides a much smaller variety of spells than Wizardry, but has given me little reason to use anything but group-damaging spells. The fighter hits hard, but can only kill one monster per turn. Big groups of monsters, as mentioned, can do lethal damage to a healthy party in a single round. Armor-boosting spells may provide some protection, but you know what else does? Just killing the whole group with a firestorm spell or two. Monsters never do anything but hit anyway, and they never resist damage spells. Maybe this will change later on, when my party is strong enough to withstand an initial round of melee combat, and the monsters are strong enough to survive an initial volley of offensive magic, but for now it seems pointless.

As I publish this, I've mapped out the first two levels and am halfway done mapping out the third. Everyone has over 8000 XP, and the gameplay loop hasn't changed much, but now the wizards have enough SP that we can cast Shadow Shield (-2 party AC for the session, non-cumulative) ahead of time and have enough left over to squeeze off two firestorm spells, provided we live long enough. Shadow Shield doesn't do a lot, but it's the only way to be proactive against surprise attacks. Fred can hit twice per round now, and equipment upgrades are few and far between, as gold rewards don't scale at the same rate that XP does.

The tougher monsters on level 2 still have a good chance of killing someone, especially the Ogre Magi, who seem to hit very hard and accurately relative to their XP worth. Fire Dragons are lucrative targets, awarding 140XP per character if you kill a group of four. I wish it were more. Combats on level 3 give about 150XP on average, but the tough ones are usually total party kills - which I don't mind, as I'm still making progress mapping it out, and it does not take long to restore the party and walk back down to it.

 

Level 1 map:

Gold XP Base A Kobolds 14 4 B Thieves 16 4 C Goblins 18 4 D Scavengers 18 4 E Dwarves 22 4 F Rogues 24 4 G Skeletons 26 4 H Warriors 30 4 I Pirates 32 8 J Zombies 32 8 K Bladesmen 34 8 L Berserkers 40 8 M Wolves 34 12 N Orcs 36 12 O Lurkers 38 12 P Ogres 40 16 Q Clue R Stairs up S Stairs down

 

Level 2 map:

Gold XP Base A Rogues 24 4 B Warriors 30 4 C Pirates 32 8 D Zombies 32 8 E Bladesmen 34 8 F Berserkers 40 8 G Wolves 34 12 H Orcs 36 12 I Lurkers 38 12 J Ogres 40 16 K Werewolves 48 16 L Maze Shadows 50 16 M Trolls 44 20 N Ogre Magi 54 20 O Stone Giants 56 24 P Fire Dragons 60 28 Q Clue R Stairs up S Stairs down

Game 451: Maze Master

 

Our next whale, Michael Cranford's The Bard's Tale, has a predecessor.

When The Bard's Tale came out in 1985, originally on the Apple II, there were plans to port it to several computer systems, including the Commodore 64 which had comparatively few RPGs at the time, and didn't even have a version of Wizardry yet.

It did, however, have Cranford's first game as a lead designer - a Wizardry clone, Maze Master. The Bard's Tale would expand on the Wizardry formula, but this one is greatly pared-down to fit a 16KB cartridge format.

I'm not going into this completely blind - specifically, I already know that it has an absolutely awful bug which can be averted through a bit of system abuse. As in Wizardry, each dungeon floor has a set of stairs leading up, but hitting random encounters can lead to a situation where infinitely spawning monsters block the stairs, making your return impossible. The trigger seems to be fighting a random encounter right after hitting a non-encounter square, so I will need to be mindful to avoid that.

To even the odds and preserve my sanity just a little, I intend to abuse a system which is also there because of technical limitations. Your characters aren't stored on the cartridge; the system doesn't support this. Instead, character sheets contain a 21-digit code, a two-way hashcode of their state, which can be entered during character creation to "restore" the character. If someone dies, I can simply "restore" their last healthy state upon returning to town by re-creating them with said code.

 

Character creation is the first simplification; your party has room for three characters, and only two classes exist; warrior and wizard. The manual advises a fighter be in front, so I aimed for a party with one warrior for protection and two wizards for maximum firepower, and rolled about twenty of them before settling on three that I liked best:

• Fred the Fighter, 18 strength, 6 int, 13 dex, 17 con
• Houdini the Wizard, 18 strength, 18 int, 8 dex, 14 con
• Jin the Wizard, 12 strength, 18 int, 10 dex, 18 con 

 

I restored these characters, bought some basic gear, and went into the maze.

 

I explored a bit, first casting the cat's eyes spell to increase visibility, and almost immediately got total party killed by a small group of rogues.

 

I soon reload the characters and try again.

 

Some observations as I explored and mapped out the maze:

• Graphical responsiveness is nearly instant compared to Wizardry.
• There's a realtime element to exploration. Light spells expire after a set number of minutes, and random encounters can occur without your input.
• Fixed-encounter squares always have the same monster types. The door directly to the north of the initial stairs always leads to a fight against rogues, for example.
• Combat is quite brutal early on. Monsters can easily deal 10 damage per hit, nobody gets more than 18 HP to start with, and surprise attacks aren't uncommon. Even with this rather powerful party, I can lose people in the first round of fighting. Every expedition to the dungeon consists of one fight, where I unload my best spells and immediately return to town once it's over.
• Monsters in fixed-encounter squares stay dead forever, even when you return to town, or even after a total party kill. Only restarting the system brings fixed encounters back.

 

On combat rules:

• Each character may spend their turn attacking with their weapon or casting spells. Despite what the manual says, there is no option to escape from combat once it has begun.
• Wizards have access to all spells from the start. Only your mana reserves limit the possibilities - Houdini and Jin have 4 points each, and the most high-end spells cost 6.
• Spells are cast by entering a "spell number" from 1-18 rather than a spell name. You need the manual to know what spells are available, what they cost, and what they do.
• All of the combat spells either do damage, lower party AC, or raise hit accuracy, in varying increments.
• There is no way to query remaining spell points during battle, and if you try to cast something when you're out of them, there will be no feedback. Your wizard just attacks instead.
• Gold amount awarded is determined by monster type, not monster quantity, and everyone gets a fixed amount. E.g - defeating kobolds gives everyone $14, no matter how big your party is.
• XP awards are convoluted. Start with a base number determined by monster type (typically 4-12 on this level). Multiply by number of monsters, plus 1. Double the result if you have only two party members left, quadruple if you have only one. Algebraically, it's [Base]*(Monsters+1)*(2^[3-Survivors]).

Progress is pretty slow; combats yield an average of 20XP per encounter for each character, and it takes 1024XP for a level up. As of this writing, I have begun to accumulate enough gold to buy some second-tier equipment, but we're barely halfway to our first level up, and floor 1 is already cleared of the weaker monsters.

And the medium-strength ones can one-shot me.

 

It might be time to reset the game but reload my characters, so I can beat up the weaker enemies again.

 

A question for readers - on this post, I have set my images' aspect ratio to 0.75:1 PAR, which accurately reproduces the C64's NTSC aspect ratio, but this comes at the cost of pixel clarity. This might not work correctly if you're viewing on a mobile phone and there's nothing I can do about that. If they do appear aspect-corrected, you should be able to click the images and see them rendered with square pixels.

My question is, what do you prefer? Images with perfect, square pixels, or images with fuzzy pixels accurate aspect ratios? I can't have perfect clarity and correct aspect ratios unless I also make the images much bigger, which I definitely don't want to do.

 

Level 1 map (so far):

Gold XP Base A Kobolds 14 4 B Thieves 16 4 C Goblins 18 4 D Scavengers 18 4 E Dwarves 22 4 F Rogues 24 4 G Skeletons 26 4 H Warriors 30 4 I Pirates 32 8 J Zombies 32 8 K Bladesmen 34 8 L Berserkers 40 8 M Wolves 34 12 N Orcs 36 12 O Lurkers 38 12 P Ogres ? ? Q Clue R Stairs up

 

Game 450: Space Harrier

 

We’re jumping a bit ahead of schedule as this past weekend, I had an opportunity to play 1985’s Space Harrier on an original deluxe motion-cabinet at Funspot, NH, and wanted to do a writeup while it was still fresh in my mind.

Space Harrier was never my favorite Sega game, though my experience with it until now had only been through ports and emulation. It seemed quite repetitive, with little to distinguish the moment-to-moment gameplay of its 18 stages other than visual variety, the pseudo-3D shooting action clumsy, imprecise, and full of unfair deaths, and its much-praised Super Scaler technology just felt visually incoherent, providing nothing to look at but a vast, empty void with a large number of smoothly-scaling objects flying at you at warp speed. It's very colorful, sure, but the monochromatic wireframes of Star Wars and even Elite offered a far better sense of 3D space and perspective.

 

This deluxe sit-down cabinet, similarly to the deluxe cabinets of OutRun and After Burner (which also feature at Funspot and are placed right next to it), tilts and pivots as you push the flight stick to move your character around the screen. It's definitely an added dimension to the trippy experience that emulation can't replicate, but it's not quite as immersive in this early iteration. OutRun's moving seat bounces as you drive across bumpy terrain and physically leans into your turns, and After Burner's moving cockpit syncs with the chase-camera perspective, tilting and pitching as your F-14 Tomcat does, but here, the pseudo-3D perspective isn't tethered to your character's movements as he zips around the screenspace, and the cabinet motion doesn't feel as connected to the action as it might have. One suspects it might have made more sense in the first prototypes where you actually flew a harrier, instead of a guy with a jetpack and a space bazooka.

I played through Space Harrier's 18 stages for the first time at Funspot, and while the motion cabinet doesn't do anything to enhance the gameplay, it does enhance the experience. I'll note that I'm very grateful for the tertiary fire button located physically on the cabinet, as repeatedly squeezing the joystick trigger gets very, very tiring. I then played through for the second time at home, using a USB flight stick to steer and the left-ctrl button on my keyboard to shoot. The game has no limit on the number of times you're allowed to continue, and no penalty when you do. Therefore I had no incentive to replay and try to play more optimally or use fewer of them.

The game controls quite well with either setup, but is still very shallow and very repetitive. Each of the 18 worlds has a distinct color scheme and visuals, but it makes little difference whether you're shooting at giant Moai heads while dodging their fireballs and avoiding stone pillars on the ground or you're shooting at harriers while dodging their missiles and avoiding steel towers on the ground. Basically, never stop moving, never stop shooting, and try not to move into oncoming obstacles.

Space can be a tough place, and you will eat flaming laser death.

Bosses, at least, provide a bit of gameplay variety, but they're pretty brief and a lot of them are still samey. There are also two rounds where you ride a Luck Dragon and just try to smash through as many obstacles as possible for bonus points, and the final stage is a boss rush and considerably easier than the stages before it. There is, of course, no plot and no context at all for why you're here or what you're doing.

I will give some credit here, for lack of a better place to mention it. Space Harrier's audio design is quite excellent, with a psychedelic FM soundtrack, clear voice samples (AHHHHH!!! .... get ready!), lots of rumbly explosions, zippy cannonfire, a distinctive 'bloop' sound of enemies unloading clusters of fireballs in your general direction, and convincing stereo separation effects behind it all.

GAB rating: Average. Another cutting edge Sega game pushes their signature sprite-scaling technology even harder than Hang-On did, but colorfully trippy visuals and break-neck speed can't make up for overly simplistic and monotonous gameplay, and unfair quarter-munching design.

Superauthenticity: Computer game aspect ratios

C64 pixel art demo by 8-Bit Guy: "How 'oldschool' graphics worked Part 1"

 

Pixels weren't always square, not even to the extent that you can say pixels were rectangular. Pixels could be fat, pixels could be thin, but it wasn't until about 1996 that many major computer games began to adopt the modern convention of square pixels, and not until 1998 that all of them did. For consoles, it took much longer, with the PS2, original Xbox, and even early non-HDMI models of the Xbox 360 employing an anamorphic fat-pixel format to produce a widescreen image in the games that supported this. Even to this day, the venerable DVD format sees some consumer use and uses nonsquare pixels for both 4:3 and widescreen videos.

This poses a nontrivial problem for modern compatibility, albeit one that's been solved in myriad ways. That SNES game with pixels that are 1.14 times wider than they are tall doesn't translate easily to a modern square pixel display! Even in the early days of emulation, when CRT monitors were common and capable of rectangular pixels, they did not necessarily support the modes that matched the original system's pixel aspect ratio.

Another convention we take for granted today is that the source image dimensions will match the screen's. Televisions and monitors in the 80's and 90's had a 4:3 aspect ratio, and it's typically assumed that any game's output format will match this - even emulators usually assume this. This is a somewhat faulty assumption; the final picture could wind up being larger than the television screen, leading to pixels being cut off at the edges. This is known as overscan. Or perhaps the final picture is smaller than the screen, and uses only a portion of the real estate. This is known as underscan. Overscan tended to be more common with consoles, underscan more common with computers, but either way, it's possible for the final picture to have an aspect ratio other than 4:3, and it's erroneous to assume that every game "targets" 4:3.

Actually, this helps illustrate the distinction between my authenticity and superauthenticity concepts. A more authentic experience would preserve overscan/underscan on systems that had them, but anyone playing on a modern display probably doesn't want this, and in situations where we do, it's better that we are able to control it precisely, which was not possible on most consumer televisions.

Regarding pixel aspect ratio, the simplest solution is to simply render your pixels as squares. The picture dimensions will be distorted, but you completely avoid introducing any kind of detail loss or artifact introduction. For blogging purposes, this is what I typically do - just as I consider pixel shaders to be inappropriate for documentation purposes, so is image scaling and/or preprocessing. I consider it a failing on the part of digital imaging technology that no widely supported image format even has an option for non-square pixels.

Square pixels

Thin pixels, nearest-neighbor scaling

Thin pixels, linear scaling

Thin pixels, cubic scaling

For actually playing the game, though, we have more options. Thumbnail-size screenshots are fine for blogging, but when I play games, I like my playfields nice and big. Modern displays might be limited to square pixels, but modern displays have a lot of them. With a bit of math, we can approximate the correct pixel aspect ratio by scaling the image integrally but unequally. In this case, 300% on the horizontal and 400% on the vertical does a decent job of getting us large, thin pixels and a pleasing end result, and "only" needs a display of 912x800 to work losslessly.

If you want it bigger than that, bicubic scaling is your friend and will get you whatever output resolution you want without compromising those pixels. This two-step approach is how I make my videos when the emulator allows it; record raw, scale to approximate aspect ratio, and bicubic scale the rest of the way.

1440p; click to enlarge

 

With CRT shaders, or even just using bicubic filtering all the way, we aren't even trying to render the original pixels as rectangles, and aspect ratios can be scaled arbitrarily in one step without needing to worry about integral scaling. It is no problem for crt-royale to scale a 320x200 source image to a 1440x1080 one (450% horizontal and 540% vertical), but the bigger your output resolution, the better your results will be.

crt-royale

 

Bicubic

These days, most emulators presume a 4:3 display aspect ratio, which generally gives you a close enough pixel aspect ratio, and default scaling generally looks good enough. A casual user doesn't really need to worry too much about it, but I feel there's value in giving it some thought. Authenticity demands that pixel aspect ratio be approximate to the originals - not necessarily exact, as exact isn't a thing in an analog television world. But superauthenticity allows that sometimes things can be even better than the originals if we don't do that. Often, the artists were working with square pixels, and while some took the system's pixel aspect ratio into account, I believe many of them didn't, perhaps reasoning that the CRT stretch would be tolerable, and not worth the effort to optimize for. For such games, 4:3 will still be more authentic - that's more or less the aspect ratio they were always played at - but nonstandard aspect ratios could indeed be superauthentic.

But the only way to know is to look at them on a case-by-case basis. I'll be presenting some examples on various computer systems for the rest of this post. You'll have to disregard the fuzzy filtered look; it's inevitable without greatly enlarging the picture, and we're only interested in scaled proportions here.

Please don't proceed if you're on a mobile device. It will break scaling and probably be a miserable viewing experience regardless.

 

Apple ][

Most Apple ][ games run at "high res" 280x192 resolution, with thin pixels. That's an oversimplification, and technically wrong when discussing a color mode, but for our intents and purposes we can make this assumption. I can't find definitive information on what the correct pixel aspect ratio is, but I know that it purposefully underscanned the image to ensure that the entire video signal is seen to the user.

The numbers I've seen range from 0.84:1 to 0.91:1, the latter of which would put the image in approximately 4:3. We'll take a look at three ratios, then; square, 4:3, and the low-end.

 

Ultima IV

Square pixels	4:3 display	0.84:1 pixels

The tiles are already taller than they are wide to begin with, and they only get taller as the pixels do. But maybe the tiles are supposed to be tall? Well, the moons are also perfectly circular in the square pixel display, but I have to defer to subjectivity; the humans look most proportionate at 4:3.

Verdict: 4:3 display

 

Lode Runner

Square pixels	4:3 display	0.84:1 pixels

It's kind of a wash here. Tiles are 10x10px, so square pixels makes mathematical sense, but I feel the 0.84:1 pixel ratio looks best.

Verdict: 0.84:1

Karateka

Square pixels	4:3 display	0.84:1 pixels

 

Taller looks better, and I'm not just saying that because Akuma's symbol is at its most circular, but it helps.

Verdict: 0.84:1

 

Wizardry: Proving Grounds of the Mad Overlord

Square pixels	4:3 display	0.84:1 pixels

 

I hardly think it matters here. I guess the door looks better as the ratio gets taller, and the text marginally nicer as well.

Verdict: 0.84:1

 

Choplifter!

Square pixels	4:3 display	0.84:1 pixels

 

The moon is at its most circular at 0:84:1, and subjectively I think it also looks best.

Verdict: 0.84:1

 

The Oregon Trail

Square pixels	4:3 display	0.84:1 pixels

 

There's not much we can use to objectively measure, but I think 4:3 looks best overall. All three look fine, honestly.

Verdict: 4:3 display
 

The Bard's Tale

Square pixels	4:3 display	0.84:1 pixels

 

A sneak preview of my next game? Maybe. This one's tough but I think 0.84:1 looks the best, though the font looks best with square pixels.

Verdict: 0.84:1

 

This end result surprises me a little bit, but most of the time, the tallest picture, which has a display aspect ratio of 5:4, looked the best to me. I think it indicates that most Apple ][ graphics work was done locally on these systems, though there could be a discrepancy in how aspect ratio was handled on monitors and televisions.

 

TRS-80

I'm going to be honest, I don't really understand how the graphics on this system work, and I'm also not 100% confident I know what I'm doing when I emulate this system. I know that this is basically a text-based computer with three display modes; a 32-column text mode, a 64-column text mode, and a 128x48 bitmap mode. In MAME, raw snapshots of the first mode come out as 192x192, and raw snapshots of the other two as 384x192. Running them, though, normalizes all of them to a 4:3 aspect ratio at default settings.

Assuming MAME is being pixel-accurate, we can calculate that characters are 6x12px, and are comprised of six glyphs at 3x4px each. "Bitmap" mode likely works by addressing each glyph (not pixel!) individually, which is why text can appear in graphical games.

 

Zork I: The Great Underground Empire

4:3	2:1

Zork already complicates things, because I couldn't get it to run in MAME unless I emulated a Model III, which uses a different display mode from the period-correct Model I. The game uses a 64-column display, but while MAME renders that with 6x12px characters for a 384x192 display, SDLTRS uses 8x24px characters for a 512x384 display, which is already at a 4:3 display aspect ratio. I don't really know which is more accurate. Like I said, TRS-80 emulation is still a bit mysterious to me.

In this example, though, 4:3 looks honestly just fine.

Verdict: 4:3 display

 

Adventureland, BASIC version

4:3	2:1

 

Here, with MAME emulation, the ALL CAPS display just looks better in 2:1. Authentic? Maybe not. But it does look improved.

Verdict: 2:1 display

Adventureland, ASM version

4:3	2:1

 

Yep, this later, mixed-case version still looks better in 2:1.

Verdict: 2:1 display

 

Dunjonquest: Temple of Apshai

4:3	2:1

 

"Graphics" look better at 4:3, text looks better at 2:1.

Verdict: 4:3 display

 

Crush, Crumble, & Chomp! 

4:3	2:1

 

This game is visual chaos at any aspect ratio, but the dimensions do seem more natural at 4:3.

Verdict: 4:3 display

 

I can't really make any conclusions except that the TRS-80 is weird.

 

Commodore PET 

Another text-based system with pseudographics. I don't have a whole lot to say about it - the only display mode is 40-columns. VICE emulates this at 360x258, which includes a border. Strip it and we have a 320x200 active pixel display. I am guessing that the actual pixel aspect ratio is 43:45.

Telengard

Square pixels	43:45 pixels	Borderless 4:3 display

Funnily enough, I think the final image looks best! Bear in mind, to achieve this, I massaged the source image in a way that would never occur on real hardware; I stripped out the border, which changed the aspect ratio, and then resized what was left to a 320x240 image. It would not be simple to reproduce this on an emulator either.

Verdict: Borderless 4:3 display

 

Atari 8-bit

The first computer system meant for games, the NTSC Atari 400 & 800's games typically ran at 160x192 with fat pixels. Altirra's raw output renders them as double-wide pixels and adds a thin border, giving it a 336x224 display resolution.

I'm going to assume the original pixel aspect ratio for this mode is 22:13, based solely on a confident-sounding post on Atariage.

 

Boulder Dash

Square pixels	Double pixels
1.69:1 pixels	4:3 display

 

This is a mixed bag to me. The status bar looks best with square pixels. The square elements look best at double pixels. Rockford looks best at 1.69:1 pixels. The boulders look best at 4:3. But since this game is all about boulders, I deem that the most important factor.

Verdict: 4:3 display

 

Bruce Lee 

Square pixels	Double pixels
1.69:1 pixels	4:3 display

Ehh... this is an ugly game at any aspect ratio. The status bar actually seems to use 320px/line while the playfield below uses 160px/line - the Atari can switch modes like that.

I guess I give it to 1.69:1 PAR here.

Verdict: 1.69:1

 

Archon: The Light and the Dark 

Square pixels	Double pixels
1.69:1 pixels	4:3 display

 

Chess tiles should be square. At double pixels, they are.

Verdict: Double pixels

 

Alley Cat

Square pixels	Double pixels
1.69:1 pixels	4:3 display

4:3 seems a smidge more natural.

Verdict: 4:3 display

 

Montezuma's Revenge

Square pixels	Double pixels
1.69:1 pixels	4:3 display

I think this is a situation where the sprites look best with square pixels, but the level layout looks far too squashed that way. I am imagining Robert Jaeger plotting out the sprites on graph paper and going "eh, good enough" as the ANTIC chip stretches them out. 1.69:1 best meets them in the middle.

Verdict: 1.69:1

 

Star Raiders

Square pixels	Double pixels
1.69:1 pixels	4:3 display

The Atari 400's killer app is a bit abstract-looking, but ultimately I think 1.69:1 comes off looking the best here.

Note, though, that the galactic chart's cells are perfectly square at double pixels. I am certain that this is because this was simpler to program, but the main view looks better with slightly narrower pixels, I think.

Square pixels	Double pixels
1.69:1 pixels	4:3 display

Verdict: 1.69:1

 

VIC-20 

Commodore's VIC-20 is character-based rather than text-based, and allows software-defined 8x8 characters and colors, making it better suited to video game graphics than the VIC-20. But it only has 22 columns, far fewer than the PET's 40, and has an effective resolution of 176x184.

Gridrunner 

Square pixels	4:3 display

Grid cells should be square. It's the law.

Verdict: Square pixels 

IBM PC

Even in the days of CGA, the PC had a variety of graphics modes. The most common were an 80-column text mode, a 320x200 4-color mode, and a composite 160x200 16-color mode. The PCJr and Tandy computers also had a digital 160x200 16-color mode that wasn't widely used outside of Sierra games until EGA became more popular.

 

King's Quest (PCJr)

Square pixels	Double pixels	4:3 display

4:3 just looks better. Both graphically and textually.

Verdict: 4:3 display 

The Hitchhiker's Guide to the Galaxy (80-column text)

Square pixels	4:3 display

 

The font looks a bit nicer scaled to 4:3.

Verdict: 4:3 display

 

Alley Cat (320x200)

Square pixels	4:3 display

 

This conversion matches the Atari's layout and dimensions very closely, but the redrawn elements (the holes, the furniture) seem to look better with square pixels.

Verdict: Square pixels

 

ZX Spectrum

A typical ZX Spectrum game has an active display of 256x192 pixels, and a border of 320x240 pixels. No need to overthink this - the Spectrum is a square-pixel beast that targets 4:3 screens.

 

Commodore 64

The C64's default mode is a 40-column multi-color character display which comes out to 320x200 resolution, plus a border, which is taller in PAL models than NTSC. VICE has an option to crop out this border entirely, although a lot of games will look too tightly framed without it, and some programs (mainly demoscene stuff) even manage to draw to the border through some voodoo.

NTSC has a pixel aspect ratio of about 0.75:1, and PAL is approximately square. This would mean that conversions from one to the other either needed to have their pixel art redesigned or else be distorted, and I doubt that the former was common.

 

Ghostbusters 

Square pixels	0.75:1

 

Neither one of these is quite circular, but the 0.75:1 PAR (which is taller than 4:3) comes closer.

Verdict: 0.75:1

 

Winter Games

Square pixels	0.75:1
4:3 borderless display

 

Square pixels looks too wide. 0.75:1 pixels looks better, but honestly looks a bit too tall. So I threw in a simulated 4:3 borderless, and this looks about right, but it isn't authentic.

Verdict: 4:3 display

 

Spy vs Spy

Square pixels	0.75:1

 

No question here on which looks better.

Verdict: 0.75:1

MSX

Games typically run at a maximum of 256x192 with a 320x240 border, like the ZX Spectrum. This is a 4:3 square pixel system.

Screenshot by Mobygames

 

BBC Micro

Graphical modes are 640x256, 320x256, and 160x256, and horizontal resolution can be changed mid-frame.

Elite 

I don't think this needs to be overthought too much. The active display is 256x256 with a 320px border, and it looks perfect with square pixels. Trying to "correct" it to 4:3 would just cause distortion.

 

NEC PC-88 

PC-88 games typically run at 640x200 with interlacing. I much prefer double-scan, but this is never the default emulation setting. Either way, a raw screenshot should be 640x400.

 

Thexder

Square pixels (double-scanned)	4:3 display

 

Square pixels actually look better! Squares are square, circles are circular, fonts look fine. I guess the giant robot looks a bit squat, but overall I prefer the one on the left.

Verdict: Square pixels

 

Hydlide

Square pixels (double-scanned)	4:3 display

 

Without much graphical detail it's hard to judge but I think I still prefer the square pixels, although some elements, like the Hydlide logo and some of the fonts, do look better in 4:3.

Verdict: Square pixels

 

Romance of the Three Kingdoms

Square pixels (double-scanned)	4:3 display

 

The square pixels look better to me, especially the map of Han-era China.

Verdict: Square pixels

 

Macintosh 

There's a thing I don't quite get. The Macintosh reportedly used square pixels, but the original 128KB model's screen resolution was 512x342, or about 1.5:1. Was that the aspect ratio of the screen? In photos, it doesn't look so. Later models upped the resolution to 512x384, but "square dots" was a design doctrine from the very first model, supposedly.

I'm not going to really analyze this, but I'm going to show a thing related to aspect ratio that I think is kind of cool. Mini vMac can emulate a widescreen Mac! 

Authentic? Absolutely not. Usefulness? Limited at best. But I like the fact that this is even possible.

 

Conclusions - computer game developers usually designed their art around pixel aspect ratios, but it isn't consistent. You'll see art that accounts for aspect ratio, art that doesn't account for aspect ratio, and art that accounts for aspect ratio incorrectly, sometimes all in the same game. The default 4:3 display correction that emulators tend to assume isn't always correct, but it's usually close enough, and almost always preferable to no aspect ratio correction, but there are exceptions.

Also, those C64 display borders are THICK.