SABS Forum

it's that time of year again!!!

Quote from: RT-55J on Sat, 2026 - 03 - 28, 05:13 PMIf you want to peruse other entries to the 2026 SMSPower compo, you can follow these links:

• Coding (https://www.smspower.org/Competitions/Coding-2026) (23 entries)
• Music (https://www.smspower.org/Competitions/Music-2026) (10 entries)
• Hacks (https://www.smspower.org/Competitions/Hacks-2026) (3 entries)

Lots of fun looking stuff here.

gonna be posting short reviews to these games off and on this week, in lieu of a stream

From a cursory glance, it appears that 5 of the entries in the coding category are "demos" in the demoscene sense (rather than "demos" in the game or productivity software sense).

(My Last Minute Demo (https://www.smspower.org/Homebrew/RTsLastMinuteDemo-SMS) should be familiar to all viewers of this forum, so I will refrain from elaborating on it.)


This is a demonstration of the almost-never-used "multicolor" mode the SMS's VDP had. It was a legacy mode inherited from the SG-1000's TMS9918 video chip, but even in that context it almost never got used. It basically was a direct-color bitmap (kinda) mode, with the caveat that you were stuck with some really chunky pixels. Like all of the SMS's legacy video modes, it doesn't work on the Mega Drive.

This demo, interestingly, is meant specifically for the SMS rather than the SG-1000. It uses the SMS's color palette, line interrupts, and increased video memory to make these effects possible.

(There was a bug when I first tried it where one of the effects wouldn't animate in some emulators (if the RNG was initialized to 0x00 it would be stuck there), but I reported it and it got fixed!)


This is a simple demo demonstrating a single effect (a 16-frame animation). It looks very polished, and has a good little tune. Very nice.


This is a very simple "Hello World" style demo with two illustrations, very similar to mine. I like it. Just brace yourself for an Auditory Experience at the start :)

I hope to see more from this guy.


I hadn't seen El Viento's intro before this, but it turns out that this is an astonishingly accurate demake of it! Only like 3 effects from the original are missing!

Apparently the author of this wants to demake the whole game like this. I wish them the best of luck.

very cool jam with good turnout. i played some of these recently and there's some impressive stuff.

i really want to participate in a game jam at some point. but i think i work too slow

The secret is to find a jam that doesn't care about when you started your project. Some jams are chill like that.

This baby can be considered the "main attraction" of this compo. It's sort of the holy grail of console alt-history, and it's a pretty decent effort too. The author manually translated the original 6502 source code to Z80 assembly, and the result is astonishingly accurate! The physics are so spot-on that I was able to do a walljump with some effort.


Graphically, I'm not particularly fond of the Super Mario All-Stars style that this is aping, but I will admit it at least makes for a good demonstration of the SMS's graphical strengths versus the NES.


The demo only contains the first level, which is fine, but the author has shown some interest in completing the thing. But who knows? Maybe someday, someone will do the opposite with Alex Kidd in Miracle World...

ERRATA: The original version of this demo had very noticeable loading times for the levels. The author originally chalked it up to some weird refactoring of the code they did to get the game to work with the SMS's single nametable. I had my doubts, so I took a look at the nametable viewer when the screen was disabled, and saw that it was taking the game 5 or 6 frames to draw a single 8px wide column to VRAM. I informed the author of this issue, and they noticed that they were accidentally loading the level tileset (an operation that takes 4/5 frames) for every single column they were writing, causing the initial level load to be nearly 3 seconds instead of an imperceptible half-second. The author uploaded a hotfix very soon after that.

This is an Alex Kidd fangame in an original engine. In it, you play a Prince Egle, the less famous brother of Alex Kidd. Unlike Alex, he doesn't know the martial art of "punching", he's been cursed to be tiny unless he eats burgers that slide on the ground after coming out of boxes, and he has a lot of momentum... hey! Wait a minute!


This game is a trick! It's a secret Mareo game! All these people on this forum that still censors the unholy word of Ni****do are sure to be offended!


Anyhow, this little demo has you going through four "parallel worlds" of mareo-likes to find a plumber who can solve your plumbing problem. The controls take a bit of getting used to, and the game likes to lag in certain situations, but it's a pretty decent platforming engine at the end of the day.

I wouldn't mind seeing this turned into a full game. I can already imagine a few other worlds this guy could add to it...

We interrupt this thread to bring you and important discovery about Lock-On Technology for the SMS. (https://nicole.express/2026/april-fools-joke.html)

I just finished trying out all 23 coding entries, and the last entry I played was AI slop, so we're going into the Slop Corner tonite!! :twisted:  :twisted:  :twisted:

---

That cover art certainly is a choice, especially when the game looks like this:


Credit where credit's due, though: I pointed out a couple bugs to the author and he fixed them, which is nice.

It's a very rudimentary CVBasic game, but I don't begrudge it for existing. I'm just confused why they wanted cover art that looks like that, when classic SMS-style box art is within the artistic capabilities of every human on earth.

---

This is a nice little single-screen shmup that would be home on the Atari 2600. I liked it. It's a fun 10-20 minutes.

However, looking at the author's GitHub, apparently the title screen (and a couple other assets) were created using Google Gemini:


I guess the awful lettering at the bottom and the word "CONPETITION" should have tipped me off, but again: why do this??

Not going to bother posting pictures of this, since it's just an electron app. It has an AI generated logo in the window at all times, which makes me wonder if the whole app is vibe-coded.

Feature-wise, it's okay, I guess, but there were a couple things I couldn't figure out. Aside from allowing you to use a combination of channels, this feels like a lesser version of the lovely sn_jsfxr (https://harmlesslion.com/sn_sfxr/).

---

Any now we come to the biggest disappointment of the jam for me: this port of Akalabeth, by somebody who beat me to the punch.

At first glance it appears to be rather competent and polished. It claims to be feature complete and winnable, with just a few bits of polish straggling behind...


...but the description contains an ominous sentence:

QuoteThe code is written entirely in C and optimized using AI so that it runs on the SG-1000 with 1KB of RAM.

The game is very unstable. Issues include, but are not limited to, the following:

• The border color on the title screen flashing (if you have borders enabled in your emulator)
• The screen flashing a bright color every time you take a step on the overworld
• The game randomly hanging when you exit a shop
• The screen going completely dark if you enter and exit a shop multiple times (game is technically still playable)
• Item counts not being reset upon resurrection (including your food, which will be negative if you happened to starve to death).
• Some dungeon palettes being hard or impossible to see with:
• The video (https://www.youtube.com/watch?v=mVic-tbk-qo) on the SMS Power channel ends when it becomes impossible to see.
• The game randomly crashing spectacularly (either immediately or after making a choice) when you enter the castle (meaning you cannot even initiate the main quest).

I think I should campaign for an AI ban for next years compo.

---

Next up: Some good entries!

Road Fighter was a 1984 arcade game by Konami in the "drive to the top of the screen" genre of racing games. I had no familiarity with it before trying out this port.


Turns out that I like it! This port feels very polished and professional, so assuming that it's accurate, then the original game is rather sophisticated and refined for a racing game of this type. It's not especially deep, mind you, but it executes its core concept quite well.


Recommended without any real reservations. This could have been a commercial release back in 1986, if Konami were to have supported the platform at all.

As its name suggests, this is a proof of concept for a port of Puzzle Bobble (aka Bust-a-Move). Everybody's played some variation of this game, and it's on a bajillion platforms. It's nice. Your mom probably enjoys it.

As for myself, I like this game, but this port ends up with the same issue I have with most other versions of this game: some colors are too dang close to each other for me to distinguish (I am not the best with colors). I can tell the colors apart with some focus, but not at a glance. Hopefully some options are added to address this at some point, though with the SMS's RGB222 colorspace I don't think it's quite viable, per se.


Anyhow, as a proof of concept, this just has a single-player mode with 10 levels that repeat and infinite lives. There's no time pressure from a lowering ceiling or any of that. There's no music right now either, but there are at least some sound effects. Still, regardless of all that it seems pretty well-made. It has some features that I haven't noticed in other versions of Puzzle Bobble, such as up and down moving the cursor to or away from the center, and holding the other button allowing you to single-step the cursor. It's rather useful for getting tricky shots.

I liked this, and hope the author continues with it.

Today we're gonna talk about two games by a rather eccentric (compliment) author.

---

This game is very weird. I highly recommend that you read the author's description in the link above to get into their headspace, and then trying the game out for yourself. See what the author means by them claiming to have made "the first master system game where you can move in 3D space," and then come back to me, okay?


...have you tried it yet?

Okay good.

So, yeah, I'm not sure how that could really be considered 3D either, but it's still really interesting. Implementing an FMV engine on the SMS, and then using it to make a Simon clone is a wild choice. It's such an outsider artist move, and I love it for that.

I don't have much to say here, I'm very glad this thing exists.

---

This game is supposedly edutainmentical, but like the previous game's claim of being 3D, it kind of stretches the definition there.


The theming of the game (again, I implore you to read the author's own words) is that you're splicing DNA from a bunch of different creatures to create a proper pet animal. This entails finding a pattern of balls in the playfield on the right that match the goal in the upper left. You have to do this in a certain time limit. I can be an interesting challenge, but the fact that there are only 2 difficulty options kinda limits its appeal. The 8x8 grid makes it too easy to find a match, while the 16x16 grid is a bit too hard, especially with how tiny the icons are.

Regardless, the hybrid animals are incredible to look at, and really make the game. Again, I'm very glad this exists.

PSA: I finished playing all 3 entries in the Hacks competition yesterday, and today I submitted my ratings for the coding and hack compos. (Still need to listen to the music.)

The poasts here will continue whenever I feel like it.

Tonight we have two different abstract territory-control board games:

---

Go is a millenia-old abstract strategy game from China. Players alternate placing stones on the board, with the goal of controlling more territory. One can capture a group of orthogonally-connected pieces belonging to the other player by surrounding every intersection orthogonally surrounding the group with your own pieces. The rules are very simple, but the strategy is deviously complex. Competent Go AI was regarded as a Hard Problem in computer science for decades.


raphnet's "Capture GO" is an implementation of a Go variant that simplifies things greatly. The board is made drastically smaller, reducing the possibility space. Four pieces are already placed on the board, removing the difficulty of selecting opening moves. And the goal is simply capturing a set number of pieces, rather than continuing to the ambigious point of when neither player desires to make a turn. These choices greatly reduce the complexity of implementing a Go AI on an 8-bit system (and in general).

I assume this variant exists as a means of teaching newer players some fundamentals of Go tactics, and in this respect I think the game works quite well. I'm not an experienced Go player by any means, but I was able to complete all 5 levels of this game after some effort.

Overall, this is a very polished implementation of a classic boardgame ruleset. Kudos.

---

Atoms is, in some respects, similar to Go. However, despites having a more complex ruleset, as a game it feels decidedly messier (but perhaps this is because I have not played against a human).

Every turn in Atoms you must place a piece (atom) on the board. You may place a piece on top of one of your own pieces. An atom will explode after n pieces have been placed on it, where n is the number of adjacent cells (so 4 for the interior, 3 for the edges, and 2 for the corners). When an atom explodes, it deletes itself from existence, and places another atom of your color on all orthogonally adjacent cells. This can cause chain reactions. Other players' atoms will change to your color when caught in an explosion. A player is defeated when they have no atoms on the board at the start of their turn (after the first turn). The last player remaining wins.


When it is your turn, pieces that are about to explode are flashing. Strategically speaking, it seems optimal to cause your pieces to explode whenever possible, because that is equivalent to placing multiple pieces. However, beyond that I haven't been able to suss out many of the strategic particulars to this game, because the AI just places pieces randomly (pure RNG) and is trivial to defeat (even in a 1v5 context). Just be aware that any chain reaction of your pieces can also be triggered by any adjacent player that can start an explosion.

In addition to the standard versus mode, there is also an arcade-style "Challenge" mode, where you place pieces of a randomly selected color (think Tetris or something), and you have to meet a destruction quota for each color in a time limit. The basic strategy for this mode is to balance all of the colors so you can meet all of the quotas, but this is easier said than done, because massive chain reactions are mathematically inevitable. If you eliminate a color entirely from the board, you get a time penalty and the board is re-randomized. It's pretty weird, and I'm not sure what the optimal strategy would be.

Anyhow, this is a decently fun game, but compared to the elegance of Go it feels kind of iffy. Regardless, this game does has the benefit of allowing up to 6 players in a match (in any configuration of humans or bots), whereas SMS Capture Go is solely a man vs. machine affair. Also, the chain reaction mechanics give this strong possibilities for the kind of wild swings in placement that make for good, easily-understood entertainment.

The Sega Master System had a "3D glasses" peripheral that could plug into the Sega Card slot. They were not traditional red-green anaglyph 3D glasses. Instead, it used a couple single-element LCDs to act as shutters, blocking one eye on one frame and the other eye on the other frame. This effect was controlled by the game software, hence why it needed to be plugged into the machine.

Anyhow, this is was two sequential frames of this game look like:


If you do not have the 3D glasses, this is practically unplayable.

Fortunately, there are SMS emulators that support the 3D glasses peripheral, such as Snepulator and Dega, so this isn't too much of an issue. They can output anaglyph 3D or one eye only:


I don't have anaglyph 3D glasses, but this is much, much more tolerable.

Anyhow, as a game this is neat, but pretty rudimentary. It's a dot-eater, but you have a jump. Enemies chase you and you can shoot them, but they can also jump. The jumping action does seem to legitimately be in 3D space. If you don't finish the level in time the enemies become much faster.

There are 4 or 5 levels (I forget), after which the game loops.

Verdict: Not half bad.

This is a simple software toy for the SG-1000 that implements playback of so-called ARPABET (https://en.wikipedia.org/wiki/ARPABET) phenomes. It's basically like a speak-n-spell sort of deal.

The software presents you with 3 modes:


ARPABET SAY brings up the keyboard for you to type stuff on:


It's neat, though I wouldn't have known what sounds some of those weirder letter sequences were supposed to be for without checking the Wikipedia page (or some trial and error). One oddity is that pressing Enter to play a sound also clears whatever you've typed.

The ARPABET TEST option just has the machine say a prewritten speech about itself.

The INSULT ME option has the machine hurl insults at you. Some of them are pretty good. Some of them are vulgar. And about two that I encountered might have flown 20 years ago, but I wouldn't have included them myself.

I didn't spend much time with this, but I appreciated how robotic the voice sounded.