I usually give my "I don't care if it's AI or not as long as it's good." opinion, except when it comes to modifying someone's hand crafted art. (both as a single piece or as a larger corpus.) It comes across as disrespectful if they haven't gone out of their way to seek the original artist's blessings.

This one floated around the office slack yesterday thanks to that attention grabbing chain chomp.

Lore accurate Mario Chain Chomp

@Thad a lot of software has been a buggy mess long before the dawn of AI. It's not a replacement for good discipline in development. There's just way more of it now because the barrier to entry is so low.

We were lucky before that software had such a high barrier of entry via the effort required to do as a hobby that it naturally gatekept against low effort slop. AI just removed that.

So coincidence? No, but understanding the actual causes is more important than passing judgement with an opinion your favorite YouTuber gave you.

Just like I don't care if my oranges are non-GMO or organic, I similarly don't care if a decomp is AI. If the author puts in the polish, then the means really don't matter to me.

I took a deep dive into this code and can confirm that it is highly likely to not be written with AI.

She's the real deal folks. Good work here.

For an AI assisted port, this is really clean imo. There's some trademark python tooling sprawl, but honestly it looks well guided in all.

Reading the write-up on the ROM is really weird, like as a technical person myself, stuff like this:

"Point your very modern phone at this very old screen and it jumps straight to the full entry, photographs and all, on the website."

I've read so much chatgpt generated documentation in the last year. There's something about this text that has that fluffy grindiness of over describing each element of a simple statement. Plus a bunch of other technical details to emphasize the technical complexity of parts that either aren't that technically complex or like it's trying to "how do you do fellow kids" into the niche.

This is weird comment to leave, like not lost on me lol.

@slider1983 that's kinda where I sit too. Honestly I thought it was pretty neat and I'd like to see where it goes.

Back in the day, pre-AI, I created a real time ray traced flatland demo with some complex-ish lighting (point+global) and some fun trope-y rendering items like wavy glass and complex shapes (for some rad shadows). It ran in a gutted SMW cart shell on a custom PCB with an FPGA. Very chintzy but it landed my goal of making my "ultimate enhancement chip" I've wanted since I was a kid. There are YouTube videos of, again pre-AI, FPGA carts doing the same thing.

This also does roughly the same thing, but with the SuperFX. This is an area I've wanted to see people do neat stuff on, AI or not.

He's certainly entitled to his opinion, and to that end, I really wished more people who comment or wrote articles here were makers who contributed more to communities directly, because at least they'd be able to build more nuanced opinions of topics like, but not solely, AI. The opinion from the maker community is a lot more murky and hard to characterize than those looking from the outside-in who tend to have hard, strict, one-sided opinions.

As an example, I use AI for bug reproduction. Exercising tough to replicate bugs can be exhausting and nobody is going to help unless you're a big established project. Devs here who regularly triage will know exactly what I mean by that. Does that make my project slop? I surely don't think so. And yet, I can pick out a developer without knowing their background usually by their thoughts on that. Whereas someone looking in generally already has a fully formed opinion that's hard anti AI and full of shame or full pro AI and also full of shame. In both cases, entirely unconstructive and out of their element.

I agree with the camp against slop code written entirely by AI. It's a recipe for uninspired, unsafe, unoriginal, barely supported junk. But I do support it in areas where it's used as a measured tool for honing an already beautiful piece of work and keeping it clean and pristine. Does the overall community outside of the maker space have the appetite for a nuanced take like that? I'm not sure it does.

@jason160uk Yes, you're referring to the shadow mask/aperture grille. And you're not wrong that computer CRTs have a higher phosphor dot pitch. That does have some impact on the overall picture, and it is noticeable in general, but the effect of animations and blends aren't that dependent on pitch usually unless the art itself was designed specifically to take advantage of a given pitch. When you factor in video over composite or even RF, the phosphor pitch is almost irrelevant at 240p and higher. At that point your pitch is high enough to characterize the low-fidelity signal. Most of the effect you see in LCDs/OLEDs at this point is determined by the ADC/T-CON and not really the screen itself.

I made sure to test this first before I leave my monthly comment here lol.

@axelhander wanted to more or less side with you and mention that I think you've been dogpiled onto for no discernably good reason.

That said, I use a CRT (some hand-me-down school monitor from the late 90s) and I love it. The colors are vibrant and pop. They feel like they're sitting just behind the glass and I love it.

For people here saying that CRTs are needed for pixel art specifically made for it: you're conflating signal quality with the display itself. On a modern OLED, you can absolutely get the same effect as long as you use equivalently low bandwidth and noisy signals. Even on my computer CRT, I can get crisp pixels and lose those interframe blending tricks if I view it over straight RGB. Again, because it's a signal factor; not the display itself.

This has classically been a problem for people who HDMI mod their consoles. Even on a CRT, those effects diminish due to the increased fidelity.

That all being said, I have no clue at all who this is for. There are some great small production batch retro-forward-screens out there nowadays and this isn't on the list from what I'm seeing.

@slider1983 I agree. I'm glad to see people doing weird fun things like this and sharing their experiences.

Seems like a positive to me. To be honest, it sounds as if AI has been a positive on the retro community. I've seen all sorts of ports going out just this year, and honestly it kinda warms my heart. That's a whole new level of accessibility that wasn't as possible.

More of this. I like this.

@dedant you're referring to trademarks, not copyright. Not defending your copyright doesn't cause you to lose it. Not defending your trademark does cause you to lose it.

Trademarks exist to prevent confusion, so you have to defend it or why even have it. Copyright has no such burden.

I just recently completed a multi-year project developing an engine to play King's Field 1 JP. (Started on the 30th anniversary, took roughly 2 years to finish, about 20 months longer than I anticipated.) It was rewarding, but it was grueling work at times.

I wish I had something like this going in. I ended up writing the GTE and SPU logic myself to reasonably match the PSX version. I ended up poring over the internals and sites like exploring tokimeki memorial, nocash, and trying to suss out info from the PsyQ docs. If this existed when I started, I would have used it. From being in the trenches of that work, static analysis tools like this are incredibly helpful, as much as I live Ghidra and all its quirks. I think making the process of decomping something as easy as the original build-up of the game via the rich PsyQ libraries is really really valuable. Very cool project.

Didn't comment on the AI part because I don't have much to say even though it's the article's focus, other than I agree. I think going into it not wanting to learn and just wanting to port something is a weird goal since PSX emulators are quite good already.

Edit: and if I could leave words of wisdom to anybody looking to do a hand decomp, keep a good accurate emulator's source at hand. I referenced mednafen's source constantly for all sorts of HW info. Also take your time. It's a marathon, not a sprint.

@Ruka they're a private company with private investors. They can say they sold ten trillion of these and it wouldn't be a crime on its face, let alone a "serious crime." It's a cute thought though.

I'm not sure how long this gravy train of theirs will last, if it even exists in the first place. They already did "the thing" that fans wanted and then went straight into a gimmick. I agree with the sentiment that they're just a couple years from insolvency.

I can smell that translucent plastic. I haven't owned anything made if it in decades, but just seeing it made my nose wiggle a bit.

Well... Huh, anyway

My gut tells me the brand is being used as nothing more than a vanity car plate would be used.

It was a fun story the first scam, but I think this whole saga has run its course.

@RadioHedgeFund by 2000, Sega of America was just a weak zombie arm of Sega of Japan, filled with yes-men and run by leaders who themselves were SOJ's favorites.

It's not something I like saying, but there's a lot of evidence to suggest that SOJ was largely xenophobic towards SOA. (And from what I understand, that internal culture flaw still hasn't changed.) When you read through the history of management and development, there's many stories of SOJ seeing SOA as nothing but a cash cow with no inherent value otherwise.

And I wouldn't say SOA was arrogant as much as they were naive. They tried so hard to not acknowledge what SOJ really saw of their branch, and it ended up killing what was left of SOA by 1996. As much as I really don't like Nintendo through history, Sega's biggest failure was not embracing their American arm the way Nintendo did.

Is the link on "shown off in 2010" correct? It directs me to a video about the development of Lemmings, which is a cool video, but if it is correct a timestamp would be really helpful.

@KingMike it was officially licensed by SEGA with the catalog number T-47803M, so it went through all the usual certification steps. Another reason it's kinda notable is because they slid right under the deadline for the end of GD-ROM disc production. Pretty sure they rushed a bunch of copies up to the last disc pressing.

@hste the ROM on this cab only exists with a couple private collectors and a museum. If you wanted to play this specific English version, you'd need to own this cab. It's not in the wild to download like Rev A and C.