Chrono Trigger Remake: Classic Characters in Today’s Graphics 💥

[The Illusive Gamer]

Hey everyone,

I've been working on a Chrono Trigger characters remake for the past month and a half, and I'm excited to share the first time I ever published something like this with you all!

I am a long-time fan of Chrono Trigger and I was looking for either a video or art of characters with today's graphics since there's a lot of talk about a remake, but I failed to find them - so I took it upon my self to make them from scratch - the exception being Robo as I found what seems to be a 3D model.

Anyway, long story short, I spent the whole month obsessing over some AI tools and Photoshop, and let me tell you, AI still has a lot of ground to cover, but with Photoshop AND AI, I think I managed to do a good job.

What do you guys think?

https://www.youtube.com/watch?v=hsfxepsIFbE

This is my only video so far as only a passion such as Chrono Trigger could make me venture into this, but I really liked this obsessive hobby so maybe I'll do a part two.

What do you think? Worth it?

Also, if so, what character that isn't here would you most love to see?

I'm open to feedback and suggestions.

Thanks for watching and replying in advance!

[Rhyagelle]

AI

[Jaxom]

If we could take AI out of gaming that would be great.

[@R+5]

Originally posted by Jaxom:

If we could take AI out of gaming that would be great.

why? in games, better ai, means more interesting enemies and fights, since they could learn to avoid some patterns or switch them to be less predictable and more challenging.

ai as a design tool, could also be useful for good random generated dungeons, if you want to introduce for example a special area or "bonus rogue-like dungeon".

ai can also help to improve the quality of animations, and help reduce the amount of "manual work", which then also helps to reduce the costs of production.

the downsides of using it for art, are not at the side of obsessive artists or "perfectionists", but at the side of production trying to rush things without polishing them enough, so the mistakes done by ai could be properly ironed out.

also, ai could improve "dynamic camera movements", when following the action in battles or giving a more cinematographic style to exploring and walking (making it a bit less boring when playing long sessions).

you cannot stop the ai revolution or the use of tools in the industry, what can be done is placing rules to how and when to use them, just like cars replacing horses.

[@R+5]

imo, one of the possibilities that could be visually interesting, would be something a bit similar the dragon quest 3 2.5 hd remake, and similar games, but with the option of using actually non-pixelated "highly polished" (hand-drawn like) animated sprites (also with smooth animations), or 3d characters cell-shaded and animated in such a way they look like sprites (what guilty gear devs did), and high quality textures that follow closely the art style of toriyama and the original sprites at the same time.

watch this "concept video":
https://www.youtube.com/watch?v=yKca4X3AKR0

similar approach
https://www.youtube.com/watch?v=UJuoH2wZw7Y

similar, but kind of blurry (which i dont like), but a bit closer to what im suggesting.
https://www.youtube.com/watch?v=J3Z64OwgEfs

imo, it could work great if done similar to some 3d versions of dragon quest. i mean, visually, like the visual appearance could be like the video offered by op. but with a more "classic gameplay style" (ie, preserving or making the new combat system extremely close to the original, or recreating it to work in 3d space, like in the first video i included)

so, something like the "fake sprites using 3d models" of guilty gear, with a cell-shaded anime style, but preserving the art direction and storytelling of the original 1:1 (so no additional dialogues or ff7 remake style shenanigans).

https://www.youtube.com/watch?v=kZsboyfs-L4

[@R+5]

i guess, short version, of how the remake for chrono trigger should look like, imo

this
how arc system works animates guilty gear and dragon ball
https://www.youtube.com/watch?v=yhGjCzxJV3E

+

this
quality of art, accurate and faithful relative to the original art style; maybe even make aesthetics a bit closer to promotional artwork
https://www.youtube.com/watch?v=hsfxepsIFbE

with a bit of this
(recycling textures like in classic games, but with better "hd hand-drawn like" sprites), allowing also to make the world look closer to 2d animation, even when using effects like motion blur or depth of field.
https://www.youtube.com/watch?v=yKca4X3AKR0

[FreshMint]

Originally posted by @R+5:

you cannot stop the ai revolution or the use of tools in the industry, what can be done is placing rules to how and when to use them, just like cars replacing horses.

Maybe in 10 years things will be different (if we aren't struggling to deal with processing costs instead) but currently I see AI struggling to make videos that are longer than 10 seconds while also having long-term consistency.
And in other ways it's difficult to use generative AI for more than a source of inspiration and occasional aid.

Feeding an AI some rough sketch of what you want and retrying until you get what you want is one thing, but even in the japanese AI retelling of some SNES rpgs you can see characters details changing here and there while it quickly becomes apparent that it's just brief generated clips pasted together, with some characters rarely truly showing varied expressions.

Same goes for OP's video: Not a single character looks like it fits the same artstyle.
I personally think that if a product keeps shifting in art-style for unintentional reasons (IE that fake Dedede show in the kirby anime, or how in this game certain cutscenes come with CGs to show a scene in more dramatic effect than sprites can usually do) it comes across as unfocused and it kind of takes the overall quality down.

This is also not even mentioning playing a game that's actively being rendered by AI- Oasis being a very clear example of how, currently, we are nowhere near actually managing such a thing. (Waiting lists, everything offscreen being immediately forgotten, inventory not making sense, input being delayed as the frames need to be generated etc etc.)

Originally posted by @R+5:

Originally posted by Jaxom:

If we could take AI out of gaming that would be great.

why? in games, better ai, means more interesting enemies and fights, since they could learn to avoid some patterns or switch them to be less predictable and more challenging.

Why would 'better AI' mean more interesting enemies and fights? Would AI 'learning patterns' truly make for more 'interesting' fights? It's not like this is an action RPG... for now.

This admittedly auto-translated piece from an interview with Akitoshi Kawazu (SaGa) might be interesting;

That's right. That's because, at the current level of technology, "interesting or not" is not something that can be evaluated by AI. For example, in a battle, you can give a numerical evaluation based on the number of turns it took, but you can't calculate whether the battle was interesting or not.

"Quick resolution = interesting" is not necessarily true, and people have different values, such as "it was an exciting battle and it was interesting" or "it was fun to be able to defeat it," so it's not possible to evaluate it.

Of course, one day we may be able to make that evaluation. But I think it's difficult at this stage.

(source)[news.denfaminicogamer.jp]
(Context: AI has been used as an aid to balance out fights, help see what setups are possible and aren't, but currently trust in generative AI to 'make' fights and scenarios isn't quite there yet.)


And of course even when AI finally becomes 'good enough', there will always be people who see good games as works of art, and appreciate the work put into them. From crafting visuals that make the game scenes what they are to designing the game with player mentality in mind.
There's also an underlying fear that AI is now getting utilized by major (and commonly seen as greedy) publishers as a way to make production of games cheaper for the sake of making them cheaper, while there's still a lot of question marks about training data in some AI (as it's currently hard to prove if data has been obtained legally- or even morally).

I probably took a bit too long to make this post but I felt like typing for a bit I guess.
But TL;DR - AI can make small things that we may think are pretty (although frequently still not exactly how we want it), but can't make large things with proper consistency, whilst some people have an actual appreciation for hand-made products.

[@R+5]

Originally posted by FreshMint:

Maybe in 10 years things will be different (if we aren't struggling to deal with processing costs instead) but currently I see AI struggling to make videos that are longer than 10 seconds while also having long-term consistency.

And in other ways it's difficult to use generative AI for more than a source of inspiration and occasional aid.

Try 5 or less years:
just like the early bitcoin "gold rush" was taken with a lot of scepticism, from ignorance of how much and how easily it could grow, thanks to the technology used and original framework, today, even most the most optimists about ai are falling short of appreciating how big, and how fast its impact actually is.

two years ago, "graphics-image ai generators" in most cases, were only able to make rough sketches with many imperfections; most of them now are a lot more accurate, precise in proportions and realistic, and at the same time, the new "ai video generators", which didnt existed publicly a year ago, can do a lot better than ai image generators when they were publicly available.

and that is for "toy level ai": the kinds of ai that big companies are making, are meant to be a lot more effective to be useful at industrial level. Their impact is already big enough, that is one of the main reasons (and also rarely mentioned in the media) why ineffective and mediocre "clean energy alternatives" have been reducing their presence and sales (such as wind energy, or water turbines meant for the ocean, which not only do more harm than good, but are also more expensive and produce less energy than a nuclear plant with a modern design and materials)

Originally posted by FreshMint:

Feeding an AI some rough sketch of what you want and retrying until you get what you want is one thing, but even in the japanese AI retelling of some SNES rpgs you can see characters details changing here and there while it quickly becomes apparent that it's just brief generated clips pasted together, with some characters rarely truly showing varied expressions.

sure, but those are minor, and short term, imperfections. wait just 6 months, and you will see a significant improvement; wait two years, maybe less, and you will see the first "commercial ai pro video generators"

the ideal, imo, would be to not only make laws and regulations that make illegal the industrial use of "creative ai tools" without rl human imputs, and rl human "polishers" (ie, an individual or team focuses in prompts to generate "stock content", and another removes and fixes their "minor flaws"; then, another team focuses in "editing and mixing" those, and so on)

and imo, another great consequence of that, will be the end of traditional "celebrity cultism": the role of professional actors could begin to become appreciated for motion capture, and "models" wouldnt need acting skills (they would only need to "look good-compatible" for the character).

Originally posted by FreshMint:

Same goes for OP's video: Not a single character looks like it fits the same artstyle.
I personally think that if a product keeps shifting in art-style for unintentional reasons (IE that fake Dedede show in the kirby anime, or how in this game certain cutscenes come with CGs to show a scene in more dramatic effect than sprites can usually do) it comes across as unfocused and it kind of takes the overall quality down.

Again, those are short term limitations. Consistency and "continuity issues" are part of the same problem ai has: their models need to evolve more, and they will.

Originally posted by FreshMint:

This is also not even mentioning playing a game that's actively being rendered by AI- Oasis being a very clear example of how, currently, we are nowhere near actually managing such a thing. (Waiting lists, everything offscreen being immediately forgotten, inventory not making sense, input being delayed as the frames need to be generated etc etc.)

we are closer than you expect: as in mentioned in a previous paragraph, you are not taking into account what can affect "speed of evolution", and therefore improvement of "ai products" (and derivatives).

Originally posted by FreshMint:

Why would 'better AI' mean more interesting enemies and fights? Would AI 'learning patterns' truly make for more 'interesting' fights? It's not like this is an action RPG... for now.

It doesnt matter the genre of game, as long as the gameplay involves overcoming challenges (of any kind), and more commonly "enemies".

basic ai, doesnt and cannot "learn" things or adapt. bad vs good ai for games and simulators used to be about how good it was, to adapt to changes and react to user input, and adopt temporarily a new pattern to be "a bit less predictable".

thin a bit about the difference in patterns possible in enemies programmed for a 1990s videogame, vs any early 2000s playstation game. every version of ai was "stupid", but more effective than previous versions. Modern ai is "smart", because it can go beyond just repeating predictable patterns.

Originally posted by FreshMint:

This admittedly auto-translated piece from an interview with Akitoshi Kawazu (SaGa) might be interesting;

That's right. That's because, at the current level of technology, "interesting or not" is not something that can be evaluated by AI. For example, in a battle, you can give a numerical evaluation based on the number of turns it took, but you can't calculate whether the battle was interesting or not.

"Quick resolution = interesting" is not necessarily true, and people have different values, such as "it was an exciting battle and it was interesting" or "it was fun to be able to defeat it," so it's not possible to evaluate it.

Of course, one day we may be able to make that evaluation. But I think it's difficult at this stage.

I think you are misunderstanding what he is saying, or what it implies. i think he is talking about ai replacing a human to generate and judge itself how good a game it made is.

and sure, thats more complex and will take more time than what im talking about.

what im talking about ai in games, not only for enemies, is as different tools that still need human input, and "judges", "curators", directors, etc.

Some online games depend on having a specific minimum amount of players to be enjoyable, right? most shooters, most mmorpgs, most competitive online games. If devs could use "virtual players" ("ai actors"), and those were believable enough to dont break immersion, they wouldnt need to depend much in attracting a lot of early adopters-players. some businesses could even rent, license or offer those "ai actors" for that purpose, and maybe players could actually pay for "ai companions" in games if they cannot afford the time or want to socialise while gaming (people that have few minutes to play after working, or studying, etc)

Originally posted by FreshMint:

There's also an underlying fear that AI is now getting utilized by major (and commonly seen as greedy) publishers as a way to make production of games cheaper for the sake of making them cheaper, while there's still a lot of question marks about training data in some AI (as it's currently hard to prove if data has been obtained legally- or even morally).

Most data, no matter if you want to consider morals or not, has been taken legally, because we have been tricked many years ago to allow data harvesting each time we use search engines, or any social platform (including steam).

Why do you think the biggest market right now, or one of the biggest, is "big data"? is not only useful for politicians, weather analysis, or any other kind of research, is also useful to train and experiment with ai learning models, and related.

also, is obvious thats what they are doing, and will keep doing, because thats why companies like apple uses cheap exploitative labor to produce their hardware, where thats legal.

so laws and regulations must be made to set rules about how to use ai at industrial level; maybe something as "an ai tool-type per n rl-person(s)". and separate "tools" from "independent-type ai consultants" (aka a "virtual worker").

i think sometime ago i saw something about why robots should be made to look like humans or anthropomorphic, in order to make the legal distinctions between humans an "human-like machines" easier to set. And that topic also relates to the idea of the "virtual woker" (which again, could translate in that context, maybe, as "one worker per one business owner", ie you want to make a small business and cannot afford to pay a human assistant, but maybe you can afford an "ai worker"; that could actually make easier for more people to start their own businesses, and make more mistakes in less time, to then be able to learn what and where they need to improve, or change)

[FreshMint]

Excuse me as I'm a bit too lazy to quote everything separately, and dont want to go deep into this convo, but here's some points.

I can agree that tech can rapidly evolve in unexpected ways, but as much as we can try to hypothesise about a world where generative ai can make things just how we want it, it's not there yet so people don't embrace it that much yet.

Don't get me started on clean energy, just because a neural network used good energy doesn'r make it much better if it's greedily slurping it- it's the amounts that cause concern, not source.

While I don't doubt that ai could aid in reducing enemy predictability, I wouldn't say that's always a 'fun' thing. What if an enemy knows to handle and migitate/evade everything the player throws at it? what if someone enjoys bosses they can strategize for?
I still remember the frustrations of a boss defending when I'm about to burst in Bravely Default- not everyone will enjoy an enemy that adapts all too well, while it's also a shame to take away figuring out bosses as some people enjoy.

The interview i linked goes in depth about how AI can be useful as a help (judging if fights are even possible as to not overload real playtesters) but aren't still there yet to judge and make 'interesting' content. We indeed still need to curate what it makes and at what point isn't it just better to do it ourselves, or do procedural generation which has worked all the way back for things like Rogue and Mystery Dungeon?

I also get that having ai players can help populate games reliant on online play, but stuffing games full of bots also contributes to the dead internet theory, and not telling players they're playing against bots, ai or not, is dishonest at best.

Also yeah, big companies will always skirt the edge of whats legal to amass as much profit as possible to please shareholders with 'growth'.
But that's exactly the issue- websites like twitter and reddit decided that the data people made is now theirs to use for training, and even when you opt our who's to say your existing content hasn't already been used to train? I've yet to hear of un-training ai...
Normal people don't process millions of pieces of content at the speed ai does, is it really fair that ai now uses it in ways artists and such never agreed on?

And of course, it's nice if people can find ways to have ai help them out with their own work, but not everyone can train an ai to specifically aid them in their goals (esp not without relying on big tech industries or huge budgets).
And it's also a question on if people can handle them responsibly- look at deepfakes or the ai slop (bad images, factually incorrect information under the guise of a guide) that's flooding search engines.

Maybe things will change, but not right now, and right now some things just kinda feel more offensive to the people that sink effort and passion into their creative works.

May not want to offer much headspace to this thread any further and possinly not respond anymore.

[Ysopet]

You did no work and you suck.