FFVII Remake Game Engine

[Guitar (pseudo)God]

Thanks for the information. Very interesting. The engine issues is akin to a project I had that was an unholy union of acquisitions and legacy technologies. We had 7(!) database implementations to try and make work to win the business.

What is BD2's next project, besides the FFXV DLC (and, of course, Versus XV)?

I don't do video game development, but Luminous looked pretty good (if overspec'd) for its first iteration. I will say that optimization presentation was rather basic stuff in my world; I consider FFXV an open beta of sorts.

Is it a 5% royalty for UE4? In my experience, these things are rather flexible.

 

[BladeRunner]

What is BD2's next project, besides the FFXV DLC (and, of course, Versus XV)?

I think they will be working full-time on XV updates for some time now. Then probably some new project from Tabata, maybe Type-Next.

 

[Ikkin]

1) Yeah, I mean PS4 to PS4. But Tomb Raider certainly has the advantage of generally being much more able to scale well, which is one of the criteria by which I'd 'score' an engine. Its predilection to scale are one reason its Pro modes are so amazing, incidentally.

It seems manifestly unfair to compare an infant engine which hasn't even seen a consumer PC port to an engine that's had years to work out the kinks in that regard. We haven't even seen how well Luminous can scale when given a proper chance; my understanding is that FFXV's Pro modes were the result of giving a limited internal team a few weeks/couple of months to put something together instead of outsourcing to one of the premiere porting companies in the industry and giving them as much time as necessary. And, even so, FFXV's High mode added 1800p checkerboarding, AF, and lighting improvements.

2) Scale-wise I wouldn't say Tomb Raider's engine is any less capable than FF15, to be honest. FF15 does some great stuff with its magic effects and the like, but it pushes a lot of high-poly enemies at once, a lot of particle effects, fire & ice fx that rival FF15, and most impressively (I think) works Uncharted-style set-pieces is to a much more open game (something FF15 tries but fails to do, imo, with the Leviathan/Titan stuff being cool conceptually but lame looking). I'm not doubting that Luminous got FF15 to an impressive place, I just think Crystal's engine is more stable and capable. It's not to say Luminous is the worst SE engine either, since the Deux Ex engine is, I think, a disappointment after all they hyped it up to be. (Much like the game). Things you say there about semi-random nature of summons or ridiculous camera angles isn't exactly a big deal, though. It's a video game: you can put cameras anywhere. Any game faces these sorts of mechanical challenges (and things like the Warp Strike - Mass Effect's Vanguard Warp Charge is literally the same move as Noct's warp strike, broadly speaking, for instance - these things have existed.

I'll give Tomb Raider equal capabilities with regards to VFX, but am I missing something with regards to its number of enemies? I've looked at walkthroughs on YouTube of parts of the game that should have had a bunch of enemies (i.e. stuff labeled as "huge battle", the encounters immediately leading up to the final boss) and it seems like it generally limits its waves to a half-dozen guys, most of whom remain at shooter distances (and therefore lower LoD). Meanwhile, FFXV can throw a half-dozen Sabertusks and ten Magitek Assassins into a fight you'd started with a mix of eight or so Garulas/Garulessas/Garulets, all in melee range, with three or four super high quality ally AI models taking part, for a total of close to 30 character models and close to 10 unique animation sets being juggled simultaneously. And then an Astral can show up and nuke the entire area. That's not even the same order of magnitude.

It's also worth pointing out that most of FFXV's setpieces skip right past Uncharted and aim for God of War, which is probably why they ended up being a bit rougher around the edges. The train setpieces kept their scope in check better but also ended up feeling a lot more polished because of it (while still being pretty awesome). And, of course, there was one super-ambitious thing near the end of the game that was just plain jaw-dropping because it did pull off its ridiculous scope.

As for the summons and warp strikes, I was under the impression that, while one can obviously put a camera anywhere in a video game, it'd still create unique challenges for the engine's streaming solution if the camera's position can move massive distances almost instantaneously, with the semi-random factor making it mostly impossible to limit the scenarios in which it can happen to simplify it.

3) Witch 2.0 is fucking amazing, but let's not forget it was running on a Quad SLI Titan setup that'd literally cost nearly 10 grand to build. I'm more interested right now in what can be used to make games.

It was also running at 4K/60fps, wasn't it? The requirements would have been a lot closer to feasibility if it wasn't 4X the standard HD resolution and 2X the standard frame rate for slower-paced games. =P

4) I agree about the PS4 version of TW3, actually - there's a lighting issue with that game that on that platform really suffocates its art. But an engine is more than one platform, but it'll be interesting to revisit this when 15 comes to PC. RE KH -- I dunno what it is about KH2.8, parts of it do look cheap - but DQ11 looks absolutely gorgous on UE.

DQ11 works much more effectively within its limitations (in large part due to what appears to be a choice to use clay shaders for skin and cloth instead of plastic shaders), but it's still handily trounced by FFXV with regards to model complexity and the presentation of materials. (DQ11 is much cartoonier, of course, but that still doesn't account for the low-res cloth and leather textures seen in cutscenes.)

As for PC FFXV, I suspect that it could look utterly fantastic if Tabata is allowed to make the version he wants, because the FFXV/Luminous pair feel almost too ambitious for the platforms they released on. Even with regards to simple improvements, 4xMSAA would pretty much wipe out the hair complaints, higher levels of AF clearly help a lot given the way the game looks in Pro screenshots, more frequent use of the highest LoD textures would give a better overall impression, and a competent CPU allowing for Ultra PC-level draw distance could make a huge difference. (IIRC, console ports tend to use Medium settings overall with Low draw distance due to their weak CPUs, no?) If they're able to re-implement some of their more complicated real-time GI solutions on top of that...

Generally speaking, I'd say this: Luminous has been built to an impressive thing and FF15 is a very pretty game, but if it was such hot shit it would be being used for DQ, KH and FF7 - why would Square want to pay Epic 5% of all their revenue (before profit!) on FF7 remake, potentially their biggest game ever? Luminous just isn't really ready for the big leagues, but that's because it's still a frankenstein's monster that's difficult to use. If they're smart FF16 will probably use something else and they can keep Luminous studio doing research, or put a smaller game on Luminous so their biggest game can maximize a tried and tested set of tools. (For instance, perhaps let BD2's next game continue on Luminous, with FF16 made elsewhere on a more tested engine). Based on what (I think) I know about who is on FF16 I think this'll actually be the case, but that team might well be learning Luminous, IDK.

I don't believe this argument for a second.

IIRC, Julien Merceron said at one point that engines need to develop alongside games, but it's a terrible idea to try to make multiple games with them simultaneously while they're still being developed. And it's obvious why that would be the case -- Squenix tried that with Crystal Tools, and it backfired horribly.

FFVIIR, KH, and DQ11 were all in development for years before Luminous had its first full game release. KH was originally intended to use Luminous, but that team had no HD experience and the Luminous engine team was too busy making the engine to provide support -- I'm sure they wouldn't have made that decision if they had any other choice, considering the massive downgrade that came with the engine change. (The Visual Works Sora model in the Luminous trailer was so good compared to the current monstrosity. D= ) And both FFVIIR and DQ11 are outsourced, which would have made the use of an in-progress internal engine even more impractical. As you yourself said, EA limited Frostbite to DICE for years even though it was one of the best engines in the industry. Giving new engines to outsourcers isn't exactly a thing that happens very often.

Reasonably speaking, Squenix should follow EA's footsteps and make one big game with Luminous at a time until they can afford to have their engine team spend significant amounts of time providing support to other teams. Limiting it to small games would probably slow progress on the engine significantly. And the sooner Squenix's major properties abandon UE4, the better, because I can't stand the way the Squenix games with final/close to final footage look on that engine. =(

...though I suppose I could live with them operating on a longer timescale with regards to Luminous as a whole if they figure out how to take that "dump a Visual Works model into the game with only slight modifications" feature and use it outside of their own engine. I'm not a fan of the appearance of environments in Squenix's UE4 games, but I can live with them. The difference between Noct and Agni, on the one hand, and KHIII Sora and the DQ11 protagonist on the other physically pains me. >_<

 

[APZonerunner]

I don't fully have the time/inclination to reply to absolutely everything, because it's just going back and forth on semantics, but I just wanted to say:

It was also running at 4K/60fps, wasn't it? The requirements would have been a lot closer to feasibility if it wasn't 4X the standard HD resolution and 2X the standard frame rate for slower-paced games. =P

It's so far from being remotely possible on regular hardware at this point it's silly. 4K/60FPS has nothing to do with it really; that demo featured 64 million polygons per frame, about 15 times what the PS4 Pro can reliably render even at 30fps, and that's without getting into the 8K textures and insane shaders. Like it's impressive, but it's some seriously future tech shit.

I don't believe this argument for a second.

IIRC, Julien Merceron said at one point that engines need to develop alongside games, but it's a terrible idea to try to make multiple games with them simultaneously while they're still being developed. And it's obvious why that would be the case -- Squenix tried that with Crystal Tools, and it backfired horribly.

FFVIIR, KH, and DQ11 were all in development for years before Luminous had its first full game release. KH was originally intended to use Luminous, but that team had no HD experience and the Luminous engine team was too busy making the engine to provide support -- I'm sure they wouldn't have made that decision if they had any other choice, considering the massive downgrade that came with the engine change. (The Visual Works Sora model in the Luminous trailer was so good compared to the current monstrosity. D= ) And both FFVIIR and DQ11 are outsourced, which would have made the use of an in-progress internal engine even more impractical. As you yourself said, EA limited Frostbite to DICE for years even though it was one of the best engines in the industry. Giving new engines to outsourcers isn't exactly a thing that happens very often.

Sure, and I think this is actually the path they should take. Speaking to people who've worked with/on it, though, it's clear that Luminous is more of a mess now than Frostbite ever was when DICE shipped a game on it -- that's all I'm saying.

One thing I've pondered about a fair bit is why its top three architects had a particularly bloody, messy divorce & departure from SE around the time Tabata took over FF15 - unrelated to Merceron, because he (while an absolute genius) to be clear was CTO for the whole of Square Enix (coming off the back of a role as Eidos CTO), so he was overseeing all engine operations across all companies and unlike how at Konami he directly oversaw Fox Engine he was in more of a management role. Part of why he moved on was that he wanted to get more hands-on with something. Luminous was the brainchild of Yoshihisa Hashimoto (who also held the CTO role but only at SEJ, and produced Agni's Philosophy etc under Merceron, with Hashimoto having the more hands-on controlling role in Luminous' design as they tried to rescue it from the gaping maw of disaster that was Crystal Tools) - when Luminous hit some major snags in 2014 he fell on his sword, so to speak, and left the company. He took his two deputies with him, and that was - I think - the start of Luminous really having major difficulties. Merceron had been gone for a year at that point.

Hashimoto & his 3 deputies were head-hunted by SE from SEGA -- they'd designed Sonic Team's excellent Hedgehog Engine as used in Sonic Unleashed on, which for that generation was a really good engine that put performance first.

People clamouring for Agni in the future should be mindful that whole thing was Hashimoto's vision/creation and he's no longer there. I do think SE will make use of it, though, but it'll be without its creator, and it seemed he had strong ideas. As he said to me back in 2013:

"Having said that, in the beginning of creating a concept, we started to discuss and really consider "What is Final Fantasy"? The theme of the demo was Final Fantasy, we wanted to have a Final Fantasy demo - we wanted to share an idea of what it is first of all. We took some time doing that."
"When we discussed the concept of 'What is Final Fantasy', we tried to narrow down what has to be there to be Final Fantasy. We filtered some of the things that we felt not necessary to remain - we just excluded them, and kept only the core, essential things needed to be Final Fantasy. We also added some things that were not there before, and this was the end result."
​

At the time it was a very different vision - that first Agni demo was running on much more reasonable hardware to the point where in the one-hour one-on-one session with Hashimoto and his top three staff they yanked a sheet off a covered table and revealed the very machine that rendered the trailer, then set it running for us live and panned the camera around and stuff. I can't remember the exact spec - I do have it written down somewhere - but it was a lot more reasonable, it was maybe a $3000 rig rather than a rig with a $9000 of GPUs alone. Hashimoto also spoke at the time about how his #1 priority for any game in Luminous would be hitting 60fps, and talked about the engine being scaleable enough to run on iOS and stuff (like UE). After he left the focus seemed to change, and when Tabata took control of all that it seems like they shifted focus towards visual fidelity above all else; The Witch demo and the insane tech demo is the result. That's not bashing on Tabata either; I have a great deal of respect for him - just it appears to be what occurred.

 

[Ikkin]

It's so far from being remotely possible on regular hardware at this point it's silly. 4K/60FPS has nothing to do with it really; that demo featured 64 million polygons per frame, about 15 times what the PS4 Pro can reliably render even at 30fps, and that's without getting into the 8K textures and insane shaders. Like it's impressive, but it's some seriously future tech shit.

Resolution and frame rate scale linearly with regards to GPU, don't they? (CPU is a different beast, obviously, but CPUs can't be SLI'd so it doesn't really matter here.) Trading 4K for 1080p would get you down to one Titan X, and cutting back to 30fps would give you half a Titan X. Or you could checkerboard the 4k and cut back to 30fps on one Titan X. That wouldn't be reasonable for this generation, obviously, but why wouldn't it be a reasonable equivalent to the original Agni demo on hypothetical hardware from 2019/2020 when next gen comes around? Am I missing something?

Sure, and I think this is actually the path they should take. Speaking to people who've worked with/on it, though, it's clear that Luminous is more of a mess now than Frostbite ever was when DICE shipped a game on it -- that's all I'm saying.

Considering the scope of what FFXV required Luminous to do, that seems like that should have been expected. =/

One thing I've pondered about a fair bit is why its top three architects had a particularly bloody, messy divorce & departure from SE around the time Tabata took over FF15 - unrelated to Merceron, because he (while an absolute genius) to be clear was CTO for the whole of Square Enix (coming off the back of a role as Eidos CTO), so he was overseeing all engine operations across all companies and unlike how at Konami he directly oversaw Fox Engine he was in more of a management role. Part of why he moved on was that he wanted to get more hands-on with something. Luminous was the brainchild of Yoshihisa Hashimoto (who also held the CTO role but only at SEJ, and produced Agni's Philosophy etc under Merceron, with Hashimoto having the more hands-on controlling role in Luminous' design as they tried to rescue it from the gaping maw of disaster that was Crystal Tools) - when Luminous hit some major snags in 2014 he fell on his sword, so to speak, and left the company. He took his two deputies with him, and that was - I think - the start of Luminous really having major difficulties. Merceron had been gone for a year at that point.

Hashimoto & his 3 deputies were head-hunted by SE from SEGA -- they'd designed Sonic Team's excellent Hedgehog Engine as used in Sonic Unleashed on, which for that generation was a really good engine that put performance first.

People clamouring for Agni in the future should be mindful that whole thing was Hashimoto's vision/creation and he's no longer there. I do think SE will make use of it, though, but it'll be without its creator, and it seemed he had strong ideas. As he said to me back in 2013:

"Having said that, in the beginning of creating a concept, we started to discuss and really consider "What is Final Fantasy"? The theme of the demo was Final Fantasy, we wanted to have a Final Fantasy demo - we wanted to share an idea of what it is first of all. We took some time doing that."
"When we discussed the concept of 'What is Final Fantasy', we tried to narrow down what has to be there to be Final Fantasy. We filtered some of the things that we felt not necessary to remain - we just excluded them, and kept only the core, essential things needed to be Final Fantasy. We also added some things that were not there before, and this was the end result."
​

At the time it was a very different vision - that first Agni demo was running on much more reasonable hardware to the point where in the one-hour one-on-one session with Hashimoto and his top three staff they yanked a sheet off a covered table and revealed the very machine that rendered the trailer, then set it running for us live and panned the camera around and stuff. I can't remember the exact spec - I do have it written down somewhere - but it was a lot more reasonable, it was maybe a $3000 rig rather than a rig with a $9000 of GPUs alone. Hashimoto also spoke at the time about how his #1 priority for any game in Luminous would be hitting 60fps, and talked about the engine being scaleable enough to run on iOS and stuff (like UE). After he left the focus seemed to change, and when Tabata took control of all that it seems like they shifted focus towards visual fidelity above all else; The Witch demo and the insane tech demo is the result. That's not bashing on Tabata either; I have a great deal of respect for him - just it appears to be what occurred.

...where is this information coming from? o_o Is this all just stuff you have personal knowledge about but never wrote down officially, or...?

And if you do have personal knowledge of this stuff, how did it affect Episode Duscae, which came out in early 2015? Did they change the engine shortly before or during the development of that demo?

In any case, if Tabata was the one who decided to go for visual fidelity over 60fps, I tip my hat to him for that. =P There may come a point where the increase in visual fidelity from halving the frame rate is too minimal to be worth bothering with, but it probably won't be any time soon given consoles' propensity to go for low-end CPUs that require massive compromises on polygonal complexity and draw distance.

(Sure, maybe the current version is too ambitious, but I kind of like having over-ambitious projects around. Shadow of the Colossus didn't belong on a PS2, but I'm very glad that I was able to play it there. =P )

 

[APZonerunner]

Resolution and frame rate scale linearly with regards to GPU, don't they? (CPU is a different beast, obviously, but CPUs can't be SLI'd so it doesn't really matter here.) Trading 4K for 1080p would get you down to one Titan X, and cutting back to 30fps would give you half a Titan X. Or you could checkerboard the 4k and cut back to 30fps on one Titan X. That wouldn't be reasonable for this generation, obviously, but why wouldn't it be a reasonable equivalent to the original Agni demo on hypothetical hardware from 2019/2020 when next gen comes around? Am I missing something?

You wouldn't just have to pare back of resolution but also on the number of polygons, ultimately; but if you do that you lose fidelity. Like, in the presentation he gave alongside Witch 2.0 Tabata was saying that ultimately this tech is potentially as much as 10 years out from being consumer friendly. You can either look at this as mad prescient R&D or as a money sink, depending on your perspective. Part of why the Witch footage looks so good is that it's supersampled, of course, but by the time it'll actually be usable those resolutions will be more common, so that's interesting.

Considering the scope of what FFXV required Luminous to do, that seems like that should have been expected. =/

Based on conversations I've had (some of them on the record - look up my 3 interviews with Tabata over 15's development!) I think nobody expected it to be as difficult or as much of a struggle as it ultimately was. I think KH had to be UE due to the timing of its development, but I think FF7's decision to use UE4 (which was very wise IMO) was directly impacted by how the 15 team was basically pushing a very heavy rock uphill in terms of working on and improving the engine.

...where is this information coming from? o_o Is this all just stuff you have personal knowledge about but never wrote down officially, or...?

And if you do have personal knowledge of this stuff, how did it affect Episode Duscae, which came out in early 2015? Did they change the engine shortly before or during the development of that demo?

In any case, if Tabata was the one who decided to go for visual fidelity over 60fps, I tip my hat to him for that. =P There may come a point where the increase in visual fidelity from halving the frame rate is too minimal to be worth bothering with, but it probably won't be any time soon given consoles' propensity to go for low-end CPUs that require massive compromises on polygonal complexity and draw distance.

(Sure, maybe the current version is too ambitious, but I kind of like having over-ambitious projects around. Shadow of the Colossus didn't belong on a PS2, but I'm very glad that I was able to play it there. =P )

Which bit do you want? I mean, you can see the job positions, timing of leaving etc for Hashimoto, Merceron etc on LinkedIn and the like. When Hashimoto left it was a fairly big news story because Square had hyped him up & put him in front of press lots of the course of two years as a big new name for them -- then he quit for "personal reasons" and the three guys who joined the company with him from SEGA and headed up the earliest Luminous development/Agni also left the same week. I have personal knowledge of the ins and outs of their departure, but it doesn't take a seer to look at that and figure out there were deep professional differences that led them to leave SE less than 5 years after joining. (Part of this might also be down to SE's culture, though, which is notorious for overwork even by Japanese standards).

Luminous being Hashimoto's brainchild and not Merceron or anyone else's is readily available information. At E3 2012 Square Enix invited people into their booth after-hours after the show had closed on Tuesday. I was there! They showed a small group of people - about 40 press - Angi's Philosophy for the first ever time in the world. Hashimoto himself introduced it. If you prod around you'll find news stories on all this, some of them from me. The next two days at E3 they trotted out Hashimoto for a slew of press interviews. I sat with him for over an hour, and he ran the demo in real-time on a PC, even handing me a controller so I could fly the camera around the Agni environment in real-time - and like I say, it was running on much more modest hardware than later demos. They really wanted to prove it was running on 'realistic tech' for the next couple of years. Alt tabbed out to desktop, went into the PC's device manager, etc.

Obviously at this point Versus was actually still Versus (the XV decision would be made about two months later, and Tabata would first be drawn into the project shortly after that) and Hashimoto seemed to be hinting a lot that Agni was a vision for a new FF - in my interview, he asked if he'd mind if he interviewed me briefly at the end, and asked me what I thought of the demo and if I'd like it as a vision for FF's future. When I answered, they took notes! They seemed real about it. Stuff behind-the-scenes changed pretty rapidly after that though, it seems.

Tabata changing the direction to be about mega high-end tech demos is a supposition on my part, though it is one based in facts: if you go and read interviews with Hashimoto from the 2012 period from me or any of the many others he spoke to, you'll see that his focus is 100% on performance and on an ease of pipeline. It only seemed like after he left and Tabata took over as the head of the Luminous project that the direction seemed to significantly change. I think Tabata's direction has merit, though I also wonder if FF15 would've run into the sheer number of development problems they had if they'd taken a more restrained approach to the engine.

RE Duscae: Hashimoto left almost a full year before Duscae came out, and based on the timeline I have the plan for the demo probably existed at that point but no work had been done. The version of Luminous used in FF15 only really became concrete around 6-7 months before Duscae was released, I think, and that was basically shortly after Hashimoto & co moved on.

 

[xXShuyaXx]

Wasn't it already mentioned before that, specially with the FFVIIR, they wouldn't use the same engine as FFXV because the game will end up looking too similar? From the trailers that we have, it would appear that they are using a re-purposed UE4, possibly using some similar codes from the Luminous engine. Why I say this is because they implemented the exact same crawl under/scale wall codes from FFXV. You can see this in the gameplay video for FFVIIR.

FFVIIR started pre-production while FFXV was still in production. And it would make sense to avoid the Luminous Engine to avoid looking overly akin to each other, despite being different games. Most AAA developers upgrade their existing engines to be better for the next title. They don't use the exact same engine, it's a upgraded variant every time.

However in this case, KH3/FFVIIR/FFXV were being developed at the same time. If they all utilized the same engine, which will technically become near obsolete when the first title using it comes out, by the time the latter games are released, the engine will be outdated, looking overly similar to the first title. However, the argument can be made that they can upgrade the engine as they develop the game; however as a counter debate, that's exactly what FFXV did. And that took a lot of their resources and time, further delaying production.

That's where Luminous Studio Pro or whatever it actually is possibly comes into the picture. This will the be the next big upgrade to the current engine and possibly be used for the next fresh title. I assume once FFXVI starts development, the Luminous Studio Pro or the incrementally upgraded version like Vers1.1 will used as the base engine. Using similar codes from the current engine, but ultimately shining by it's own powers.

So why UE4? Because UE4 is an open source developer friendly engine that can be re-purposed for many projects. It does provide more options, however it does have it's own limits. But it is also fast, efficiently reliable and cheap.

 

[Hey Everyone]

I doubt they're just going to abandon it after putting so much recources into developing it.

Who knows maybe they will make a Kingslavie game, and Versus XV after all
heh heh

 

[Guitar (pseudo)God]

You wouldn't just have to pare back of resolution but also on the number of polygons, ultimately; but if you do that you lose fidelity. Like, in the presentation he gave alongside Witch 2.0 Tabata was saying that ultimately this tech is potentially as much as 10 years out from being consumer friendly. You can either look at this as mad prescient R&D or as a money sink, depending on your perspective. Part of why the Witch footage looks so good is that it's supersampled, of course, but by the time it'll actually be usable those resolutions will be more common, so that's interesting.



Based on conversations I've had (some of them on the record - look up my 3 interviews with Tabata over 15's development!) I think nobody expected it to be as difficult or as much of a struggle as it ultimately was. I think KH had to be UE due to the timing of its development, but I think FF7's decision to use UE4 (which was very wise IMO) was directly impacted by how the 15 team was basically pushing a very heavy rock uphill in terms of working on and improving the engine.



Which bit do you want? I mean, you can see the job positions, timing of leaving etc for Hashimoto, Merceron etc on LinkedIn and the like. When Hashimoto left it was a fairly big news story because Square had hyped him up & put him in front of press lots of the course of two years as a big new name for them -- then he quit for "personal reasons" and the three guys who joined the company with him from SEGA and headed up the earliest Luminous development/Agni also left the same week. I have personal knowledge of the ins and outs of their departure, but it doesn't take a seer to look at that and figure out there were deep professional differences that led them to leave SE less than 5 years after joining. (Part of this might also be down to SE's culture, though, which is notorious for overwork even by Japanese standards).

Luminous being Hashimoto's brainchild and not Merceron or anyone else's is readily available information. At E3 2012 Square Enix invited people into their booth after-hours after the show had closed on Tuesday. I was there! They showed a small group of people - about 40 press - Angi's Philosophy for the first ever time in the world. Hashimoto himself introduced it. If you prod around you'll find news stories on all this, some of them from me. The next two days at E3 they trotted out Hashimoto for a slew of press interviews. I sat with him for over an hour, and he ran the demo in real-time on a PC, even handing me a controller so I could fly the camera around the Agni environment in real-time - and like I say, it was running on much more modest hardware than later demos. They really wanted to prove it was running on 'realistic tech' for the next couple of years. Alt tabbed out to desktop, went into the PC's device manager, etc.

Obviously at this point Versus was actually still Versus (the XV decision would be made about two months later, and Tabata would first be drawn into the project shortly after that) and Hashimoto seemed to be hinting a lot that Agni was a vision for a new FF - in my interview, he asked if he'd mind if he interviewed me briefly at the end, and asked me what I thought of the demo and if I'd like it as a vision for FF's future. When I answered, they took notes! They seemed real about it. Stuff behind-the-scenes changed pretty rapidly after that though, it seems.

Tabata changing the direction to be about mega high-end tech demos is a supposition on my part, though it is one based in facts: if you go and read interviews with Hashimoto from the 2012 period from me or any of the many others he spoke to, you'll see that his focus is 100% on performance and on an ease of pipeline. It only seemed like after he left and Tabata took over as the head of the Luminous project that the direction seemed to significantly change. I think Tabata's direction has merit, though I also wonder if FF15 would've run into the sheer number of development problems they had if they'd taken a more restrained approach to the engine.

RE Duscae: Hashimoto left almost a full year before Duscae came out, and based on the timeline I have the plan for the demo probably existed at that point but no work had been done. The version of Luminous used in FF15 only really became concrete around 6-7 months before Duscae was released, I think, and that was basically shortly after Hashimoto & co moved on.

Fantastic stuff -- thanks!

The fact that FFXV was as good as it was is even more of a miracle than I thought.

I'm sure there's enough out there, and I take Tabata's word, but what were the circumstances regarding the director switchover and departure from the Versus specifics? It seems to me like the top brass told the team that:

1. You're getting this out in 2016, come hell or high water on this engine we've dropped in your lap.
2. Get rid of the FNC crap -- we've had 4 games of it already.
3. Nomura, this project is off the rails. Tabata, you fix it while we have our top creative guy work on the way-overdue KH3 and this remake Hashimoto wants to do.

I'm very interested to see how Luminous and the BD2 development methods pan out going forward. Truth be told, that's what endears me to Tabata (and anybody behind-the-scenes). They dragged a business unit kicking and screaming into an Agile/flat-hierarchal development team.

 

[Hey Everyone]

UE4 is a really strong engine and can work in a variety of different styles (for instance, Street Fighter 5 and Tekken 7 both use UE4 but couldn't look more different).

SE has historically struggled with the engine/tech race, and I think when 2.8 comes out you'll see how much polish they can achieve when they're in a pre-defined engine with proper tools. Also, CC2 is doing the heavy lifting on 7 development wise, and they're a studio who have a history both with Unreal and third party engines in general.

If they want to continue to chase their own engine they need to have one studio develop games in that engine and iterate on it. EA did this with DICE and Frostbite, and it took three generations of the engine and over ten years before the engine and tools were mature enough to be deployed right across all of EA. SE is a long, long way from that - the struggles with FF15 prove that. UE4 is about as mature as engines get, so... it's a no brainer.

I think the main reason they aren't using Luminous for FFVII Remake, and KH3 is because they don't want to go through a hellish development cycle again, because they probably know that if they tried Luminous they couldn't be working on FFVII or KH3 at the same time.

 

[Ikkin]

You wouldn't just have to pare back of resolution but also on the number of polygons, ultimately; but if you do that you lose fidelity. Like, in the presentation he gave alongside Witch 2.0 Tabata was saying that ultimately this tech is potentially as much as 10 years out from being consumer friendly. You can either look at this as mad prescient R&D or as a money sink, depending on your perspective. Part of why the Witch footage looks so good is that it's supersampled, of course, but by the time it'll actually be usable those resolutions will be more common, so that's interesting.

Polygon count is more dependent on CPU and memory constraints than GPU, isn't it? SLI doesn't really provide benefits in either of those regards, IIRC. And to the extent that it does depend on GPU, linear scaling down from 4K/60 to 1080p/30 or checkerboard-4K/30 should still apply.

With that said, Witch 2.0 came across as being even less optimized than Agni's Philosophy with regards to raw poly count, and that demo included individually-modeled toes inside Agni's shoes for no reason other than that no one bothered to cull them. The hair solution in particular seems like an obvious place to cut back without a huge impact on the overall effect.

And Squenix has always had a "mad prescient R&D" team called Visual Works. The current Luminous stuff seems like an attempt to bridge the gap and make some of that expertise usable in real-time.

Based on conversations I've had (some of them on the record - look up my 3 interviews with Tabata over 15's development!) I think nobody expected it to be as difficult or as much of a struggle as it ultimately was. I think KH had to be UE due to the timing of its development, but I think FF7's decision to use UE4 (which was very wise IMO) was directly impacted by how the 15 team was basically pushing a very heavy rock uphill in terms of working on and improving the engine.

Fair enough. It's not exactly uncommon for projects to end up being way more of a struggle than they were initially expected to be. XD;

It's been said that FFVIIR was already underway when Squenix trolled with the PS4 remaster in late 2014, though, so I'd think there would be timing issues there, too.

Which bit do you want? I mean, you can see the job positions, timing of leaving etc for Hashimoto, Merceron etc on LinkedIn and the like. When Hashimoto left it was a fairly big news story because Square had hyped him up & put him in front of press lots of the course of two years as a big new name for them -- then he quit for "personal reasons" and the three guys who joined the company with him from SEGA and headed up the earliest Luminous development/Agni also left the same week. I have personal knowledge of the ins and outs of their departure, but it doesn't take a seer to look at that and figure out there were deep professional differences that led them to leave SE less than 5 years after joining. (Part of this might also be down to SE's culture, though, which is notorious for overwork even by Japanese standards).

I was talking about the whens and whys of Hashimoto and Merceron leaving, so that answers that question.

Luminous being Hashimoto's brainchild and not Merceron or anyone else's is readily available information. At E3 2012 Square Enix invited people into their booth after-hours after the show had closed on Tuesday. I was there! They showed a small group of people - about 40 press - Angi's Philosophy for the first ever time in the world. Hashimoto himself introduced it. If you prod around you'll find news stories on all this, some of them from me. The next two days at E3 they trotted out Hashimoto for a slew of press interviews. I sat with him for over an hour, and he ran the demo in real-time on a PC, even handing me a controller so I could fly the camera around the Agni environment in real-time - and like I say, it was running on much more modest hardware than later demos. They really wanted to prove it was running on 'realistic tech' for the next couple of years. Alt tabbed out to desktop, went into the PC's device manager, etc.

I guess I just wasn't paying enough attention to the major technical figures when reading up on Agni stuff. ^^; I definitely remember it running in real-time on PC hardware that wasn't outrageously different than console specs, though. Wasn't there a PS4 port of the demo shown at a Sony event, too?

Obviously at this point Versus was actually still Versus (the XV decision would be made about two months later, and Tabata would first be drawn into the project shortly after that) and Hashimoto seemed to be hinting a lot that Agni was a vision for a new FF - in my interview, he asked if he'd mind if he interviewed me briefly at the end, and asked me what I thought of the demo and if I'd like it as a vision for FF's future. When I answered, they took notes! They seemed real about it. Stuff behind-the-scenes changed pretty rapidly after that though, it seems.

Now, this is interesting. Do you think that Agni was in the running for FFXV at that point? And if it was, could the departure of Hashimoto and his crew have been related to Squenix sidelining the Agni project?

Tabata changing the direction to be about mega high-end tech demos is a supposition on my part, though it is one based in facts: if you go and read interviews with Hashimoto from the 2012 period from me or any of the many others he spoke to, you'll see that his focus is 100% on performance and on an ease of pipeline. It only seemed like after he left and Tabata took over as the head of the Luminous project that the direction seemed to significantly change. I think Tabata's direction has merit, though I also wonder if FF15 would've run into the sheer number of development problems they had if they'd taken a more restrained approach to the engine.

Ambition is a double-edged sword, certainly. But it's always hard to deal in "what-if"s, especially since a less ambitious engine wouldn't necessarily mean a less ambitious project in other ways, just a less visually-impressive one, and I suspect that FFXV benefits heavily from its top-tier combat animation and open-world lighting.

RE Duscae: Hashimoto left almost a full year before Duscae came out, and based on the timeline I have the plan for the demo probably existed at that point but no work had been done. The version of Luminous used in FF15 only really became concrete around 6-7 months before Duscae was released, I think, and that was basically shortly after Hashimoto & co moved on.

Interesting... so they wouldn't have had much time to work on Duscae after Luminous had solidified. No wonder it looks so different from the final game!