Possible explanation for missing character designs?

[Kiske]

I have been thinking a lot about why Suzuki and co. elected to show a photoshopped Ryo. The last update gave us a glimpse into how they have managed to regain some of that shenmue "feel" back from the sterile unreal engine pictures of the past. Shenmue has a color and texture that is all its own.

Part of what gave rise to that look were the limitations of working on what is now considered weak hardware (Dreamcast) back in the day. One of the ways that SEGA used to make shenmue look so far ahead of its time is the texture work that was dedicated to creating the faces of NPCs throughout. A clever design allowed them to bake the lighting and shadows into the textures. This gave otherwise flat polygonal skeletons an incredible amount of depth.







Those faces were not photo-realistic. They were, in a sense, painted by hand. It was crafted. Each shadow was placed with purpose, and not the result of some modern lighting engine. I have been wondering recently if this is what Yu and his team are struggling to recreate. It would certainly clash drastically with the rest of the game. Especially given the emphasis on light. This can be seen in that morning market battle scene.



Ryo's face doesn't look off only because of the photoshop, but because the lighting has been painted on. Notice how we also see the backs of the heads of two other NPCs. The one to the left has a definite shape to it. It is highly polygonal compared to the older games. With such harsh light, artificial shadows would make no sense on such a three dimensional shape.

If I were to imagine how that NPC looks like from the front, this image of jade empire (xbox) is much more reasonable. Something where the lighting is guided more by the polygons than by an expectation of what the texture should represent.



Modern games also follow this approach because it works well with a good lighting engine, something shenmue never really had.



Think about it. Why does this face look off?



Bake in some more of the lighting into the texture (as well as some obvious changes) and suddenly it looks a lot better. Not perfect, but better.



Again, another example of how baked in shadowing is giving way to modern lighting. Looks good, but still a bit off.



It's hard enough to get this working on Ryo. It will be a huge undertaking to get it looking right on a great variety of NPCs. One of the magical aspects of shenmue has always been how each NPC is immediately recognizable, slightly cartoonish, and, as stated before, hand crafted. This is very hard to recreate. It requires a lot of people working very hard. This may be the reason why they are still trying to get things looking right before they show us.




Just my opinion in trying to make sense out of a few things. Fell free to disagree and tell me why I am completely wrong haha. In the end, I just want the game to look as good as possible!

[khmlight3]

I agree with you great post! This game will look better when it comes out!

[Hazuki00]

Great post! It amazes me how a "few changes" makes Ryo look a lot better!

[killthesagabeforeitkillsu]

Think about it. Why does this face look off?



Bake in some more of the lighting into the texture (as well as some obvious changes) and suddenly it looks a lot better. Not perfect, but better.

While I agree that lighting is important, what makes the difference here are the "obvious changes". As you said most of these textures were handcrafted and there is a certain artistic touch that is hard to replicate.

So there is a need to use the old model structure and build upon them. At least from the characters we already know.

[ChiefNeo]

Think about it. Why does this face look off?



Bake in some more of the lighting into the texture (as well as some obvious changes) and suddenly it looks a lot better. Not perfect, but better.

While I agree that lighting is important, what makes the difference here are the "obvious changes". As you said most of these textures were handcrafted and there is a certain artistic touch that is hard to replicate.

So there is a need to use the old model structure and build upon them. At least from the characters we already know.

For sure. It always amazes me how the original shenmue NPCs were very, very simple in terms of geometry. Much like world ofwarcraft, art direction carried that game in a lot of ways.

[shredingskin]

We can still manually paint "dark spots" in the ambient occlusion map and it's standard procedure for most games.

The game will look different because everything is different xD

[Anonymous81]

I think this is essentially true in a sense. I'm no developer, but I suspect it's less that they're struggling with it, and more that they've been testing ways to get as close to it as possible during preproduction as they can before proceeding. They know exactly what they want, and want to find the best and most efficient means of achieving that so that it's out of the way once they start creating content and they already know what they can and can't do in terms of avoiding straying from the precise visual effect they're aiming for.

I suspect that's what is meant by exiting preproduction and moving on to production. They keep mentioning nearly being done "testing" things, despite there not yet being any real feature content to fully test. I think these visual elements are part of (among many other elements) what they mean by that.

As shredingskin says above, it's not as if they can't still simply (and don't commonly already) do these sorts of things by hand, or at least in part by hand. It's more that there's also all of these other lighting and shadowing (both source and global) methods available to them that they have to test their assets with to achieve the aesthetic design they envision. While the tools are more convenient and capable, and can be customized as much or as little as they need them to be, they also aren't as intimately familiar with them as an engine they built themselves from scratch like they had in the first two games. That's a double edged sword when trying to create something contemporary yet true to something almost 20 years old.

Basically, I suspect they simply haven't shown us what they'll end up creating now that they've settled upon, or have nearly settled upon, the technique they plan to use for everything, because it just doesn't exist yet as they've only now reached a decision on how EXACTLY to proceed. Clearly their goal is to get as close to the photoshopped face as possible, and that's very particular and I can see it taking a lot of, "Maybe if we try this... okay that's looking better, but let's see what happens when we apply X... hmm okay that's further away. How about Y?" Etc.

I have a feeling that's the phase they're now exiting.