Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Categories
YL2 Update December
odes
Administrator
Originally posted Dec 30, 2017 at 2:29pm.
Hi all!
Hope you're all having a great time during these holidays.
We're back with another news update.
So... December... I'll be honest with you - the character creator has been a struggle for us. We have so many things we want to realize with it, but organizing all those ideas into something workable has proven challenging. With the adaptive rig tech that we created and showed off previously, we're trying to raise the bar of what is currently possible to create in a character editor. I really believe we have something truly unique with this tech, but the question remains how we should organize our app around it and how to create content for it in such a way that we can maximize the potential of it to its fullest extent.
Our original idea with the character creator was to offer a large selection of sliders, affecting the character model being authored in different ways. This was the direction we've been heading in for some time now. However, upon implementation and content authoring for this, we started to realize there are a lot of problems with this approach. To be more precise, there is great complexity emerging from the usage of a lot of sliders that affect the same areas in the model. Using several of these sliders at the same time means we have to create "corrective shapes" (and maps) for our mesh, so combinations of these sliders won't cause the model to look weird when applied together. That means for each additional slider we add, we have to create a whole array of corrective shapes to cover the possible combinations that can be created using these sliders. Adding just one extra slider to an area in this system would vastly increase the number of possible combinations, and the number of required corrective shapes.
We realized that such a system is simply not feasible in the long run. We would have to limit the amount of sliders to reduce the amount of possible combinations for it to be manageable, but we're not that keen on developing ourselves into a corner.
In addition to the problem explained above, we also realized something else. When we first started developing this system with sliders, we intended to have only 1 slider for making the character chubby or muscular. Moving the slider left would make the character chubbier (in several stages, including hyper), and moving it right would make the character more muscular (again in several stages). But here's when we realized that if we were to commit to this system, these stages would be the only ones we could do. There would be no more variety to chubby and muscular shapes than the ones offered by this slider and its stages. Granted, there would be other sliders to affect the mesh in other ways of course, like increasing the size of tits and ass for example, but the actual shape for "chubby" and "hyper chubby" would only be just those two. Surely, a chubby character or a hyper chubby one for that matter, could look in so many different ways. Bellies would accumulate fat differently, along with butt and arms for example. If we only had that one slider, there would be limited variety. We didn't like this, since it wouldn't harness the true potential of the adaptive rig tech. So we started thinking in a different direction...
Just in case you don't buy these concerns, please understand that with our character creator, we're aiming to do something very different from what you've seen in others. Of all the character creators I've seen (AAA games included), nothing comes even close the size and shape differences we want to offer in ours. There are most definitely character creator systems out there that offer a great deal of options, probably more than we are able to implement being just the small team that we are, and yet they are not able to affect the character so immensely in sheer shape as we intend to. Characters in our app won't just be able to be small or big, muscular or chubby - they can be BIG or ˢᵐᵃˡˡ, MUSCULAR or CHUBBY (and everything in-between), if you catch our drift. Naturally this comes with its own set of challenges and problems.
Ok, so what's the plan then?
Well, we still intend to offer many sliders. Things like masculinity-femininity, height, scale (not same as height), tit size (including flat-chested), belly size, ass size, hip width and waist width, in addition to being able to select species, genitals, ears, feet, along with facial and genital customization. But when it comes to the actual "body type" of the character, instead of having these as separate sliders that cause a nightmare of complex combinations and corrective shapes, we came up with the idea to offer a selection of pre-authored body types to choose from, that you can blend in to your liking.*
* P.S. Note:
We have begun implementing a system that will allow you to blend in body types using a mask (in addition to a blend parameter). So if you for example like the belly area of a certain body type, you can blend in only that part while ignoring the rest by using a mask.
Now, before you completely dismiss this idea, just take a look at the variety of bodies offered by the furry fandom. There is such an immense diversity of creatures and body types, that encompassing even a fraction of them would not be possible using just sliders. If we intend to make this community any justice (and we do), we have to explore new options. With our system, we are truly trying to make something that hasn't been done before.
The benefits of this idea is severalfold. First of all, we bypass the complexity of combinations completely since each array of corrective shapes would be compartmentalized to each body type. That means if we ever decide to add more sliders to affect the body somehow, then we would only have to add 1 additional corrective shape per body type. Second of all (and I suppose this is connected to the first one), we could keep on adding more and more body types as time goes on. We wouldn't be limited to just a single chubby-muscular slider, but rather we could (in time) offer several different interpretations of "chubby" and "muscular" (and other body types). Thirdly, it makes authoring body types so much easier for us, since each body type would be its own shape that we sculpt in Zbrush, without having to worry about how it blends with other body types.
Again, with such a system you would still be able to tweak height, scale, sizes, femininity-masculinity among other things, but the actual body type would be pre-authored (although you'd able to tweak how much it influences your character). This body type selection would be limited to the body. Faces and other areas would have their own customization options.
We think this system is much more realistic for us to implement, while also offering far more variety in the long run.
First glimpse character creator
While development regarding shapes has been a back and forth journey, we have still managed to make some solid progress laying out the foundations for the character creator. We now have a working system in place for genital and species selection. The methods we use for this system have been developed with scalability in mind, so we can easily add more content to it as time goes on.
Below are recordings from the character creator. (Obviously extremely early footage.)
https://gfycat.com/InferiorRedHogget
When a selection is made, the system fetches the required models and merges them together. This goes for both their skeletons and mesh data (including all blend shapes authored in each individual mesh), forming a single object hierarchy and model. This is good for several reasons. Firstly, the draw calls are reduced when welding the meshes together, but also any visible seams present in the formerly separate objects will disappear. Here's an example showing the difference welding makes.
(Welding objects together to decrease draw calls and remove seams, in this case in neck area.)
In YL2, we have tried to keep performance in mind throughout the development from the very start, so if we're ever faced with the choice of either getting results fast but with no regards to performance, or getting results slightly slower but performant, we usually go with the latter option.
Obviously there are still a lot of things missing from this system. Things like selection for eyes, ears and tails (in addition to shapes) have yet to been implemented. These are the things that we intend to work on next.
Summary
This month we've been laying out the foundations for our character creator. You're now able to select the character's species and gender. After going back and fourth between different ideas for how the user is supposed to configure the model, we have come to the conclusion that our adaptive rig implementation will benefit the most from a body type selection system.
In the coming months, we will integrate body types and detail options into the character creator.
I know this post turned out to be more of a text type of post, but sometimes that's just the way it is. Hopefully we will have more flashy things to show next time.
- odes
Comments
Nothing has changed.
Don't know yet. Things are shaping up nicely though and we have made a ton of progress this month.
@pocket84
I don't understand your question.