Unity Acquires Ziva Dynamics

Is Emma's voice done via speech synthesis?

 

This video suggests it is an actress' performance: https://forum.unity.com/threads/unity-acquires-ziva-dynamics.1229805/#post-7839750

On this level of fidelity I think it is unavoidable to actually perform the animation and the audio together. Although I may be wrong and there is a tool to resync the audio and the animation performance somehow.

 

It would make a difference of Unity acquires Epic.

 

Somehow I doubt it.

"We've decided to keep Unreal engine a separate product".

 

Yes. And unlike small outfits, who should usually only care about what is available in their tools now, a large organisation knows they'll be investing significantly in terms of training, processes, custom code, infrastructure, etc. for the tools they adopt. So for them future direction is pretty important, even if they know details will change.

 

On the other hand, large organizations are less pressured to choose the turn-key products. They usually have the expertise, manpower and know-how to work on custom pipelines (asset, graphics, anything).

 

Best quality digital character I've seen yet, looking forward to trying this tool out. I sure hope Unity do the work of integrating it properly into the engine.

 

I can only imagine creating a rig from the solved data

 

By "large" I really meant the people between "indie" outfits and those with the firepower to do their own stuff.

But, off the shelf engines are used in big budget games. Though without income from royalties it's reasonable if Unity isn't aiming for that.

 

Honestly I'd have far fewer issues with all these acquisitions if Unity just made it so they all had non-release evaluation versions that throw up a watermark or don't support export so they could be reasonably evaluated. Like a full evaluation with no expiration, it's just that it's not suitable or license compatible with released work.

I say "non-release" instead of "non-commercial" because, well, free games exist so that's kinda a bit of a loophole issue.

 

I meant it in this way: bigger organizations won't complain about SRPs, they likely have Tech Artist(s), who knows how a graphics pipeline works, so they can make whatever their vision happen. Meanwhile a small indie studio usually doesn't have the manpower to easily realize the "anything they dream up". And this is true to many parts of an engine.

So, they go through the evaluation, and the direction and probable features are important, but they may be more lenient in some areas, because they can supply some of those features themselves.

 

It's not as simple as indies not having the time, it's also not having the money. That SRP issue that you encounter is often a matter for enterprise support or an expensive ticket. This is the reality of any engine, Unity or Unreal, you will end up staying in the safe area as indie or being comfortably budgeted to pay the professionals to do it.

 

Unless its free or seriously cheap with a perpetual one time buy license, I will be sticking with Character creator which so far is able to match metahuman, whilst also being able to handle stylized characters and now with CC4, non-humanoid characters too.

CC is a breeze to use and does everything I need at a decent pricepoint, so if unity is expecting this to catch on with anyone that is not a large sized company, they need to make sure it matches CC and either exceeds it in affordability, or feature sets (that are relevant to average users and companies, not what was shown on the blog)

 

This is just machine learned animation for muscles and deformation. It's character related but not for generating characters like your use case.

 

Right then I literally didnt understand what the blog post was even about. I would love it if there was a little less marketing speak in the blog posts than of late, as it makes it hard to tell what is being offered half the time, and what the intentions are... (And admittedly makes me more inclined to skip through it / skim it)

Thanks for pointing out and correcting me, I will go back into my dark corner now

 

I wouldn't say that (skin fold for ex), but it's pretty good on it's own, more versatile and definitely more complete and mature, like use a photo ad it does 80% of the modeling for you.

 

Agreed, although with the right inputs if using the photo method, or the right time investment if doing manually you can match metahuman entirely, but its not like its a few sliders - it takes learning the program and really needs to be used by an experienced artist to get that level of quality (which luckily we leave this work to our artist to handle otherwise I wouldnt attempt it )

But you dont end up with characters that all look the same! Which I cannot say about metahuman, they all have a single "style" and thats fine for some use cases, not so much in others


EDIT:

@hippocoder the blog posts first bold line literally says "and realtime character creation" so you can see why I was confused - I still dont know what this is now based on all this

 

Not really, if you can match the geometry in some way, the animation of metahuman is still superior, which is why I mentioned the skin fold, metahuman have really impressive deformation on its blendshape, and I haven't seen CC match that (yet?). BUT CC is compatible with metahuman so who cares, he win by making the enemy a friend!
https://www.reallusion.com/iclone/live-link/unreal-engine/metahuman/default.html

Side by side comparison in their tutorial does show you how so much more stylized figure in CC are, although a random quidam wouldn't see the difference, it's all realism to them.

 

Are you sure about that? First there's what @MadeFromPolygons pointed out, and I'm pretty sure the talking head said it was "created", not just "animated". But also, Ziva's website talks about using one existing character to create another.

I'd just grab the trial and find out, but I don't have (or adequately know how to use) Maya, which it plugs into.

 

I went through Ziva's public website (did not subscribe or asked for beta use, so they may have extra things in the oven, behind whatever they offer publicly). The current publicly available workflow is:
- you make your character or use what they are selling and base yours on theirs ($8,800 per pair, one female, one male)
- you upload your mesh
- they rig it, apply the muscle maps and all "magic" they need to make your mesh deforming on the right ways
- you perform the animation and they apply it to your character
- download the end product and put it into Maya and eventually into your game engine, presumably

As far as I understand currently they do not have character creation service (where you choose hair color, style, skin color, etc to build a unique human character).
But, there is the ZivaRT, which is in closed beta for now, where they mention that they will have body, face and clothes services, which I guess a step in this direction.

Plus now, that Weta and Ziva are under the same umbrella, they may unify their techs, so they can achieve more faster together.

 

And all I need to figure out now is getting my custom HMC to work with this.

 

From my understanding of setting up your head mesh for training in ZivaRT, you will mainly need Wrap3 to wrap your custom mesh to the Ziva source model. And then subdivide it for cleaner solving.

A demonstration of this process would have been helpful and show how it works in Unity

 

The blog does not do a good job of explaining new features, but nor does anywhere in Unity. I am a long-time user of Unity, but still do not know what many of the new features do, DOTS for example. And the reworking of the Blog to make it hard to find content means I rarely bother using it anymore. It used to be great.

 

Agreed I used to read daily, I read maybe once every couple of months now as the rework made the blog awful in terms of finding content.

 

It's kinda disappointing too, because I'm a visual learner and I rather have someone show me what they are talking about vs a blog laden with marketing speak...

I'm familiar with machine learning and 4D scanning, but most indies may not be up to speed on it. I had to look at the actual Ziva website to get ANY type of insight to what it is

 

The blog has been little more than a marketing arm for investors for a long while now. What I used to check regularly because there was actually useful content I now check maybe every other week.

 

Cool acquisition, and congrats to Ziva (I see you are also in Vancouver!)

BUT is it just my monitor or does the demo video need some basic colour correction? To me, the character looks under-lit and colours a bit over-saturated. This makes the darkness in the mouth and around teeth look bad, freckles and moles exagerrated, and the character doesn't really look like it's "in" the bright white room.

Here's a quick photoshop tweak to show what I mean: (Unless, of course, it really is my monitor calibration?!)



I may have gone a bit too pale, but this was just a quick edit.

 

Both my screens are widely uncalibrated, one oversaturated and the other undersaturated, and your correction still looks many magnitude better and natural in both of them!

So basically same services as Eisko/polywink?
https://www.eisko.com/

back to Ziva:
Generally stuff like that signal a preferred claimed software, but I wonder if the result still operable in others (if they use a default file format).

Also it seems their system don't track the eyes, it's dead in their reproduction (and the iris is too big).

 

Hello, I would like to know when this asset is going to come out?

 

https://store.zivadynamics.com/

 

Where can I learn more about this?
I cannot find any footage/current users/information what it is about.

Why is the more information button password locked?
For what reason a company offering services would keep their product information hidden?

 

May as well roll your own solution

 

If it saves several artists a month or two of work, it pays for itself. But it's definitely targeted at AAA space with that price of entry.

 

Yep. There is NO WAY we can use that at all

 

$8800?
With a price like that, it better be a great product to use.

 

And that's PER CHARACTER!!

 

Are you positive about that? It says "unlimited anatomical, bipedal characters" after all.

 

That's not that much of an unusual price for industrial software.
My employer in the automobile industry literally pays millions anually on various software tool licences.

 

I know how the whole Unity vs UE thing is getting old, but... I find it interesting how UE started in the enterprise AAA industry and has slowly but surely approached indies by making their stuff more accessible, both in terms of usability as well as pricing. Unity, however, started in the indie scene, but now more and more of their products and services are targeted towards enterprise.

 

It is interesting indeed. One should keep in mind however that without the major revenue games like Fortnite, Epic would likely not be willing to afford this aproach..

 

It's also not much compared to hiring a person to make a character. At the high end of the industry you could easily pay a person far more than that just for their time to make one character.

So if you can buy a bit of software that speeds them up significantly, gives a major boost to quality, or both, then that price tag suddenly looks pretty good.

 

As far as I know they've always had modding tools available, which means that in any studio it's common to have people around who are familiar with UE.

They've also been supportive of indies / startups who approach them since before Unity was well established. About 15 years ago at a trade show someone from Epic offered my small team a license deal without the then-massive up-front price tag (I think it was royalties instead), and I got the impression it was a standard offer.

I think it's more that both sides make noise about it when Epic works with another big name, moreso than their support for smaller teams being anything new.

 

I'll need to look at it again just to be sure. But that's very steep if you're a single developer

 

If 50 a month means I can stop and start the sub and it's for authoring then sure. I'd sub one month, get some work done. Pause. Sub again later.

But if we're talking lock-ins then I think these prices are missing a tier since that looks like AA indie team pricing, bit out of reach of small guys.

Cos for small indies, you look at what the bright memory chap did. You use facial capture from a phone if you have to. That's budget and looks decent still.

 

So if Ziva is hidden behind a massive paywall, I guess it's safe to assume the Weta tools will be as well (perhaps at twice the price, who knows). Basically, we don't get to play with those tools, ever, despite Unity claiming they're trying to get those tools out to the masses. A handful of wealthy studios aren't what the masses are though

 

Well, wait. It's not Unity that has set up this paywall. That's how Ziva was operating previously already, right?
I do believe Unity has more plans than just keeping status quo of the company they bought.

 

You've been here since 2015 and you really believe this?

 

Well I believe they will use it in future cutscenes for sure because as someone who can lipread... well the Enemies demo is the nicest rendering but the lipreading is bad, frankly.

But the prior Ziva demo, I could actually lipread her. Big difference. So for that alone, it increases accessibility (I cannot sign). So Ziva has a lot more potential than the odd Adobe remonetisation strategy beyond 'just graphics'.

 

None of this really addresses the core of what I said. Unity acquisitions either remain entirely status quo for how they functioned as a company or gradually vanish into the ether. I don't think Unity will change anything about how Ziva Dynamics functions or their pricing structure. Historically, that just hasn't been how things play out with their corporate acquisitions. The only stuff where we see changes to that sort of thing in any meaningful capacity is when they acquire assets like TextMeshPro.

 

Not much, but they say, they're working on it.