Facegen

[New Horizon]

If oDD says we're safe using it, then that's good enough for me.

[Ishtvan]

If oDD says we're safe using it, then that's good enough for me.

 

I dunno, using illegit software is one thing, but when you consider distributing something that a living person or corporation has an active copyright on, that's taking a much more serious risk. So far all of our departments have taken great pains to avoid that risk. That's why we're recording our own SFX instead of using an SFX library, and why Spring is making sure not to use paintings that still have active copyrights. Why should copyrights owned by Facegen be an exception?

[sparhawk]

You mean the same Odd, who thinks that it would be to much hassle for companies to go after softwarepirates, because they are doing it for free and it would not be worth to pay the lawyer for this?

 

If the models are bit tweaked then it might be ok, otherwise it can become pretty risky.

[Renzatic]

As long as Odd tweaks the resulting models enough so that they're not immidiately discernable from a regular Facegen generated head, I wouldn't have a huge problem with it.

 

I'd recommend he redo the textures from scratch for a starter. That seems to be the most obvious tale tail sign to me.

[Springheel]

Putting ourselves on their radar would be a BAD idea. That's the main reason we have to be so careful about Thief assests, because we're already on their radar as a Thief-type game.

 

I'm not terribly worried about this myself. IF the company ever noticed (highly unlikely) and IF they could prove they were actually their models (again, unlikely), then we could just remove them from the mod.

 

I think we should certainly include our own faces...I don't see how they can claim ownership over our own likenesses.

[oDDity]

So are we going to start ripping off 3d content libraries for prop models too? Might as well since no one can prove we didn't make it from scratch and just happen to put all the vertices in the same place for hundreds of props.

 

I can see your argument maybe working for one face, but if we make 200 faces available in whatever format, that in itself is suspicious coming from a mod with a small modeling team. What if they can prove that all those face models share mathematical features that are specific to their algorithm?

 

No problem, I'll just tweak them all a bit.

[sparhawk]

I guess we should delete this thread. And if somebody downloads a model somewhere and tweaks it, then he should at least delete the original model as well.

[Fingernail]

yes, we can deny all knowledge.

[oDDity]

We've no choice anyway, there's no way I can model 150-200 heads, so it's either this or make do with one head per model.

[sparhawk]

Well, I don't know anyway wether a model is done with it or not.

[Ishtvan]

I guess if the model and not just the texture is sufficiently tweaked so they can't prove it was generated with their algorithm, that's fine by me. Just hope they didn't put any secret ratios we don't know about in there or something.

[oDDity]

Yes, and using photofits should help too.

Here's what I got when I fed in my photo.

(Yeah, ok, I'd had a rough day when this was taken)

[Dram]

PUT THAT ON THE COVER PAGE OF DARKMOD SPRING!!!!! PLEEEEAAASSSSEEE!!!!!!!

[New Horizon]

PUT THAT ON THE COVER PAGE OF DARKMOD SPRING!!!!! PLEEEEAAASSSSEEE!!!!!!!

 

Dram...we're trying to 'avoid' drawing attention to the fact we might use Face Gen. LOL

[sparhawk]

Yes, and using photofits should help too.

Here's what I got when I fed in my photo.

(Yeah, ok, I'd had a rough day when this was taken)

 

I always imagined you like that.

[oDDity]

THE best thing about ths is we can have full lip sysnc and facial expressions.

I not sure how it's goign to be done from a codeing perspective though. THe only way I can think of is to do it on a bark by bark basis, so someone will have to go though all te sound speech files and mark down which phenome animations will have to play on which frames during that speech.

 

lip sync.avi

[sparhawk]

For lip syncing, I would expect that we must analyze the soundfile while it is playing and calculate the mouthpositon depending on the current frequency. Or it can be prepared in sync with a particular soundfile to include a motion database, but then it would work only for known soundfiles. I doubt that this would be possible to do without access to the soundengine.

 

That avi is AWESOME!!!

[oDDity]

It's not as simple as that to get GOOD syncing though, that little anim took me about an hour to do just for a few words, and it still isn't matched perfectly.

The problem will be the belnding betwen the frames. Some mouth sounds are snappy, and some are more rounded, like saying a 'pee' or a 'wow'.

IFyou just play the phenomes and let them all blend the same, it'll look like shit.

[sparhawk]

Since we don't really know when a particular part of a soundfile is played, I think lipsyncing will not be possible within Doom 3 right away. Ishtvan said something about an open source sound library (or a free one) that might be used, though. With this it might be more realistic to do this.

[oDDity]

Yeah, but as I said, it could be coded in by having a text file of each soundfile which inculdes the name of the sequence phenome anims to be played.

It's really just the reverse of stating which frames of an anim to play a sequnce of sounds file on, since all we need to do is to specify which sequence of anims is to be played with a sound file.

[sparhawk]

Yeah. I know. But lip syncing would be cooler if it were be able to be done automatically. If we provide textfiles that describe the motion then the only thing we still need to know is when a sound stops playing.

[Dram]

so spar, currently we cannot find at what sound level/pitch/etc a sound is playing right? if we can then we could do this relatively easily, if not then we could only do it either externally, or make a small program which analyzes a sound and writes to a text file some values. These values would be added to the sounds soundshader. Then the game would use those values to determine the facial expressions of the person.

[Darkness_Falls]

Won't the animatability (facial animations) be a dead giveaway Facegen was used?

[sparhawk]

Why? I don't think you can animate the face with Facegen anyway, do you? I only saw controls to modify the face, but not to have complete control over expressions.

[New Horizon]

Can't see how that would give anything away, since we would have to come up with some way to have the animations ingame anyway. Facegen doesn't provide any type of framework to get them working in D3. It would be our own blood, sweat and tears.