The probability of simulated model entities in TV & adverts

Questions, speculations & updates on the techniques and nature of media fakery
I, Gestalta
Member
Posts: 149
Joined: Wed Jul 25, 2012 9:00 pm

Re: The probability of simulated model entities in TV & adve

Unread post by I, Gestalta » Sat Jun 07, 2014 5:20 am

dblitz wrote:Okay, I get your point, but why the overlay if it's just a real girl anyway? Are we ruled by robots, aliens or very ugly people? B)
I'm not sure if the sunglasses are a "lampshade" of sorts, but in the case that the question wasn't completely rhetorical:

I think one of the most plausible interpretations of this en-masse digitization is, to an extent, akin to a modern video game's public-beta stage---except the user is not told that they are playing a video game. B)

It's a great way to not only work out issues which may pop up in real-time demonstrations and/or fine-tune their software and methods; but it's also an excellent way to slowly "normalize" what sort of visuals are to be expected when one turns on their television. Yet another benefit of these "mistake"-laden instances is that, at any given point, a somewhat bottom-of-the-barrel organization such as RT can be scapegoated for fakery (no matter how innocuous the fakery may be) and most people wouldn't be led to question any other "news" organization. "Yeah, I used to watch RT, until they started using computer graphics to make their news more interesting. Looks like its back to a more trusted news network for me"! Lastly, and this goes more toward my last point & perception-management tactics in general, "if they can't even fake a news reporter properly, how could they have faked Boston, or 9/11, or NASA?", etc. It's funny how these guys can give themselves more credibility by, ironically, lying more.

It's fascinating to me how imperceptive the common person is to this type of imagery and the anomalies therein. There are even people who cannot tell the difference between 24fps and 48fps; or that a film being viewed on a television with the "smoothing" feature activated is not how a film is meant to look. To me, the difference is as stark as B&W films vs. color.

lux
Member
Posts: 1914
Joined: Sat Oct 01, 2011 10:46 pm

Re: The probability of simulated model entities in TV & adve

Unread post by lux » Sat Jul 05, 2014 12:16 pm

How they did this very realistic CGI "Audrey Hepburn" for this recent commercial ...


full link: http://www.youtube.com/watch?v=W55yAyXT5s8

... is explained here.

I suspect a version of this technique is what is being done with Snowden and probably others.

sunshine05
Member
Posts: 307
Joined: Mon Jan 28, 2013 6:00 pm

Re: The probability of simulated model entities in TV & adve

Unread post by sunshine05 » Sat Jul 05, 2014 1:37 pm

Thanks, Lux. I think this is how they created Snowden and other questionable entities too. I'm going to paste the text of the article.

It’s been around for a while now, but that doesn’t mean I’ve stopped gawking at that Audrey Hepburn Dove Chocolate commercial. It’s a lovely ad, but you’ve got to admit that it’s a little spooky. Like J’Adore by Dior before them, Dove Chocolate (Galaxy in the UK) brought back an iconic actress for a TV ad. A seemingly real Hepburn “Moon River”s off of a bus and into the backseat of the “chauffeur’s” car. And like the J’Adore by Dior ad before them, the Dove Chocolate ad is pretty amazing: HOW DID THE DOVE AD DO IT?!?!

Perhaps my grasp on special effects is rudimentary at best (it’s definitely rudimentary at best), but I don’t have any idea how the people behind the commercial did what they did. What is the technology? Is she 100 percent animated? You know, like a Pixar character? Is this an Andy Serkis/Smeagol thing? Or are we looking at an actual person? Could there be an actress out there who looks exactly like Hepburn? Or did they seamlessly piece together old footage of Hepburn? SO MANY QUESTIONS.

I can no longer let these questions go unanswered. I must know how on earth they made the Audrey Hepburn Dove/Galaxy Chocolate ad! Fortunately, the Internet is up to its eyeballs in relevant information. According to PR Newswire, “the visual effects team at Framestore … used the star’s entire film catalog and all available photographs as reference.”

Whoa. Her entire catalog and every photo? That’s a lot of material to wade through. How did they go about incorporating the footage and pics into the ad? “Many scenes required frame-by-frame hand animation techniques to ensure the model was authentic from every possible angle.”

Frame-by-frame hand animation? Tedious/impressive. Wait, a model? Are they calling a CGI’d Hepburn a model? Or is there a real person playing Hepburn? Like I said, I’m a n00b about this stuff. Per The LA Times, according to a publicity handout form Dove, “they first tried hiring an Audrey Hepburn double, then tweaking her features with computer-generated imagery.” I see, I see. There is a real person!

That wasn’t good enough, so they decided to go all-CGI.” Oh. So there isn’t a real person. BUT SHE LOOKS SO REAL!

I need to go to the source. I need to go to Framestore’s website and hope they have a detailed explanation of th— WELL! Would you look at that? They do. How’d ya do it, Framestore?!

The process started with an exhaustive search for the perfect Audrey Hepburn double. The hope was to find someone who could give [director Daniel] Kleinman the performance that he needed, and ideally to share as many of her features and characteristics as possible. Framestore then completed a round of extensive pre-shoot facial scanning at its in-house Capture Lab in order to deliver the complexities demanded by this project. This included a FACS (facial action coding system) session that allowed the team to record more than 70 possible muscle movements and capture high resolution textures for building their CG Audrey.

They found their Hepburn double and recorded her face for the ad. However, their work was not done:

After the shoot, Framestore set about building a facial rig using the FACS head scans as reference for the multitude of shapes the human face can achieve. For each of these shapes, combinations had to be carefully created to allow the face to blend convincingly between expressions during the animation process, which was pivotal to authentically reconstructing the actress’s face. Facial scanning usually provides exact shapes of the person who needs to be recreated, but in this instance, the scans only provided a template based on the double who, although she looked similar to Audrey Hepburn, was nonetheless a different person with different nuances.

The actress’s face and movements were just a foundation/template for the actual ad. Though I don’t understand it entirely, that FACS thing sounds pretty cool. The 3D team built the model of Hepburn, making use of the star’s entire feature film catalogue, plus all available press and documentary photographs as reference.

Ah, so “the model” is the CGI’d Hepburn, and “the actress” is the Hepburn double. Got it. They scrapbooked the old footage and pictures onto the actress’s face, which created the model. However, as there was no technical lens information or measurements available for such old footage, it was quite an inexact science, involving tirelessly tweaking to refine the model from every possible angle.

Like I said, tedious/impressive. Another obstacle: ”perfect[ing] the complex look of human skin.” I’d say they managed to do it. Say it with me now: She looks so real.

CGI-ing lifelike skin wasn’t the toughest part, however:

The biggest challenges in recreating an authentic and unmistakable Audrey Hepburn proved to be the eyes and smile. Although the actress was cast for her eyes, and originally the team had hoped to use the real eyes and build the CG face around them, as post-production progressed it became clear that recognition was key to the success of this ad and, close though the actress was, full CG was the only way to get it right.

http://www.bustle.com/articles/22563-ho ... ake-a-look

lux
Member
Posts: 1914
Joined: Sat Oct 01, 2011 10:46 pm

Re: The probability of simulated model entities in TV & adve

Unread post by lux » Sat Jul 05, 2014 9:59 pm

I think this ad serves as a lovely proof for the skeptics who disbelieve that the technology for human computer simulations of just about anyone is here now.

Evil Edna
Banned
Posts: 117
Joined: Thu Nov 07, 2013 2:32 am

Re: The probability of simulated model entities in TV & adve

Unread post by Evil Edna » Sat Jul 05, 2014 11:22 pm

Amazing! The official Dove video had been sitting on youtube for four months (uploaded March 2nd) but disappears the same day it's linked from cluesforum!

"This video is not available. Sorry about that."

But there's more whack-a-mole to play, as there's another copy here, and a brief description from FoxTV of how it was (apparently) done here.

It's still not perfect though - there's a certain spasticity to the head movements, like it's an animated mannequin, which of course it is, at least from the neck upwards. It's easily detectable as CGI, whereas sim-Snowden is not; so they've retained some slack, or reserve, in the plausibility stakes, for use only in critical operations (like Snowden).

Though they were nowhere near as ambitious with sim-Snowden. While the sim-Hepburn prances around the set in this ad, Snowden has largely been very static in his sim-terviews. So maybe they were constraining to the tech limitations there too.

hoi.polloi
Administrator
Posts: 5061
Joined: Sun Nov 14, 2010 7:24 pm

Re: The probability of simulated model entities in TV & adve

Unread post by hoi.polloi » Sun Jul 06, 2014 12:06 am

I agree. There could be a hundred methods of CGI animating a face photo-realistically, one of the most painstaking digital methods being photoshopping each frame of a rotoscope animation on video. It seems the Peter Joseph, Edward Snowden, Dylan Avery and 'vicsim relative' ones are hybrids of sorts, using actors, some CGI components, motion cap, augmented reality, video, data (face/eyes/nose) scrambling, makeup and so on. Hepburn here looks alright, but she definitely has an artificial, static "I am a mannequin and you are observing me very carefully" kind of look.

hoi.polloi
Administrator
Posts: 5061
Joined: Sun Nov 14, 2010 7:24 pm

Re: The probability of simulated model entities in TV & adve

Unread post by hoi.polloi » Sun Jul 06, 2014 5:51 am

http://www.nydailynews.com/news/nationa ... -1.1828865

This article is hilariously revealing, thanks to a Fakeologist for pointing to it. Sorry if this is a repeat of something already posted. So Snowden appears on a video at a supposedly leftist ACLU (a place and YouTube channel Simon and I have found to have very suspicious CGI characters played as real) conference where a woman gets a seizure from a video. Snowden heroically detected this and using his own blinking screen guided her through it.

Don't be afraid to stare into the over-blinking eyes of epileptic simSnowden, folks. You don't have epilepsy, do you?

This 'robot' screen in which he appears from the shoulders up (much like his interviews where the Snowden entity is barely seen) also was used at a TED conference, the article states. "Snowden" is using a live video to help ACLU attorney Ben Wiz[ard?]ner help a friend recover from video. He appears in video interviews. He gave a TED talk, where the very concept of it is that people are physically in the presence of a physical person giving a compelling speech, through video.

He is fucking video and nothing more.
snowdenbot14n-4-web.jpg
You do not have the required permissions to view the files attached to this post.

sunshine05
Member
Posts: 307
Joined: Mon Jan 28, 2013 6:00 pm

Re: The probability of simulated model entities in TV & adve

Unread post by sunshine05 » Sun Jul 06, 2014 2:04 pm

"Snowden" also had an opportunity to ask Putin a question via video.

Image

Image

http://www.dailymail.co.uk/news/article ... d-Q-A.html

Flabbergasted
Member
Posts: 813
Joined: Mon Nov 12, 2012 12:19 am

Re: The probability of simulated model entities in TV & adve

Unread post by Flabbergasted » Sun Jul 06, 2014 4:27 pm

Cute! Get your own snowdenbot: with and without earplugs. Stubbles are extra.

Image

brianv
Member
Posts: 3959
Joined: Sun Oct 18, 2009 10:19 pm
Contact:

Re: The probability of simulated model entities in TV & adve

Unread post by brianv » Sun Jul 06, 2014 4:52 pm

Flabbergasted wrote:Cute! Get your own snowdenbot: with and without earplugs. Stubbles are extra.

Image
Shadow of glasses on face is free!

Libero
Member
Posts: 333
Joined: Fri Aug 24, 2012 8:21 pm

Re: The probability of simulated model entities in TV & adve

Unread post by Libero » Sun Jul 06, 2014 11:34 pm

This was a nice catch from Abirato of fakeologist.com regarding an observation made by the host of the "No Agenda" show:
At around the 45:00 Adam makes an observation that Sim Snowden is still missing a pad from his glasses’ nose mount one year later. Ouch that’s torture- if you’ve ever worn glasses you will understand.
http://fakeologist.com/2014/05/31/more- ... m-snowden/


Image
http://en.wikipedia.org/wiki/Edward_Snowden

edgewaters
Member
Posts: 91
Joined: Tue Jan 22, 2013 3:49 am

Re: The probability of simulated model entities in TV & adve

Unread post by edgewaters » Sun Jul 20, 2014 6:21 am

Yep, nosepad is missing in every single photo of him, ever. Not only would it be torture, he would constantly have to adjust his glasses to keep them from slipping off. I had a pair with a missing nosepad once. It doesn't stay on your face unless you hold it or tape the middle bit to your nose or something.

Closer view:

Image

I think they're being very cheeky with this pic. Like they're showing it off as a fake.

If you take a good look at all the photos, it's clear that the bit where the nosepad goes is not, actually, broken off. It's just impossibly not there. You can see from the 'intact' spectacle that the two sides don't connect directly to the bridge - the intact side is attached to the piece coming off the bridge, that has the nosepad on it. The other side just ... I don't know. The bridge just kind of dematerializes behind the rim of the spectacle.

simonshack
Administrator
Posts: 6776
Joined: Sun Oct 18, 2009 8:09 pm
Location: italy
Contact:

Re: The probability of simulated model entities in TV & adve

Unread post by simonshack » Sun Jul 20, 2014 11:15 am

edgewaters wrote: Yep, nosepad is missing in every single photo of him, ever.

Indeed ! :blink:


Image
http://nycprowler.com/2014/06/02/oliver ... d-snowden/

Image
http://www.digitaltrends.com/web/edward ... eech-live/


Image


Image
http://www.theguardian.com/world/2013/d ... -documents



Even opticians have noticed Snowden's limp spectacle(s)... B)

"EDWARD SNOWDEN APPLIES FOR EYEWEAR ASYLUM"


http://silverliningopticians.com/journal/snowden/

"Edward Snowden is in desperate need of Eyewear Asylum. We will grant him that if he concedes to never again wear those metal semi-rimless frames with a broken off nose pad–see below."

Image

EYEWEAR ASYLUM FOR SNOWDEN NOW ! :P

icarusinbound
Member
Posts: 393
Joined: Mon Nov 28, 2011 8:49 am

Re: The probability of simulated model entities in TV & adve

Unread post by icarusinbound » Sun Jul 20, 2014 12:56 pm

The missing glasses-pad is inarguable, excellent observation. And how about the optics of the lenses themselves, or especially the one for Snowdon's right eye? (left as viewed). It is represented as being a much-higher diopter thickness, which is evinced by those surreal indents, below his right temple. Human frailties....perhaps.

And yet....the eye itself, overall, is never noticeably displaced, from the viewer's perspective. Oculus dextra, with it's intercept of the eyelid as viewed through the lens sometimes/often too perfectly meeting the lower eyelid (as visible directly from behind the glasses). The right pupil is (in my impression) too symmetrically represented, which adds to the attempted 'realism' of projected contact with the viewer, but not in a genuinely realistic fashion

Is this all ophthalmologically credible? Granted, photographs that are taken/selected in any circumstance will always be the best representation of the subject, but, there could be a strong impression that a photographic work-up, a montage, is going on around Snowden's right eye.

[edit]What the blazes is that dark object above/behind Snowden's right ear, in the upper picture??[/edit]

ps I think the lettering on the side of the (aways) left leg of the glasses says 'Superdry'. An omni-present brand. But...is that non-script/copperplate logotype somewhat atypical?? I couldn't find it...perhaps I'm just being short-sighted https://www.google.com/search?q=superdr ... 74&bih=595

simonshack
Administrator
Posts: 6776
Joined: Sun Oct 18, 2009 8:09 pm
Location: italy
Contact:

Re: The probability of simulated model entities in TV & adve

Unread post by simonshack » Sun Jul 20, 2014 1:35 pm

icarusinbound wrote: ps I think the lettering on the side of the (aways) left leg of the glasses says 'Superdry'.

Wouldn't Snowed-in's glasses be meant to be this Burberry model ?
Image
http://www.opticsplanet.com/burberry-ey ... enses.html

Post Reply