This weekend marks the return of HBO's "Game of Thrones," with season five expected to bring more blood, intrigue and Daenerys Targaryen’s famous fire-breathing dragons.

Pixomondo’s VFX supervisor Sven Martin, who designed the flying monsters, recently revealed how his team was determined from the beginning to ground the look of the dragons in reality. To that end, they turned to a very unthreatening animal for a little inspiration: a chicken from a local grocery store. 

"I was looking for an animal so we can really discover how the muscles underneath should work,” Martin, whose studio took over the dragon design in season two, told The Washington Post. “I called over all the animators and they all had to just play with the chicken … You could feel how the muscles underneath are moving and what are the restrictions, where the bone can’t go. We built our dragon basically the same way."

The dragons' chicken similarities did not last long, however. For season three, Pixomondo accounted for their quickly growing bodies and early flight by using the wingspan of a bat and the posture of an eagle in mid-flight. As for their temperament, Martin told Vulture his team was inspired by the angry reactions of the common swan, as well as the raised frills of lizards. 

"One of the things that we played with that’s also new is frill and what it would look like when they’re mad. There are real lizards that have it. Or you may have seen it before in the first 'Jurassic Park', where the dinosaur is screaming at the guy in the car."

As anyone who has watched the show can tell you, Daenerys' dragons have only become meaner, larger and more complicated to generate. “They are growing with every season, and the textures are getting 10 times more complex,” Martin recently told Variety. All this spells good news for fans looking forward to seeing what adult dragons are capable of and bad news for all those characters who stand in the dragons' way. 

Check out a demo reel of how Pixomondo created season four's dragons below.

Related on MNN: