Coquettish Rebirth

Nolan soon discovered that it is really difficult to learn the design of big scenes from comics.Or rather, apart from Jia Hongjian, it seems that others are not able to learn big scenes from comics, because whether it is Japanese or American comics, how many big scenes can there be in the actual picture?Can the big scenes in comics compare with Hollywood blockbusters?Can it compare with Michael Bay's explosion scene?Incomparable!

Of course, even though he said so, Nolan couldn't just give up like this, so he still learned from comics and other big scenes in movies, and on the other hand, he stepped up and started drawing the storyboard.Nolan is of course not a professional painter, so the draft he drew is just a rough sketch. To put it bluntly, it is similar to a cartoonist's draft. In fact, the drawing is not detailed, but it roughly outlines his head. It's just a picture of the imagined scene inside.

It still takes a long time for Nolan to draw the storyboard draft. After all, compared with editing and writing, this thing is a method that Nolan is not familiar with.While he was working on the shot draft, Jia Hongjian was not idle.He is instructing everyone how to capture facial expressions in Pixar Studios!

That's right, expression capture!Although the technology of facial expression capture was "eliminated" by Nolan, Nolan was unwilling to use it desperately, and Jia Hongjian also "spoiled" Nolan and nodded that he would not use this new technology.But it doesn't mean that Jia Hongjian doesn't study it, right?After all, whether we have it is one thing, but whether we need it or not is another matter!

Speaking of which, this facial expression captures it.In fact, it's also quite interesting. Generally speaking, the motion capture is made more detailed in terms of expressions.The technology of motion capture has actually begun to appear in the 50s, but the motion capture technology at that time was relatively primitive, using mechanical capture technology.What is mechanical capture technology?In short, it is a set of things like the exoskeleton that the U.S. military is developing. It is connected to the torso and joints of the human body, and then through the movements of the human body, the mechanical skeleton and related mechanical joints perceive specific The magnitude and angle of motion can then be transmitted and recorded.In fact, to put it bluntly, this technology was not first used in movies.Instead, it exists as a spatial positioning technology!

Even into the 80s.This kind of technology still exists and is used to a certain extent, but it is not used in movies, but in the field of robot control—for example, after only making a robot hand, and then wearing a mechanical glove by a real person next to it .The robotic arm can completely simulate the movement characteristics of the human hand wearing the glove.However.Such a mechanical motion capture has a disadvantage.That is, the data is relatively lagging. It may take half a second for the robot arm to follow the movement after the person moves.This kind of delay may not be a big problem if it is placed in film shooting.However, if it is used in the control of mechanical arms and robots, it may be a bit of a headache.

In addition to the earliest mechanical transmission, the optical motion capture system that is most used in Hollywood today appeared in a few years later.This kind of optical motion capture system is easy to say, it relies on the traditional positioning method in the three-dimensional environment as the core of the technology - the most traditional method of special forces and the like to determine the target distance in the wild is to erect a large Point the thumb at the target, then close and open the left and right eyes respectively, and multiply the horizontal image difference between the target with the thumb through the different angles of the eyes to multiply a coefficient, and then you can get how far the target is from you!

This method is what the People's Liberation Army calls the "eye-jumping method". It is the most old-fashioned positioning method for special forces, snipers, artillery, etc. without a rangefinder.It is said that this old-fashioned positioning method can basically achieve a distance of 200 meters after rigorous training, and the distance judgment gap does not exceed 5 meters!This method actually uses trigonometric functions and occult methods to locate. Simply put, it is similar to the positioning of the human eye—if the human eye only opens one eye, what is actually seen It is flat, and it is easy to fail to distinguish the difference between distance and distance, especially the distance between distance and distance - for example, if you use sewing thread to thread the eye of a needle, if you only use one eye, it is easy to miss the eye of the needle!

And if you look at it with two eyes, because the angles are different, the picture that people feel after processing by the brain is actually a three-dimensional picture, and it is actually easy to judge the difference between distance and nearness!And this principle can be developed into an eye-hopping method, and it can also be developed into a solution for optical motion capture!To put it simply, it is to put a lot of cameras around the actor in a room, and then put some "light spots" on some key points on the actor's body that are easy to distinguish from the background, and then as long as the two cameras can simultaneously Capture a point, then through the calculation of the computer, theoretically the position of this point in space can be determined!

So after the 90s, when making movies and games like motion capture, it was to let the personnel wear a dark monochrome tights, and then paste some circles on the head, hands, arms and joints. A blank sheet of white paper, and then a dozen cameras were placed around the captured person to shoot continuously.Then basically you can calculate a "skeleton" motion model in space through computer calculations, and then use CG technology to get textures and skins for binding, which can make a relatively good motion capture. , you can make the movement characteristics of virtual characters more like real people!

In addition to this optical capture, there are also acoustic capture systems and electromagnetic capture systems.Acoustic capture is relatively simple, that is, replace the camera in the optical capture with an ultrasonic sound generator, then replace the white paper on the personnel with an ultrasonic receiver, and then replace the connection between the computer and the camera with a The computer is connected to the receiver on the person, so that the distance between the ultrasonic receiver and the ultrasonic sound source can be calculated to calculate the position of this point in three-dimensional space.And the electromagnetic capture system is similar to this ultrasonic wave, except that the ultrasonic wave is changed into an electromagnetic field.

But what about the ultrasonic system and the electromagnetic system?Each has its own limitations. Although the ultrasound system is cheap, the problem is that the accuracy is poor and the lag is serious.As for the electromagnetic system, the investment is relatively large. Although it is accurate, it requires a room without any metal objects as the capture site. Otherwise, the electromagnetic field will produce magnetic field distortion due to the existence of metal objects, although the corresponding speed is fast. , but it is cumbersome to operate.

So overall.Such a motion capture.Optical capture systems are still popular now.For example, Jia Hongjian and his company set up a system for Moshen Company, a well-known motion capture company in the United States. It was a new system launched in 99, and the overall computer configuration was Pentium 2-450cpu. Plus ten digital cameras.Such a system that can collect up to 70 points in real time.At the same time in such a system.There is also a data processing system developed by Moshen Company itself.With such a data processing system software, the computer itself will analyze the collected photos and then perform data calculations.What is finally presented to the staff is the output bone motion model in a three-dimensional space!After the model is taken out, it can be used directly as the skeleton of the cg animation, as long as the textures are done and bound up!

Jia Hongjian and the others set up such a system this time, but they didn't actually spend any money, because they "invested" with the concept captured by facial expressions.Just like Nike and Jordan jointly launched the "Flying Man" series of sneakers, just took Jordan's name and image, but this sub-brand can also be regarded as Jordan's shareholding, and the small money for the shoes sold every year is divided. Give Jordan some!But this time, Jia Hongjian and his DreamWorks are cooperating with Moshen Company to develop a brand new facial expression capture system.After this system comes out, Pixar will release a cartoon to "advertise" this system to show the awesomeness of facial expression capture.At the same time, the Demon God Company will also provide DreamWorks with a set of this system for free, and even the small money sold by this model of expression capture system in the future will be distributed to DreamWorks!

This time, according to Jia Hongjian's proposal, the equipment brought by Moshen Company is all old equipment temporarily modified, mainly for proof of principle-such an expression capture system is said to be pasted on people. The white pieces of paper are pasted on people's faces, the position of the stickers has changed, the placement of the cameras and even the processing procedures have to be changed accordingly, and now, the staff of the Demon God Company will I am busy pasting small pieces of paper on the tester's face, and then trying to see if different camera positions and different numbers of cameras can bring about any different effects.

Jia Hongjian didn't understand and didn't want to understand what the specific staff members were trying. He just looked at the pieces of paper pasted on the tester's face and felt that he was "intensively phobic"—a big old man had a sticker on his face. Dozens of small round pieces of paper half a centimeter in diameter!And they are all pasted horizontally and vertically. When the face is not moving, the pasting on the face is like playing Go, and there are small white dots horizontally and vertically!It looks really weird!Then, when the tester started making expressions, the small dots on his face started to move with his facial muscles... Anyway, it looks weird!

After some testing, these staff found that facial expression capture can actually save money—it doesn’t need to bring ten digital cameras like motion capture. In fact, it seems that they only need to bring in a helmet, and then take out a bracket. The bracket extends in front of the actor's face, and then point two digital cameras at the actor's face!Because the face area is not big after all, basically if the digital camera is in the center, it can take the whole face!So as long as two cameras are placed side by side, this kind of facial muscle movement can be made into accurate data!

But at this time, the Pixar employees put forward a new idea-can a camera be pointed at the actor's mouth?Because American animators always pay attention to lip-synch when doing animation!Even the tongue movements have to match the real people!It's not like the Japanese animation where you just open your mouth and shut your mouth!If there is a camera facing the actor's mouth, even if the lips don't need to capture facial expressions, the staff can make models and mouth shapes according to the scenes shot! (To be continued..)