Coquettish Rebirth

Why do you say that making a missing frame out of nothing in the middle based on two photos from a digital camera is so different from doing special effects?You must know that with the two photos before and after, it is different from making a static CG alone. The former requires you to make something out of thin air, which is in line with scientific principles!It is to be coherent with the front and back!To use an inappropriate analogy, it is equivalent to taking a step, from the moment the right foot leaves the ground until the right foot hits the ground. If the missing 20 frames in the middle need to be created out of thin air, then this is created out of thin air. The number of frames, it must make the movements of the characters absolutely seamlessly connected with the two photos before and after!

It is impossible to say that the 20 frames in the middle look very scientific, but compared with the front and back, it is a little bit out of touch, a little different, and the audience may suddenly jump when they watch it!This kind of rush is very bad!If you simply make a static CG, it is a random painting. It is normal to draw whatever you want, as long as it conforms to the principle of human body movement.But this kind of supplementary frame is like doing cloze to fill in the blanks. There is only one correct answer, but there are too many gaps in such a paragraph!

For example, when Cheng Longfei is in the air, not only the hands, feet and limbs of a person generally move, but hair, clothing, even wrinkles on the skin, shoelaces, etc. may also move!How to discover all the surprises between these two pictures, and then create an intermediate form?This is simply harder than everyone coming to find fault!And most importantly, these are not two pictures taken by the same camera!This was taken with different cameras from different angles!What's wrong with this shit?There is no way to even get a program to detect where the image has changed!The machine can only compare the data of two pictures at most, and compare the colors of each point separately, right?But the two different angles are so different from the original shooting, so the program can't do it intelligently!

Even if the program finds out, so what?For example, within one thousandth of a second, Cheng Long's hair moved slightly by a distance of two millimeters. For example, when making the middle frame, it is impossible to say the distance of the xy axis where the hair tip is located in the before and after photos. An average calculation, and then fake a hair - the hair itself has hair roots, and the hair roots are connected to the scalp, so the trajectory of the hair strands is not moving in a straight line!This should be done according to the principles of physics!Therefore, to repair the trajectory of a hair, it is necessary to create an "intermediate state" of a hair out of thin air.That's enough energy.Not to mention how much hair Cheng Long has!right!Everyone uses Bawang anti-hair loss shampoo!Plus clothes, shoelaces and so on, if you really rely on people one by one to adjust similar to PS, it will really be exhausting, and it might be a 360-degree bullet time like Cheng Long's Jumping Building.It will take a whole year to do it!

Spielberg led the team.Naturally, he couldn't say to use such a stupid method to make this special effect.He thought of another good way, a relatively easy way - that is to build a 3D model by relying on photos from different angles!Then use the 3D model binding.Render the surroundings, make a CG computer image, and then let these parts move under limited conditions according to the physical rules, and then capture the dropped frame in the middle!Then make an image and put it on the film, and then edit it with the original film!

What?Does a cg animation have internal physical rules?Of course there is!This and that are all the same as making a game, not only the model but also the texture, but also the rendering, and finally get into the physical rules!In this way, the whole can try to make this "character model" walk by itself!Only in this way can we save effort!

For example, a simple scene of a person walking on the street, if there are no physical rules, then it will be a waste of time and how strange is it?Even if the modeling and textures are the same as the real thing, but because there are no physical rules, the ground is not necessarily raised!In other words, a pebble on the ground may be just a texture, and this person's feet can overlap with the pebble!It won't kick the stone away with one kick!Is this fake?

At the same time, a simple walking scene will have no skeletal system and no physical rules. As a result, when a person walks, it looks like a human figure made of building blocks that are not directly connected. The position of the neck may become very strange. , maybe the hand swing is wrong!This requires people to adjust bit by bit, every frame needs to be adjusted, every frame needs to be coherent and scientific!Is this possible?If there is only a little error in the second frame, only a little error in the third frame, and then there is a little error in every frame, and the accumulation will be strange and incomprehensible!

So, in the world of special effects and cg, we came up with a smart solution. Just like making 3D games, we have a built-in physics engine and a built-in physical principle of bone movement, so we don’t have to adjust many simple things?It's like using motion capture to capture a person's walking movement, and then just like the show in the extras of a film and television drama, there is something similar to a human skeleton walking in the computer, and the movement of each bone during the walking process The rules of the model are used as the standard and stored, so when we walk as a human in the future, don’t we just need to bind the established model and texture to this bone, and then his movements will be completed naturally?Is it possible to walk like a real person?Does it mean that we set him to start walking from here and end there, and we set the starting point and end point, then the model will walk by itself?

In this case, as long as the initial database and physical principles are accumulated, and the various physical principles and light reflection principles in the entire virtual world are specified and collected, will it be faster and faster to create special effects in the future?This is the secret to Hollywood's industrialized special effects!

So Spielberg thought of a way to supplement the intermediate frame, which is to model.Let's build a 3D model, and after applying various physical principles, let the established model and the actions of the two frames before and after be the starting point and end point respectively, and then let the model move automatically, and then add frames!However, since the 3D model is going to be done like this, why not just use the 3D model to make a completely fake bullet time special effect in the air, and go to such trouble to find the middle frame?On the one hand, this was requested by Jia Hongjian, and Jia Hongjian was the boss.Another aspect is to make 3D models for special effects.That is not born with various physical principles!For example, to make human hair in 3D effect, the best workstation and the strongest technology now are far away from making hair that is one-tenth of real human hair!Moreover, the machines and various texture resolutions of this era are not enough. Even Hollywood needs real actors to act in front of the curtain.Just using special effects to mimic the background or something.Spielberg and the others made a special cg with 360 rotations in the air.Are you afraid that the audience will not be able to see that this is a dummy?At that time, various lighting conditions, various wind speeds, various wrinkles on the face, and even the effect of the skin and the texture of the clothes.There are no photos as a basis for these, can they be made to look like real ones?

Don't be kidding me!In 2012, the backgrounds produced by various cgs were fake and real, but which company said that it could make a seamless one, and no one could tell that it was real people?No!Therefore, even if this kind of stuff is to be made with CG special effects, and it is aimed at people, the special effects must be hidden in real photos, so that one frame is true and the other is fake, so that it can be imitated on the basis of reality!Otherwise, the effect would be really horrible!

However, modeling in this way is also troublesome, especially when it involves various "small parts" on the body - such as hair, shoelaces and other scenes that need to move. Generally speaking, if you are doing character modeling, that is Ask a person to take a picture of the front, side, back, etc., and let the modeler follow it.But there is a requirement for this, that is, the people being photographed should not be surprised!For example, it is impossible to stand at attention when taking a frontal photo, and squat for me in a profile photo. This is a hair model!In general modeling, although it is possible to take one picture per second, due to various requirements and the impossibility of modeling every hair accurately, the personnel turn around and take pictures, that is It's feasible - anyway, the model is just to build a model like ** without hair!

But the special effect Jia Hongjian requested is different. This is to try to be as real as possible, so as not to let people see that the added frames are fake, so the precision of the modeling is very high!It is really possible to model fingers and even shoelaces, hair and wrinkles!And for modeling, it is best when Cheng Long is jumping in mid-air, so the modeling accuracy is the best!But at this time, the 60 digital cameras took pictures with a difference of one-thousandth of a second, and the position of the hair may have changed slightly!Therefore, although the photos taken in this way are from different angles, it is not enough for modeling!

Ever since, after discussion, the Nikon company specially bought 240 digital cameras, which were combined with the previous four circles to become 240 cameras. , while the other 120 were filmed at the same time!Only in this way can we capture the appearance of Cheng Long from different angles at the same moment in the air, and then accurately model it!Then let this model change according to the time!

This seems to be very, very troublesome, but Spielberg and the two teams he led were very excited about this troublesome request-you know, this is a brand new shooting method, which has never been heard or seen before. of!Moreover, there is another key point!If they can do well this time, then after returning home, can everyone get together and discuss how to make a software, a system that automatically models and then automatically complements frames after having different photos? ?That's right!When Hollywood does special effects, the core idea is to turn everything into tools and accumulate everything!Even if it costs a lot of money, as long as it is not foolishly relying on human experience every time, but becomes an automatic stuff, then it will be done quickly in the future!This is the increase in labor productivity!If it can be done successfully in the future, even if it is selling this frame-filling software, how much can it sell for!Not to mention that if they don't sell it, if others want to do similar special effects in the future, they have to find them to do it!This is a business opportunity! (To be continued..)