Not everyone wants 3D movies, a lot of people think that traditional film converted to 3D effects movie, but often shoddy disappointing. James Cameron think so, but he was determined to break the prejudice, spent $ 18 million to the already after the culmination of the Titanic made into a 3D movie, the original broadcast in 15 years once again return to the silver screen.
April 10, 3D version of the Titanic has been released in the country. Whether you've seen the film, should be able to convert 2D movies into 3D movie technology has been curious about it.
Briefly, 2D-3D conversion technique requires image embedded portions 2D 3D computer graphics (CG) of the space. To create a more harmonious natural 3D images, require large-scale adjustment of hundreds of staff members carried out manually. Even a huge workload, but one can avoid the high cost of purchase of 3D equipment, but also on the other hand, the classic old movies and repair process, eliminating the need for face old style no longer the actors, the lack of late-coming niche charm awkward, so the 2D-3D conversion technology to obtain the favor of a large number of 3D film director. Even the direct shooting 3D movies, it is inevitable to use 2D-3D conversion technology, such as "Avatar."
Cameron once said: " In the 3D conversion technology, the most difficult thing is how to achieve its 'way to the right' to convert ." Director need for each scene to create a kind of three-dimensional visual effects to make decision, the audience needs to determine the extent of the object with different scenarios "close contact", the 3D viewing experience to the audience has a critical influence. For example, monsters and bullets need roared out from the screen, the audience shouted body after shrinking Wow. And belong to the background of the scene, you need to "embedded" into the screen to go, which need to rely on 3D graphics rendering technology to create a sense of perspective and a sense of distance. For almost every frame of the movie in every picture should increase this effect "lots of flesh", but the technology used in different scenarios and very different, even just a little change camera angles, it will bring the overall conversion technology changes.
The primary concern of which there are the following three points:
1. parallax depth is formed
When you create 3D effects, the key step is to make the depth of field. The basic technical principle parallax (Parallax) object when the human eye is formed. Briefly, the parallax difference between the viewing direction is generated by the same target from two points some distance. See from the target angle between the two points, two points called the parallax, the distance between two points known as the baseline. Just know the parallax angle and the base line length, you can calculate the distance between the target and the observer. It is through human beings in this way, the perceived depth information of the object observed. At the same time, some animals in nature also will use motion parallax, moving on their own to get a different point of view. For example, two fields of view do not overlap pigeons, there is no stereoscopic vision, but they head bobbing to obtain depth. The embodiment shown in FIG., As the distance increases from one point to the other side of the test, moving slower than distant objects near object.
Both draw on them disparity processing methods can be taken or converted in 3D movie technology.
However, just make this clear enough, in fact, not only can shoot 3D movies put out two stereo camera is sufficient.
First of all, the director must do the layout of the scene enough "confidence" clearly at the time of film screenings, which objects "almost certain", those scenes to "push of a thousand miles" in order to determine the way the optical axis of the camera, and distance difference angle difference. Imagine shooting team there must be such a stage manager, armed with a stack of scratch paper during filming, to draw more complex calculations than the map on the map, the calibration time shooting position two cameras.
In the 2D movies into 3D movies, due to the possible presence clips and original point of view during shooting to switch, so you first need to screen for each frame are taken to restore the original angle of view, and then determine the position of the 3D virtual stereo camera, about a 90-minute movies up to one hundred thousand-frame image, the amount of calculation and workload are very complicated.
Next, the conversion process, should be considered for different processing techniques to the positive and negative parallax.
Elevation difference means, two sight intersect behind the screen when the 3D effect i.e. the formation of the "embedded" in the screen to. At this point can be simply considered, two substantially parallel line of sight, the average of two with human interpupillary distance of 6.4 cm, on the 3D screen, left and right images should be a difference of about 6.4 cm displacement. However, research shows that most people are not good at the two brain differences more than 2.5% of the images together as one, so the screen for the screen width is less than 2.5 meters, displacement difference if the two images are not restricted, it is likely stereoscopic imaging effect can not be achieved. Therefore, depending on the width of the screen, to the image disparity is handled in different ways.
Negative parallax refers to the line of sight observed when the two intersect is formed in front of the screen. That is, the feel of this object should be run out of the screen close to the audience. The human brain for processing limitations such images, even in the face to be poor. If the number of parallax is too negative, switching frequency is too deep, will cause those suited viewing. At the same time, really good 3D transition effects, it must take into account the different positions of the auditorium have the same viewing experience, whether front or rear, will have the same negative parallax effect.
其实，如果直接拍摄一部 3D 电影，场景可以被事先安排，拍摄角度的变化，视察的变换就不会有频繁的切换，同一画面中的场景遵守相同的深度规则，符合人眼观察物体和人脑处理观察所得的方式。然而，将一部电影转换为 3D 就完全不一样了，很可能在同一场景中为了实现不同物品的深度效果，会带来冲突的信息。如果这种冲突处理不当，就会引起观影者的极大不适应，为影片效果大打折扣。
2. 为电影画面增加深度信息
在 3D 电影制作之前，导演需要拥有一个“深度预算”，即为负视差和正视差的范围。控制在深度视差范围内将确保眼疲劳控制到最低限度，这关乎在某时观看2-3小时的电影而非几分钟的节目人眼的舒适度。该范围要求日益严格。例如，英国天空电视台最近对大部分素材规定了2% 正视差和1% 负视差（以帧宽度的百分比来计量它）的深度预算，短期冲击效应分别不超过4% 和2.5%。
一旦导演确定下来理想的深度预算，就可以来设计场景中的深度信息了。
首先，需要将每幅图想转化为一个深度图，如上图所示，黑白色表示景深范围，白色最近黑色最远。最本原也是最粗略的判断方法是，明亮的更近，深色的更远。这是一些可以将 2D 图片自动转换为假 3D 场景惯用的手段。
幸运的是，实拍出的电影，凭借于常识，大概可以推断出场景中物体之间的位置关系。所以，对整体场景复原建模之后，再确定每幅画面的拍摄角度，就可以对深度信息了然于胸了。然而，一幅一幅手工增加深度信息，仍是一个很大的工作量。这就是为什么泰坦尼克号 3D 版本需要 450 个人员夜以继日地工作才行。
其实，目前的 3D 制作软件都具有非常出色的目标追踪功能，一旦场景和模型确定之后，绑定摄像机视角，就基本不会出现物体深度信息混乱的情况。然而卡梅隆却不愿意享受这样的便利，他十分苛刻地要求工作人员逐帧检查每幅图像转换效果，有人抱怨称这个过程是“十分让人心烦意乱的，就像是用指甲剪修剪草坪”。
一旦每幅图像的深度值都被确定了，那么就可以对每幅图像制作出立体成像。有些视频软件也具有简单的 2D 到 3D 转换效果，比如说 QQ 影音。然而该类软件只是从颜色上加以处理，然后对每幅画面做一个位移差，完全是忽视了场景中物体的立体结构，无法考虑不同视角带来的角度变化。
复杂的转换需要重建物体的网格模型，增加立体信息，包括曲面的曲率信息和深度信息等等。由此得到画面上的复合深度映射，也就是说每个像素点上代表的都是不同物体的深度值的叠加。以确保物体进出视线时，是循序渐进的。确保物体旋转的过程中，具有平滑的过度。这些，都需要符合人类的日常视觉体验。所以，卡梅隆导演要求工作人员在完成双视图之后，必须反复检查每一帧的立体成像效果，精益求精。
3. 3D 的真实效果
在正常的视觉过程中，我们主观判断一个物体的特征时所搜索到深度信息的方式，将加剧 2D 转 3D 的难度。比如说，我们在看 3D 电影的时候，所感知的并不一定是导演所营造的 3D 效果，人脑具有玄妙复杂的信息处理能力。
在高质量的 3D 电影转换过程中，必定会考虑大脑的处理信息方式，作为辅助技巧增强影片质量。然而难度在于，人脑对于深度细节的处理又有着非常微妙的变化方式。比如说，如果两个物体仅仅是遮挡关系，那么立刻就可以判断出，被遮挡的物体应该是远离观察者的。
然而，如果观察类似于下图的这样一个景象，我们会在脑海中构建一个透视图，借助于常识经验，物体会随着距离的远离而变小变模糊，所以观察者判断更大更清晰的物体离自己更近。
对比于人类处理深度信息的这些方式，并且灵活运用这些方式的组合。如果仅仅用视差来转换 3D 深度信息，简直是弱爆了。即便是增加了处理深度信息的算法，也无法灵活地随机应变组合运用，甚至在同一场景中对于不同物体会带来冲突的处理方式。这就是为什么 3D 转换目前还不能用自动的方式来进行，人脑视觉处理信息太复杂了，我们捉摸不透它的规律，还需要迎合它的处理机制。这正是最大的难点。
4. 泰坦尼克号 3D 版的转换经验
卡梅隆称，之前诸多 3D 转换电影的效果不尽如人意，主要原因还是因为其投入的时间和精力不足，太过于追求抢占市场先机，无法全身心投入，确保影片转换质量。同时，他也强调说，在此之前拍摄《阿凡达》的经验将有助于泰坦尼克 3D 版本的转换，这是其他的电影人所没有的切身体会。
不过，阿凡达中使用的 2D-3D 转换技术又与真人电影有所不同，因为原本就存储了物体的三维模型，所以其转换过程只需要在计算机渲染的时候，多增加一个视点信息就足够了。
在泰坦尼克 3D 版本的转换中，卡梅隆与著名制片人威廉·谢拉克（William Sherak）的团队强强联手。带领着 300 名艺术家的团队逐帧去绘制出每幅画面的轮廓线，建立起每个物体的三维网格模型，继而增加视点信息，得到其深度映射图。这项工作不但需要很充分的经验，还需要特别设计的软件来完成，为了泰坦尼克号，全球 3D 影像巨头 In-Three 公司专门就制作出了辅助该步骤完成的软件。
卡梅隆导演在之前接受 IBC 采访时称：“至少目前，不要妄想 3D 转换有什么自动的过程。这纯粹是一个高度主观并且重复性的机械性的人工劳动。必须这么一帧一帧地抓住各幅画面中的深度信息。”
所以，这部电影的转换用了超过两年的时间，一年用来做前期的准备工作，一年用来转换。卡梅隆称，他心甘情愿花费这样的人力物力财力来转变这个电影，并不仅仅是希望搭载 3D 电影的风潮，更重要的是希望能够将泰坦尼克号带回到更大的屏幕上去，为观众们带来更加震撼的视觉体验。
这位苛刻的导演称，即便这样精益求精，《泰坦尼克号 3D》仍不是真正的 3D，只有2.8D，最好的镜头也只做到了2.99D。不过，他说，其他的 3D 电影充其量也只达到了2.4D。