Canopus Video Editing System Base Technology
Canopus非线性编辑系统基础技术白皮书（中英文对照）
www.canopus.com
© 2005 Canopus Co., Ltd. All rights reserved.
Canopus and EDIUS HD are registered trademarks of Canopus Co., Ltd.
Canopus和EDIUS HD为Canopus Co., Ltd公司的注册商标
All other trademarks are properties of their respective holders. All specifications are subject to change without notice.
其它商标属于各自持有者，如没有特殊注释所有规范受其约束。
About this White Paper
关于本白皮书
Proprietary technologies designed to meet the rigorous demands of the industry are the key to why Canopus video editing systems outshine the rest.
专利技术成为在严格的工业标准下，Canopus视频编辑系统胜过其它产品的关键。
This white paper describes the technology for handling video images that is the foundation of all Canopus video products.
本白皮书描述了所有Canopus视频产品处理视频影像的技术基础
Canopus Video Editing Solutions Deliver
Canopus非线性编辑解决方案概述
Speed to guarantee world-class, real-time performance
世界级的速度保证及实时性能
Canopus video editing systems feature high-performance, real-time video processing for editing of SD and HD formats.Accuracy to avoid image degradation
Canopus非线性编辑系统实时处理标清格式和高清格式的视频时拥有高性能，极大的避免了图像的损失
The beauty of original, high-quality images can get damaged during any or all three phases of video processing—capture, editing and outputting. Canopus has developed its own algorithms and processes that work to minimize degradation of image quality while maintaining real-time performance.
高质量完美的图像在三个阶段会带来损失－采集、编辑和输出。Canopus利用独有的压缩算法在保证实时性能的同时把图像损失降到了最低。
Flexibility to support a wide range of formats
适应性支持更广泛格式的视频数据
Even though Canopus develops its own video processing technologies, our products have always been compatible with industry-standard components. Canopus editing systems can be seamlessly integrated into your existing work flow, and support for open standards makes it easy to edit mixed video formats.
即使采用自主开发的视频压缩技术，Canopus的产品也遵循工业标准，Canopus非线性编辑系统可以完美的结合您的工作流程，轻而易举的支持开放标准的视频格式的混合编辑。

About Canopus
关于Canopus
Headquartered in Kobe, Canopus Co., Ltd. is Japan’s largest maker of computer graphics and video editing products. Canopus released its first video editing system in January 1998. Since then, Canopus hardware and software products for video applications have gained international acclaim as video editors around the world recognize the extraordinary qualities of Canopus video technology. Today Canopus offers a complete line of video capture, video editing and graphics accelerator solutions. Canopus products are delivered to customers in 55 different countries, through sales offices in the United States, China, Germany, the UK and Australia.
总部位于神户的Canopus Co., Ltd.是日本最大的计算机显卡和视频编辑产品的制造商。公司第一款视频编辑产品于1998年1月问世，自此，Canopus应用于视频编辑的硬件和软件产品由于非凡品质的视频技术得到了全世界视频编辑人员的认可和国际组织的赞誉。今天，Canopus提供的完整视频采集、视频编辑和图形加速器解决方案，通过美国、中国、德国、英国、澳大利亚等分公司，产品畅销55个国家和地区。

Technology Built for Speed and Flexibility
技术源自于速度和适应性
The high performance technology built into Canopus nonlinear editing systems efficiently handles a wide range of audio and video formats, making the creative process extremely fluid. Our deep understanding of the PC platform’s architecture, extensive research and experience optimizing hardware and software for the PC, and our long-standing commitment to providing real-time editing solutions to our customers continue to be factors in our ongoing efforts to provide technology that is fast, accurate, flexible, and reliable.
高性能的技术使得Canopus非线性编辑系统可以有效的应用广泛的视音频格式，并使这项创新的处理相当的流畅。我们很透彻的理解PC平台的架构，深入的研究和丰富的经验使硬件和软件针对PC进行了最优化处理，我们承诺将通过持续不断的努力提供给用户的非线性编辑解决方案快速、准确、灵活、可靠的技术。


1.
Scalable technology
1.
可升级技术
Scalable technology allows product performance to increase in proportion with the speed of the PC’s CPU. This technique, which has been perfected by only a few select companies, relies on a balanced combination of the following technologies:
可升级技术可以使产品性能随着计算机CPU速度的增加而提升，这项技术依靠下列一些技术的稳定结合被少数的几家厂商完美的运用。
• PC technology
计算机技术
• High-speed software codec
高速软件编解码
• Easy algorithm optimization
简易优化运算法则
• Support for multi-processor systems
支持多处理器系统
• Optimized software that leverages the CPU’s built-in instruction sets
依据CPU建立的指令设置优化软件
• Buffering technology for faster hard disk drive access
缓冲区技术加快硬盘访问速度
Scalable technology guarantees that Canopus products place no limits on real-time processing, because they can　take full advantage of increases in PC performance capability. The key to putting scalable technology to use is the appropriate allocation of hardware and software processes.
由于PC性能的提升带来的优势，可升级技术保证了Canopus产品在实时处理方面无限的空间。可升级技术的关键是利用硬件和软件处理合理的分配。
Improvements in the processing speed of PCs are made every day, but the speed of fixed hardware stays the　same. That is why Canopus focuses on the design of both software and hardware. As a result, Canopus systems　depend on software rather than hardware for many of the arithmetic processes including video effects and codec　applications. By eliminating the possibility of fixed hardware speed impeding performance, improvements made　in CPU speed are directly reflected in NLE performance. Scalable technology means that improvements in CPU　performance are directly reflected in overall system performance.
计算机的处理速度每一天都在进步，但是同一种硬件的速度是不变的。这就是为什么Canopus的设计定位在软件和硬件二者兼顾。正如结果所表现的，Canopus系统对于多种视频效果和编码算法处理依赖软件远胜于硬件。排除固定硬件瓶颈限制的可能性，CPU速度的提升直接影响非线性编辑产品的性能。可升级技术意味着随着CPU的提升直接带来整个系统性能的提升。

2.
Software vs. hardware
2.
软件和硬件
Why should the software perform mathematical processes?
为什么需要软件完成精确的运算处理呢？
When hardware is used to perform mathematical processes (for example, codec encoding/decoding), the main　advantage is that it takes the processing burden off of the main CPU, freeing it up to perform other tasks. This is　a point some of our competitors like to state in their comparative advertisements.
当硬件被用于完成精确的运算处理时（比如：编码/解码），主要的好处在于基本脱离了CPU的运算，解放它执行其它的任务，我们的一些竞争对手喜欢在他们的对比表体现这一点
Why did Canopus choose a different route? The answer can be found by looking at the basic computer architecture,　as seen in the figure below (note that the arrow thickness denotes relative speed).
为什么Canopus会选择不同的路线呢？答案是我们发现着眼于计算机平台基础会带领我们进入以下的流程。（注意箭头的粗细表示影响速度的关系）
[attach]20198[/attach]
Dedicated hardware and hard disk drives are usually connected on opposite sides of the I/O. Therefore, data transmission from the CPU to the hardware must go through the very slow data transmission path between the CPU and I/O. When processes require transmission between the CPU and memory only, data is transmitted at high speed and therefore the processes can be completed in a very short time.
专用的硬件和硬盘通常是被连接到I/O接口的两端，因此，在CPU和I/O接口的数据之间的传输，从CPU到硬件必须经过非常缓慢的数据传输过程，当处理命令仅仅是在CPU和内存之间传输，数据经过高速的处理，因此处理过程可以在很短的时间完成。
Let’s look at an example where codec processes are executed on hardware and the effects this has on software.
让我们看一个通过软件添加一个滤镜由硬件上的编码器执行的例子：
For image data to be decoded, processed, and encoded, the following round trip route will be taken:
图像数据经过解码，处理，编码，按照下面的流程进行：
[attach]20199[/attach]

Codec processes executed on hardware

利用硬件完成编码处理

For images to be processed in both the hardware and CPU, even if it is only being decoded, large amounts of data must be transmitted through the PCI bus twice. Additionally, each node must consider the timing of the transmission, which cannot surpass the speed in the bottleneck (lowest-speed) area.
图像经过了硬件和CPU的双重处理，即使仅仅是在被解码的过程中，大量的数据必须通过两次PCI总线传输。另外，每一个节点必须照顾到传输的时间不能超过瓶颈（最低速度）范围的速度
On the other hand, the route the image data needs to take when a software codec is used and all processes are executed on the CPU is as follows:
另一种方式，当软件编码器被用于通过CPU处理执行所有的处理任务时，图像数据如图所示：
[attach]20200[/attach]

Codec processes executed by software

利用软件完成编码处理

In this scenario, all processes can be completed in the high-speed transmission area so high-speed CPU and memory performance can be fully utilized. Compared to the data transmission speed between CPU and I/O, the transmission speed between the CPU and memory is 50 times faster. Valuable time is not wasted when a software codec is used. This is especially important when working with HD video, where very large amounts of data are processed. Considering the improvements that continue to be made in CPU performance, executing processes on high-speed hardware only increases needless time spent on data transmission and decreases performance of the system as a whole. However, it is necessary that the CPU handle the software codec at a sufficient speed.
在这个环节中，所有处理完全在高速传输区域内，所以高速的CPU和内存的性能可以被充分的利用。比较数据在CPU和I/O接口之间的传输速度，CPU到内存的速度要快50倍，当软件编码器被利用时，宝贵的时间不会被浪费。这在高清环境下编辑时是非常重要的，在这个环境中有大量的数据需要处理。考虑到CPU的性能在持续的进步，使用高速硬件执行处理只不过是加大了在数据传输中不必要的时间花费并降低了系统整体的性能，那么使用CPU在充分的速度下控制软件编解码器就是个必然了。
Canopus extensively studied image compression of DCT base images in order to develop a superior software codec that optimizes CPU speed and power.
Canopus广泛的学习了DCT（离散余弦变换）基础这种图像压缩技术为了开发出一种可以优化CPU速率的软件编解码器。
PCI bus bandwidth is another reason why data transmission with hardware should be minimized. In HD video editing, a total of 250Mbytes/sec of I/O bandwidth is required (150MB/sec for video and over 100MB/sec for HDD (hard disk drive). HD systems also require other devices (e.g., gigabit networks) that utilize PCI bus bandwidth.
PCI总线带宽是为什么数据传输应该最小化使用硬件的另一个原因。在高清视频编辑中，I/O接口至少需要250Mb/s的总带宽（150MB/s用于高清视频，超过100MB/s用于硬盘读写）。高清系统也需要利用PCI总线带宽来支持其它的设备（例如千兆网络）
Unfortunately, PCI bus usability is much lower than expected—if it exceeds 70%, it is said to be very efficient. When image processing is executed in devices on the PCI bus, the system does not have much power left for executing other processes.
Flexibility is lost when all processes are executed on the hardware.
然而，PCI总线比预期的效率低得多，如果超过70%就可以说非常有效了。当在PCI总线上的设备执行图像处理时，系统无法分离出更多的能力用于执行其它的处理。当所有的处理都在硬件上执行时将失去系统的灵活性。
You may think, “If executing only some of the processes on the hardware is inefficient, why not execute all processes on the hardware?” That is a good question. Suppose there were an ideal hardware product that could execute all processes, from codecs to effects, including various composites. PCI bus transmission would only be required for input and output, thereby eliminating the bottleneck. This piece of hardware would probably be very expensive and, more significantly, it would be fixed at the design stage, without flexibility to accommodate changes in the system as technology progresses. For how long could you plan to rely on this system? Think back through the years you have worked in the video industry… Is what was the “latest hardware” on the market when you began still so “hot” now? With computer and video technology continuing to make remarkable progress every day, upgrade costs must be considered even more than initial costs.
你可能在想
“如果仅在硬件执行一部分处理效率低，那为什么不在硬件上进行所有的处理呢？”这是个好问题，假设有一种完美的硬件产品，可以执行所有的处理，从编码器到效果，包含不同的部分。PCI总线仅仅是需要排除输入和输出方面的瓶颈，但这个硬件的整合或许是非常昂贵的，并且值得注意的是，它被限定在一个固定的时期内，完全失去了技术进步给系统带来的扩展性的改变。那你计划使用这个系统多久？回想您在视频行业工作的这些年，现在市场上“最近的硬件产品”还是您刚开始那么热的产品吗？随着计算机和视频技术日新月异的发展，高消费比起基本消费一定会非常慎重。
Compared to hardware, software can be much more responsive as technology advances, with shorter development time to market and without “limits” on making functional improvements. Software is inherently more flexible than hardware, and scalability applies not only to speed but also to functionality.
相对硬件而言，软件能在短时间内开发上市并在功能改进上没有上限，从而更加接近技术响应，软件比起硬件具有固有的灵活性，同时无论在速度上还是功能都是无可限量的。
Real-time processing with scalable technology
可升级实时处理技术
The graph below compares the real-time processing of two video streams + one title by utilizing hardware versus a real-time processing system based on scalable technology. Real-time processing can be executed in the area below the horizontal real-time line.
下面的图表中体现了通过硬件和基于可升级技术的实时处理系统制作2层视频＋1层字幕的对比，在水平线下的区域中表现实时处理。
In the scalable technology system, as the processing becomes heavier, the processing time becomes longer and, when it reaches a certain point, real-time processing is no longer available. The load, however, is basically stable.
在可升级技术的系统中，处理变化非常大，当达到一个可靠的点，处理时间变长，就不再需要实时处理，无论用什么方式加载，都是基本稳定的。
On the other hand, the hardware processing system can execute processes in real-time up to a limit but, when it goes over that limit, rendering is needed and processing speed suddenly drops, meaning that overall processing time increases.
另一种，硬件压缩系统中只能实时执行处理到上限，当超过这个极限时，就需要生成同时会突然出现丢帧现象，这意味着增多了全部处理的时间。
After that, since the hardware-based system was not designed for executing processes using software, the processing time needed rises at a much steeper rate compared to the system using scalable technology.
从这以后，以硬件为基础的系统没有办法的采用软件进行压缩处理，比起使用可升级技术的系统来，处理时间直线上升。
[attach]20201[/attach]
Pursuing the optimal balance between hardware and software processes Canopus editing systems do, however, selectively employ hardware. Importing and exporting from external devices cannot be executed without hardware. Through fully understanding the structure of data flow, Canopus engineers have alleviated bottlenecks by ensuring that hardware processes do not get in the way of high-speed software processes. So long as slow transmission between CPU and I/O is minimized, the negative effects of relying on hardware will be mitigated for the system as a whole.
Canopus编辑系统实际上做的是有选择性的使用硬件来达到硬件和软件处理的平衡，从外界设备的导入导出无法脱离硬件，通过全面的解读数据流的结构，Canopus工程师已经突破了通过硬件处理无法达到高速软件处理环境的瓶颈，从CPU到I/O接口的低速传输做到了最小化，消除了整套系统依靠硬件处理的消极影响。
Canopus allocates all processes that can be completed between I/O hardware and external devices to hardware (including scaling of image data, digital filter processing, and adjustment of quantifying bit numbers). Because these processes can be executed without intervention of the CPU, data transmissions between the CPU and I/O are not increased.
Canopus分配从硬件I/O接口和外部设备到硬件所有的处理（包括变换像素比，数字量化处理，调整量化标准）。因为这些处理可以不需要CPU的参与执行，CPU和I/O接口之间的数据处理不会增加。
The appropriate allocation of processes between software and hardware is critical to the configuration of Canopus’s high-speed editing systems. System performance is further improved by utilizing a buffering network to absorb the differences between the video transmission rate of the external I/O and the processing rate of the software.
适当的分派软件和硬件之间的处理对于Canopus高速编辑系统的架构是很关键的，系统性能的提升是通过缓冲区处理外部I/O接口视频传输速率与软件处理速率的差异处理完成的。

1.
Optimizing speed and performance
3. 最优化的速度和性能
In-house development
自身的发展
Canopus develops and engineers all of its own systems—hardware and software—in house. This is what makes overall system optimization possible, including hardware architecture, codec, and disk I/O.
Canopus研发工程师自主开发硬件和软件，使制造出包括硬件、编码器和磁盘存储在内的全面优化的系统成为可能。
Optimizing software at the algorithm level
最优化的软件算法
Peephole optimization (local optimization) methods are typically used in software development. But that alone will not drastically speed up processing. By optimizing at the architecture or algorithm levels, dramatic improvements can be achieved. A full understanding of the CPU is required for algorithm optimization.
Peephole优化的方法（本地优化）是有依靠软件的发展为代表性的。但是仅单一的发展软件是达不到处理速度本质上提升的。通过优化结构或者算法可以达到引人注目的进步。算法的优化需要全面的理解CPU。
Most video editing applications do not make use of SSE and MMX, which are multi-media command sets (even when they are used, areas used are typically limited for peephole optimization only). EDIUS, Canopus’s nonlinear editing software, is built on Canopus’s own architecture design, which is optimized using SSE and MMX, and was developed to speed the rendering path up to its limit. “3D Now!” is also used where applicable.
大多数的视频编辑应用无法使用SSE和MMX这种多媒体命令程序（即使他们用到，也仅仅是限制在Peephole优化的范围），EDIUS，Canopus的非线性编辑软件，建立起了Canopus自己的设计理念，优化使用SSE和MMX技术是发展渲染速度到上限的路子，“3D Now”在这里也有用到。
For plug-ins, as well, Canopus pursues speed without sacrificing accuracy by optimizing at the algorithm level.
做为插件，Canopus利用优化算法，没有丧失清晰度的情况下发展速度。

Maximizing disk performance
最优化磁盘性能
By reducing “seek” in optimal scheduling of disk I/O, reading HD broadband data was made possible on IDE software RAID systems. This is also significant for editing DV footage. The ability to read an HD stream is equivalent to being able to read four DV streams concurrently. This facilitates multi-layered composites and editing on slow, fragmented disks. It also means that unnecessarily high-speed hard drives no longer need to be used.
减少磁盘读写的寻道时间使高清广播数据流使用IDE的RAID系统成为可能，这也影响到了DV场景的编辑。读HD数据流相当于同时读4个DV数据流，这推动着多层合成和在低速多块硬盘上的编辑，这也意味着未必像从前那样需要使用高速硬盘。
Reducing memory access
降低内存访问
Adjusting memory access will improve efficiency somewhat, but this optimization method is to be expected.
调整内存访问可提高少许效率，但是这种优化方法是有发展空间的
An even more effective mechanism for improving speed is eliminating unnecessary copies. For example, in a system where programs for reading data from the hardware, for processing images, and for outputting from the hardware are all run separately, many unnecessary copies will be made. Many popular video editing applications are designed this way. Full-frame playback of HD is problematic with this kind of technology. EDIUS software was designed considering “data flow” from input to output, resulting in systems that do not make unnecessary copies of video data.
一种更有效的提高速度的方法就是减少不必要的复制。举个例子：在一个系统中程序从硬件读数据、处理图像、输出图像所有都是分开处理的，这里产生了很多不必要的复制。许多流行的编辑软件愿意采取这种方法，全帧高清画面回放面就临着这个问题，EDIUS软件采用了原始的数据流从输入到输出，结果就是在系统中没有产生不必要的视频数据复制。
Canopus Video Editing System Design—Keeping Total Workflow in Mind When it comes to speed, faster is simply better. But when considering “image quality,” as accuracy is improved, the amount of data increases and this can degrade the efficiency of the system as a whole. Increased amounts of data mean increased storage costs and lower throughput.
Canopus视频编辑系统设计保持全部工作流程智能化，那时达到速度更快将更简单。但是考虑到图像质量的提升，总数据量的增加会降低整体系统的功效，增大的数据量意味着增大存储空间并降低制作能力。
[attach]20202[/attach]

As you can see, it is important that the NLE system be designed with optimal components and is not “overspec’ed.”
正如你所看到的，非线性编辑系统保持原始结构不做转换是非常重要的。
Of course, without considering costs, the ideal system may center around a high-speed PC and be one that can process huge quantities of data. But in reality, return on investment (ROI) is a very important aspect of an editing solution; one that must be considered. You should think about the system in its entirety, ensuring
the optimum specifications are achieved for the right ROI.
当然，不考虑成本，理想的系统应该是围绕着一个能处理大数据量的高性能PC，但事实上投资回报率是非编方案一个非常重要的因素，也是必需要考虑的因素。你应该考虑到系统各个方面的因素，确保最合适的性能换取最佳的投资回报率。
When considering the system as a whole, compression technology becomes a critical feature. Compression that can reduce the quantity of data without any degradation is ideal and is the only way to achieve the balance between quality and data size. Canopus compression codecs achieve this balance.
当考虑整套系统时，压缩技术就成为了很重要的指标了，压缩如果能在不降低质量的情况下减少数据量是最理想的，达到质量和数据量之间的平衡是唯一的方法。Canopus压缩编码就达到了这种平衡。
1.
Standards based on workflow
1. 基于工作流程的标准
How can “optimal design norms” be determined for an editing system? Canopus came up with the answer by examining the total workflow in video production: from shooting to editing to delivery. Here is what we found…
最理想的设计规范能怎样决定一套非线性编辑系统？Canopus在视频产品全部工作流程测试过程中提出了这样的问题：从采集到编辑到输出，这里我们发现了什么……
With the performance (image quality) of VTRs as the norm, input/output of images with higher quality is possible and can be edited with minimal degradation.
把录像机的性能（图像质量）做为标准，输入/输出的图像可能是高品质的也会在编辑时有少许损失。
Norms based on VTRs
录像机基础规范
There is hardly ever a time when a camera or a VTR is not used in the workflow of post-production or broadcasting including coverage, editing, and distribution. When video and broadcast professionals choose editing machines and transmission devices, they look for products that meet the image or product quality levels of the VTRs or cameras they are using. In other words, in a workflow using VTRs and cameras, the image quality level is based on the image quality level of the VTR or the camera used, and that level can be assured so long as the performance levels of the peripherals are equal to or higher than those of the VTR or the camera. Even the quality of computer graphics depends on VTR image quality, when they are recorded on tapes for distribution. The capabilities of the editing machine are also important—it must be able to perform without degrading VTR performance and image quality.
当在电影电视后期制作阶段或者包括报道、编辑和发行的工作流程中很少有用不到摄像机或录像机的，当视频和广播专业人员选择编辑设备和传输设备时，他们会寻找和自己使用的摄像机或录像机画面或产品质量等级相当的产品。换句话说，在使用录像机和摄像机的工作流程当中，画面质量标准是基于的录像机和摄像机上用到的画面质量的。这个标准只要等同于或者高于录像机或摄像机的标准，就可以被确定。当他们被从磁带上录制时，计算机显卡回显的画质依赖于摄像机画质。编辑设备的能力也很重要――它必须能够完成不低于录像机的性能和画质的任务。
Design based on usage
用法的设计基础
There are times when the best image quality available is required, for example in compositing. Canopus systems have been designed for efficient editing of images shot using video cameras, and therefore do not cover composite editing. In composite editing, multiple tracks must be processed simultaneously, so the highest quality images must be processed at high-speed on multi-tracks. In some competitors’ systems, high image quality editing has been adopted, but it can only process 2 streams + 1 track in real time. For multiple tracks exceeding this limit,
rendering must occur. In that case, it is more efficient both functionally and cost-wise to use dedicated compositing software. The bottom line is that purchasing an over-spec’ed editing system without true consideration for its actual use is likely to mean reduced efficiency and lower ROI.
有些时候必须用到最佳的画质，例如在合成过程中。Canopus系统原意是高效率的编辑用视频摄像机采集的画面，因此不包括合成编辑。在合成编辑过程中，多条轨道需要同时处理，所以高质量图像必须在多条轨道上高速处理。一些竞争对手的系统中，采用了高质量的图像编辑，但是仅仅能实时制作2层视频+1层字幕，超出这个限制的多轨道必须进行渲染。如果这样，用专业的合成软件无论在功能和价格上都是更有效的。如果底线是不考虑实际用途而购买一套过于强大的编辑系统就意味着低效率和更低的投资回报率了。
2.
Accuracy for preserving the beauty of original images
2. 精确度为了保持原始画面的完美
There are times when the Canopus codec, with its 8-bit, 1440 resolution, is said to be insufficient when compared to the standard 10-bit, 1920 resolution. The current specification has been chosen based on careful studies of sufficient balance without becoming over-spec’d.
有时当Canopus编码采用8bit，1440采样时，有人说比不上10bit，1920采样。当前的标准是建立在充分平衡、仔细的研究后不超出规范的基础上选择。
Quantifying bit numbers
比特量化方式
Canopus considered quantifying bit numbers for the following VTRs:
MPEG IMX, BetacamSX, DVCAM, DVCPRO50, D-1, D-2, D-3, D-6, DVCPRO HD, HDCAM (8-bit when compressed)
Canopus认为比特量化应遵循以下录像机方式：
MPEG IMX, BetacamSX, DVCAM, DVCPRO50, D-1, D-2, D-3, D-6, DVCPRO HD, HDCAM (均采用8-bit量化方式)
These VTRs are widely used in broadcast environments. This means that even the broadcasters agree that 8-bit is enough for high quality performance. Thus, increasing quantifying bit numbers to 10-bit for improving quality in images used during the editing process can be considered over-spec’ing in a workflow assuming the usage of VTRs.
这些录像机是被广电行业广泛的采用的，这意味着即使广电的用户也接受8bit量化方式足以满足高质量特性。因此，在编辑处理期间增加到10bit量化来提升图像质量对于工作流程中采用录像机用户是超出规范的。
Increasing data amounts from 8-bit to 10-bit means making increases by 20%. That leads to requirements for increased storage capacity and higher data transmission speed.
从8bit到10bit增加的数据流相当于增加的20％，这导致增加存储容量和高速数据传输需求的增加。
Another reason 8-bit is appropriate for high speed processing is that computer technology tends to be based on multiples of eight. Processing 10-bit data requires 16-bit buses and registers, which ends up requiring double the standard 8-bit load.
Video effects processing is designed for the final results to have 8-bit accuracy. For example, chroma keying requiring advanced interpolation, 32-bit accuracy may be needed for international calculations. Computing with unnecessary accuracy requires large amounts of memory and slows down the process. Therefore, eliminating waste by careful consideration of the algorithm implementation with sufficient accuracy is important.
另一个原因是8bit适合计算机中的8进制管理技术进行高速处理，10bit数据需要16进制的管理，结果就是需要双倍于8bit的导入，视频效果处理原理是保证8bit精确度的最终结果，举个例子，色键需要高级处理，内部可能需要32位的精确度的计算，计算机需要在内存和低速下载区进行大量不必要的数据处理，因此通过仔细的考虑执行充分保证精确度的算法对于排除损耗非常重要。
Canopus systems guarantee accuracy at each level of effect processing due to careful planning and design.
Canopus系统在精心设计下保证了每一次效果处理的精确性。
Pixels and sampling methods
像素取样方式

[attach]20203[/attach]
The Canopus codec exceeds that of the two popular VTR systems and can edit by oversampling. Some internal computations, for example composites, are executed in 4:4:4 sampling, minimizing degradation during processing.
Canopus编码器超过了对编系统可以做到超出取样范围编辑。许多类似于合成这样的内部的运算按照4:4:4方式取样，在处理期间把损失降到最小。
YCbCr as the standard format
YCbCr被当做标准格式
RGB is widely used as the norm in PCs, but YCbCr is the norm used in video equipment. Many editing systems include software to convert to RGB for easier processing on PCs. In such systems, color-related problems, such as images over 100IRE being blown-out or discoloration of rendered parts, frequently occur.
Images shot with cameras include elements over 100IRE, therefore being able to process them correctly is crucial.
RGB是在计算机领域中被广泛用到的标准，但是YCbCr是视频设备中的标准。许多包括软件在内的编辑系统为了方便在计算机部分的处理采取转换成RGB的方式，在这些系统中，类似图像放大超过100IRE或者渲染部分变色等颜色失真的现象经常出现。
So Canopus editing systems simply use YCbCr to avoid unnecessary problems.
所以Canopus编辑系统完全采用YCbCr方式来避免不必要的问题、
32-bit audio processing
EDIUS Pro processes floating 32-bit audio, matching audio-dedicated software. SDI can I/O a maximum of 24-bit of quantifying bit numbers, therefore floating 32-bit, which will cover this specification, has been chosen.
32-bit音频处理
EDIUS Pro处理可变32bit音频用于专业的音频软件，SDI可以做最大24bit量化的输入/输出，因为可变的32bit包括这些参数所以被选择。
1.
Reasons for adopting a compression codec
3. 采用压缩编码器的原因
First, we wish to make it clear that Canopus products can process uncompressed data and full HD with alpha channel (1920x1080i). However, there are reasons for Canopus emphasizing its compression codec over processing uncompressed data.
首先，我们希望使用Canopus产品可以清晰的处理无压缩数据和带有Alpha通道（1920*1080）的HD格式。然而，这也是Canopus强调它的压缩编码器胜过无压缩数据处理的原因。
Why compress?
为什么压缩？
When you calculate the amount of data for uncompressed HD images, even with larger hard disk capacities and faster transmission, it still is too much to handle realistically.
当你使用高清无压缩图像的全部数据时，就需要大容量的磁盘空间和快速的传输。对于实际操作还是相当多的。
[attach]20204[/attach]
Directly related to costs
直接关系到成本
In uncompressed HD editing systems, over half of the total system cost goes to storage. Not only is large capacity needed for storing enormous amounts of uncompressed HD data, but also products with the best performance and interfaces to meet the very high requirements for transmission speed in real time must be included. In most cases, these costs by far exceed that of the editing machine itself.
The high-speed, high-quality Canopus codec contributes to a significant reduction in total costs, while preserving quality.
在无压缩高清编辑系统中，超过半数的成本被用到了存储方面，处理如此庞大的无压缩高清数据不仅仅需要大容量的空间，还需要高性能的传输满足实时编辑的高速传输需求。这些成本远高于编辑设备自身，高速高质量的Canopus编码器在保证质量的同时提供了非常有效的降低成本的帮助。
Directly related to performance
直接关系到性能
In HD video editing on PCs, enormous amounts of data are transported between the HDD to I/O, CPU, and memory.
在计算机上编辑高清视频，庞大的数据流在硬盘到I/O接口、CPU和内存之间传输。
For uncompressed HD video, huge amounts of data are transmitted on every bus, especially between the HDD and I/O, which becomes a bottleneck. Using compression, CPU power is used for compressing and expanding.
无压缩的高清视频，庞大的视频数据在每一个总线上传输，尤其是在硬盘和I/O接口之间形成了瓶颈，使用压缩技术，CPU的能力可以被用于压缩和扩展。
However these tasks are handled on the high-speed bus between the CPU and memory. Therefore, the amount of data being transmitted on the low speed bus between the HDD and I/O is 1/7th compared to uncompressed data.
不管怎样这些任务都需要控制在CPU和内存之间的高速总线中，因此，全部的数据在硬盘和I/O接口的低速总线之间采用1/7的压缩数据。
This, as you might surmise, has great effects on the system as a whole.
正如你猜测的，整套系统中可以顺利的完成编辑。
In the future, when an editing system is configured on the network, data compression will be a must. Even with the fastest fibre channel server system available today, it is very difficult to play back three uncompressed HD streams
on five editing machines constantly, in real time.
不久的将来，当一套编辑系统在网络中使用时，必须要经过数据压缩。即便是今天我们看到的最快的光纤通道服务器，在连续实时回放五台设备的无压缩高清数据流时也是非常困难的。
In addition, since Canopus’s editing software (EDIUS) is designed in-house together with our codec, the software is able to fully harness the codec’s capabilities.
另外，自从Canopus的编辑软件EDIUS使用自主开发的编码器后，软件可以充分的利用到编码器的优势。
Excellence of the Canopus HQ codec
优异的Canopus HQ编码器
The Canopus HQ codec provides image-quality stabilization by keeping the average bitrate low, dynamically adjusting compression rates between accurate images and simple images and using variable bitrates, which can be decoded by frames. Additionally, to maintain minimal CPU load for compression and expansion, Canopus’s original technology has been introduced to the mathematical methods of the CPU. While many other HDV editing systems are forced to reduce quality and compress image data to even realize pseudo-real-time processing, EDIUS users
can go on with their editing processes without worrying about decreased image quality.
Canopus HQ编码器通过持续恒定的低码率提供高质量稳定的画质，在高精度和简单的图像之间动态调整压缩码率提供可变压缩码率方式，并可以通过单帧解码。另外，在导入压缩和展开的过程中保持最小限度的使用CPU，Canopus的专利技术引入了CPU的算法，当许多其它HDV编辑系统还在被迫的通过降低质量来压缩图像数据实现虚假的实时处理的时候，EDIUS用户已经能进行编辑处理而不担心降低图像质量了。

Canopus 的白皮书都来了。