ehaofar quyteri benedic

[Stephaine Zitzow]

For Netflix, there are three main goals of color management. The first is to ensure a predictable and repeatable way to view images and convert between different color spaces, based on a variety of capture devices and display types. This ensures that creative intent is maintained throughout production and post processes. The second is to ensure all color decisions and VFX work are done in a scene-referred color space, in order to maximize creative flexibility in post, and third, to ensure that our picture archival masters are of high quality and future proof.

A tech blog post was written by members of Creative Technologies in order to give some historical and scientific context around what we mean by color management, some tips on how to achieve it, and how it helps enable certain archival assets to be generated.

A color pipeline represents the essential color space definitions, transforms, and deliverables for a project. This usually requires knowledge of the source(s), working, and final delivery color spaces to ensure that all points in a color pipeline are aligned. This must be defined in coordination with all departments responsible for viewing, sending, and receiving images on a production.

This is based on the initial image capture device or source color space. Some examples are Sony SLog3 / SGamut3.cine, RED Log3G10 / WideGamutRGB, ARRI LogC / WideGamut. For certain workflows, this could include archival footage or graphics already in a display-referred color space such as Rec. 709 / BT.1886 or sRGB.

In this example, the working color space is Log (logarithmic) and the display color space is Rec. 709 / BT.1886. The Output Transform separates the two, and is only baked in for the Rec. 709 / BT.1886 streaming master. The archival master (NAM) remains in Log color space.

Confidence and trust in the color pipeline means that productions spend less time on support and troubleshooting and more time on creative decision making. Fixing or re-doing work due to color issues can be very expensive and time consuming. This is why it's so important for all parties in the production and post-production stages who are responsible for viewing, sending, receiving, and/or manipulating images to align on the color pipeline.

A color pipeline like the one described in this article enables the possibility of viewing images in different scenarios with more confidence. Whether it is an SDR or HDR display, a monitor or projector, the appearance of the image should be consistent and creative decisions should be maintained. This is only possible with a solid color management system.

Color management, and specifically defining a working color space that contains all of the information captured (or CG generated), helps to protect all of the creative work in a future proof picture archival master. For example, color managed images from past seasons can easily plug into new seasons or projects, VFX or Animation elements can be used across projects, or simply the archival masters from current projects can live on in the archive for historical or remastering purposes.

A high-quality color management system needs high bit-depth processing in order to transform images between color spaces non-destructively. We recognize that certain non-linear color spaces (such as Rec. 709 and Log flavors) exist so that high-quality work can be done at lower bit-depths, but increasingly there are linear representations and complex operations that require 16-bit floating point or higher precision to be done non-destructively. We would consider 10-bit a bare minimum for any professional imaging system handling SDR images, and 12-bit for HDR images. Ideally, all systems would be capable of floating point (16-bit half or 32-bit full float) in order to reduce the chances of clipping or other destructive operations.

This is a crucial point of communication which increases the likelihood of a smooth color experience through a production. Once a primary camera is chosen, or even before, conversations should be had with all parties responsible for the image. This includes the DoP, DIT, colorist, VFX facilities, and Netflix representatives responsible for the project. While certain decisions may not be able to be made at this stage (for example the exact final colorist may be TBD), it is important to have these conversations as early as possible. What camera and recording format will be used? What is the working color space? Will they use ACES? If not, is there an agreed upon show LUT?

Calibration of all displays to industry color standards is essential. This ensures that the color transforms used are relevant, image appearance is consistent, and color decisions are meaningful. Exact methods of display calibration are outside of the scope of this document, but the important standards and target color spaces can be found in our Color Critical Calibration Guidelines.

On-set is usually the first place where color and images are judged, so it is arguably the most important to get right. Hopefully, the pre-production conversations have informed an agreed-upon and widely communicated setup of on-set video signal path, CDL + LUT boxes, and calibrated monitors. The goal of on-set monitoring is usually to have a non-destructive and repeatable color pipeline, in order to give flexibility and consistency to the downstream dailies process.

Dailies is where the color pipeline has the biggest effect on post, since the same camera RAW files which will be used in finishing are used to generate color-baked dailies and editorial media. The on-set color decisions should be communicated via ASC CDLs and be passed along into ALEs which are ingested into the Avid. Those CDLs and the same show LUT / Output Transform from on-set should be applied in the same working color space so that editorial sees the same color that was seen on-set.

The color pipeline culminates at the point of the online conform of original camera files (OCF) and final color grading. Ideally, the colorist should have access to the color decisions (CDL + LUT) that was used for dailies, in order to have a starting point similar in appearance to what has been seen throughout the process. Color grading should be done in the working color space, ahead of the Output Transform / show LUT, in order to protect those decisions for the archival deliverables explained below. This can be achieved by grading in ACES, or by setting up the grading system with careful color management settings to ensure that all creative decisions are applied in the working color space (order of operations is key).

When this type of color management is employed, it enables the delivery of the NAM (Non-graded Archival Master) archival deliverable in a single unified color space. For more information about this deliverable, please see the NAM specification.

As explained above, it is important to ensure that images are stored in the widest possible color space until the very final stages of the color pipeline. This is easily achieved by using ACES, which keeps images in a high dynamic range, wide color gamut space, and allows the user to simply select the Output Transform for the target viewing display. This can also be achieved while working in the dominant camera color space with careful, pro-active color management and the use of conversion and viewing LUTs. As an example, the diagram below shows a color pipeline at the on-set, editorial, and finishing stages.

The goal of the Netflix Open Connect program is to provide our millions of Netflix subscribers the highest-quality viewing experience possible. We achieve this goal by partnering with Internet Service Providers (ISPs) to deliver our content more efficiently. We partner with over a thousand ISPs to localize substantial amounts of traffic with Open Connect Appliance embedded deployments, and we have an open peering policy at our interconnection locations. If you are an ISP with a substantial amount of Netflix traffic, review this information to learn more about the program.

The Netflix Open Connect program provides opportunities for ISP partners to improve their customers' Netflix user experience by localizing Netflix traffic and minimizing the delivery of traffic that is served over a transit provider.

There are two main components of the program, which are architected in partnership with ISPs to provide maximum benefit in each individual situation: embedded Open Connect Appliances and settlement-free interconnection (SFI).

Open Connect Appliances can be embedded in your ISP network. Embedded OCAs have the same capabilities as the OCAs that we use in our 60+ global data centers, and they are provided to qualifying ISP partners at no charge. Each embedded OCA deployment will offload a substantial amount of Netflix content traffic from peering or transport circuits. Multiple physical deployments can be distributed or clustered on a geographic or network basis to maximize local offload.

If you have substantial Netflix traffic destined to your ISP customers, deploying embedded OCAs is usually the most beneficial option. However, embedded OCAs are not always deployed, depending on your traffic levels, data center limitations, or other factors.

Netflix has the ability to interconnect at a number of global data center facilities and public Internet Exchange fabrics as listed on our Peering Locations page. We openly peer with any network at IXP locations where we are mutually present and we consider private interconnection as appropriate. If you are interested in interconnection, please review the information on the Peering Locations page.

ISPs who do not currently participate in public peering might want to consider that a single IX port can support multiple peering sessions, providing direct access to various content, cloud, and network providers. In addition to Netflix, many large organizations such as Akamai, Amazon, Facebook, and Google/YouTube widely participate in public peering and combine to deliver a substantial percentage of traffic to a typical ISP.

90f70e40cf