Vectorscopes
The skin tone line does not exist
Disclaimer: I wrote this very late in the evening/night and did this for fun. I may be wrong on some things, but I would love to hear about it. You can easily talk to me about it on my twitter, specifically this tweet: [link]
Table of Contents
Introduction
If you are an editor, colour grading artist or have just looked at a couple of colour grading tutorials on YouTube, you have probably heard of a vectorscope and a very useful thing in that vectorscope called the skin tone line. This skin tone line is supposed to indicate the perfect skin tone that you want to achieve when colour correcting and grading your footage.
A vectorscope found in After Effects. The arrow points at the skin tone line.
This seemingly arbitrary line is even used by some as a strict rule that they want to follow as much as possible in order to get satisfying skin tones. But not much information is found about this line and some people have questions about how to properly use it. Taran Van Hemert, editor at Linus Media Group, asked on Twitter “Confirm something for me: The skin tone line is mostly relevant for illumination by sunlight, correct?” [1]
That made me curious and I wanted to figure it out. I originally thought that the line was relevant for any kind of neutral light and not specifically sunlight. So, it would be a quick google to find the answer and one tweet. Instead, I couldn’t find any proper source calling this a skin tone line. Skin tone was never mentioned. Instead this line had two names: the I axis/bar/line and the burst line [2] [3]. I will call it the I line. It’s brother, the Q axis/bar/line and also the burst line, will also become important later on.
What is the I line?
According to a manual of a physical vectorscope [4], the I line is used to help detect if the luminance and chrominance (colour) signals are in-sync in analog video. A colour bar image can be generated and shown through a vectorscope. If the signals are in-sync, everything will line up, but if they are not in-sync they will not. This allows you to make adjustments to the signal in order to correct the synchronisation, so that the colours are accurate.
Source. On the left the signals are in-sync. On the right they are not in-sync.
This is important, because in analog video and specifically NTSC encoded video, the signal that carries the chrominance (colour) is completely separated from the signal that carries the luminance (brightness) [5]. However, they still need to stay in-sync and we can’t always use a colour bar image. To achieve this, there is a specific signal called colorburst [6]. This signal also gets plotted on a vectorscope and will always fall on the two burst lines (the I and Q line). Because it will always be the same signal and should always show up in the same place on the vectorscope, it is very useful to help detect if the luminance and chrominance signals are out-of-sync.
But why do we have the I and Q lines and what do they refer to? Because NTSC needed to remain compatible with black-and-white televisions, it couldn’t use the RGB model that we all use nowadays. Instead, it needed to keep the luminance, called Y, and add in the chrominance. In order to represent the chrominance, they decided to use two values called I and Q. The colour space that this resulted in is called YIQ [7]. It is very similar to the more well known YUV colour space.
YIQ and the skin tone line
The I value represents the orange-blue range and the Q value represents the purple-green range. These ranges were chosen in order to take advantage of the human colour-response characteristics. Our eyes are more sensitive to changes in the orange-blue range than the purple-green range. This allows NTSC to allocate more of the chrominance signal to the I value and less to the Q value, which in turn produces a better looking image [7].
The oranges that we are really sensitive to, are roughly our skin tones. So, when the I value was chosen to represent the colour range that we were most sensitive to, the value happened to be roughly our skin tone hue. It most likely is not perfectly on the skin tone hue, but it is close enough to it for colour encoding. I can’t find any good sources that say whether or not this was purposefully the skin tone or just a coincidence, although most point to it being a coincidence.
Because back in the days of NTSC we needed a way to detect if the luminance and chrominance signals were in-sync, our vectorscopes got two lines to help with that and because of the need to optimise the chrominance signal, one of those lines ended up roughly indicating where the skin tones lie.
Can the I line function as the skin tone line?
According to the book Video Demystified: A Handbook for the Digital Engineer by Keith Jack [8], in a section about automatic skin tone correction: “It should be noted that the phase angle for skin tone varies between companies. Phase angles from 116° to 126° are used; however, using 123° (the +I axis) simplifies the processing.” The phase angle in this case could be seen as the hue but from a different starting point. This tells us that skin tones in different software packages can vary by at least 10° in the hue. That might sound not that much or a lot, but below is a GIF to visualise it.
Animated GIF showing 10° difference in hue.
The I line does sit in that 10° range, so it could function as a skin tone line. However, there is a wide range of hues that can also be good skin tones. That means that a skin tone line should only be used as a rough guideline.
Conclusion
I have not been able to find any source that can say that the I line was meant to indicate the skin tone. It was simply meant to indicate the colours that we are more sensitive to and to help sync up signals in analog video. The I and Q lines are a relic from the analog video days. The I line roughly indicates skin tones, but there is a very wide margin to that.
How should the skin tone line be used and is the skin tone line mostly relevant for illumination by sunlight? From what I have learned from this, I would say the following:
The I line/skin tone line is meant as a very rough guideline for skin illuminated by a neutral light, not necessarily sunlight, and captured on a properly whitebalanced camera. The I line/skin tone line is too rough of a guideline for different types of neutral lights to make a big difference. Where the skin tones should actually fall on the vectorscope, is a matter of personal taste. The I line/skin tone line does fall within the acceptable hues for skin tone and can be used as an easy tool to keep the skin tones consistent.
Some applications have abandoned the I and Q lines and simply show one line that they call the skin tone line, but the same still counts. It is merely a rough indication of the skin tone hue.
Sources
[1]: https://twitter.com/TaranVH/status/1195442987729162240
[2]: http://www.peachpit.com/articles/article.aspx?p=2161673&seqNum=3
[3]: http://www.valtechvideo.com/partneri/leader/5851V-1.pdf
[4]: http://www.tek.com/dl/25W_7247_1_rfs.pdf
[5]: https://en.wikipedia.org/wiki/NTSC#Color_encoding
[6]: https://en.wikipedia.org/wiki/Colorburst
[7]: https://en.wikipedia.org/wiki/YIQ
Yes, I know some of the sources are Wikipedia, they are indeed not proper sources, but Wikipedia is a great collection of information and a great way to find proper sources. It is very late right now and I want to go to bed, so I do not care right now.
