Samsung SUHD Calibration
By Chris Heinonen on
On April 2nd in New York, Samsung held a press event to celebrate the release of their SUHD TVs. While these have actually been shipping for a bit now, some of the other models were not officially announced yet and the prices were unknown. Far more interesting to me, and many of the reviewers on-hand, was a special event Samsung held before this. We got an in-depth class on what changes are present in HDR and wide gamut content, and how our calibration and testing procedures might need to change.
Led by Kevin Miller and David Mackenzie, this nearly three hour session went in-depth on updated light targets, gamma curves, and other issues that we need to be aware of when calibrating TVs this year. This probably would have bored the general audience at the later event, but for those of us that calibrate TVs all year long it is very interesting. Typically new TVs have not required major changes in the way that they are calibrated. Since HDTV has come out, we have had the same basic targets for a proper calibration: Rec.709 gamut, around 35 foot Lamberts of light, and a gamma of 2.2-2.4 or BT.1886. Many of these have to change with the addition of High Dynamic Range content and the larger color gamut on new displays.
1000 cd/m2 of light. Usually when you calibrate a TV, you want a light output at the peak of around 35 foot-Lamberts, which is 120 cd/m2. With High Dynamic Range displays, we now are going to have highlights that can hit 1,000 cd/m2, or almost 300 foot-Lamberts in brightness. Not only can current TVs not hit these targets, but your eyes would really hurt if you wanted something that bright for more than a few seconds.
So why can we do this with HDR content? First, we now have 10-bits of data to work with instead of 8-bits. Where previously our range of levels for White was from 16-235. 10-bits gives us over four times the range for white, so a value of 235 that we consider pure white now can still be a reasonable foot-Lambert level, a highlight can be closer to 1,000 and much brighter. For non-HDR content we still want to target 35-40 ftL of brightness, but we can now have very bright highlights that look more realistic.
New Gamma Curves. Since we now have a larger range of values for white, our current gamma standards will not work well with HDR displays. If we gave an HDR display a gamma of 2.4 or even BT.1886, the main image would be very washed out. Instead we now have a new gamma standard, SMPTE 2084, for HDR content. The curve for this rises much slower than a traditional gamma in order to account for those highlights at the top. Of course you can’t use this curve with non-HDR content, or everything will be very, very dim because of how slowly it ramps up compared to 2.2 or 2.4. Since there is a standard for this, hopefully most, if not all, HDR displays will ship with it as an option.
Larger Gamuts. Along with HDR, most of the new high-end displays are supporting the DCI/P3 color gamut. This larger color gamut corresponds to what you see at a movie theater, and produces shades of colors that were not possible in the home before. In an example that was once given to me, Lightning McQueen in Pixar’s Cars is a completely different color on toys compared to your TV, because the HDTV color gamut doesn’t contain the color of red that he is meant to be. Now that shade of red, which you would have seen at the movie theater, is possible for the home. Another great thing is that since movies have been using the format for years in the cinema, there is a large amount of content ready to be available at home. Watching clips of films side-by-side, one with the P3 gamut and one with the HDTV gamut, it is easy to see the difference. Colors pop off the screen but are still exactly what the director wants you to see.
The one downside here for some calibrators is that getting these colors requires using Quantum Dots (or Nano Crystals, or whatever a company wants to call them). These Quantum Dots produce a much different spectrum of light than a traditional LED or UHP light source, and so a colorimeter can usually not read the light correctly. If you have a spectrometer, like the i1Pro2 I use or a better model, then you can read the light and get accurate results. For calibrators, or home users, that only have a colorimeter it is going to be hard to impossible to calibrate one of these correctly. It also is going to be a reason or excuse for some of us to upgrade our equipment.
New Test Patterns. Another recommendation that came out of this session is one that I’ve already been doing on reviews: Using APL patterns instead of windows or full field. APL, or Average Picture Level, patterns produce the same level of light output from a display no matter what test pattern is up. Previously many calibrators would use small windows on a plasma and full field on an LCD because of the differences. Now with local dimming, HDR, and OLED technologies out there, APL patterns around 18% provide the best results for calibration. They best mimic real-world content in terms of light output, and that is important when trying to calibrate. Many pattern generators support APL, like the previously reviewed DVDO AVLabTPG, but for those that don’t have this you might need to upgrade.
Samsung also provided us with a suite of test patterns designed by Florian Friedrich. Some of these patterns are combinations of multiple tests onto a single frame, making it easy to use. With a single screen you can check color, tint, sharpness, brightness, contrast, and more. It also shows the image processing that displays will do, and the problems that it can cause. In one case, a TV that displayed a 1×1 pixel checkerboard and 2×2 pixel checkerboard pattern correctly failed on a 3×3 pixel pattern. Being able to see this all from a single screen is great, and he provided some fantastic test clips shot in 6K resolution and scaled to UHD to show off what a display can do.
Now that these displays are out, and hopefully coming for review soon, there will be some differences in the data you see. As a reviewer, I now will have to calibrate each display twice: Once for HDR content, and once for standard HDTV content. Otherwise one will always look off. Hopefully these TVs will be smart enough to automatically switch between an HDR mode and a non-HDR mode when they detect HDR content, but we have to wait for that to appear first.
That said, when you watch HDR and P3 content, it looks fantastic. You get colors you didn’t see before, a larger dynamic range that is closer to real life, and it really makes a huge impact. Unlike the higher resolution, it’s a difference you can see on any size screen, and from any distance. I really can’t wait for these displays to arrive, along with content, because it is the first thing that makes me want to upgrade from my reference plasma.
Finally, I couldn’t resist this photo. At their press event with all the members of the media, where they want to show off the absolute best that their new $6,500 reference TV can do, what does Samsung do? They use Amazon HDMI cables just like I do and every other reviewer I know.