Select Page

Calibrating the JVC HD100 with a DSC Chart

The following article was published on this site in 2006. This is part 1 of 3 of a series of articles about calibrating a camera with a chart. When I moved the site to WordPress the original article was unlinked and since then I received several requests for the original TrueColor configuration. Apparently there are still a lot of HD-100 out there. So, here it is, by popular demand, the original text with the TrueColor v.1.0 at the end. I will post the other two parts shortly.


 

Recently I bought a DSC Labs chart (http://www.dsclabs.com) to calibrate my HD100U. My goal was to achieve 1:1 calibration in order to use as much colorspace as possible. Basically I want to leave the creation of my look in post, working on the highest amount of valid bits possible. According to Scott Billups this can yield up to 20% more color resolution than factory settings. The chart that I used is the FrontBox 16:9 HD Pro. At ~$300.00 you might think that it’s a lot of money for a chart but I can tell you one thing: I’ve seen the light and I am a believer :). I would buy a Chroma DuMonde chart if I could afford it. The difference in image quality after using the DSC chart is dramatic. I would not shoot anything without adjusting the camera with it. The FrontBox series has a nice backfocus chart in the back, much better than the Siemens star found in the Fujinon manual.

I spent two and a half days tweaking the settings in the camera, testing the results and retesting several times. I believe I found the configuration that I was aiming for. This article describes the steps that I performed to calibrate the camera. If you find that I seriously missed something I will be very interested in hearing about it. My email is phciccone at gmail dot com. Please forgive the lack of an email link, the scumbags spamming the whole world made posting that kind of stuff very dangerous. Here is how it went. First I checked backfocus by using the backside of the chart. Then I placed the color chart in front of the camera and lit it with two photoflood lamps at equal distance. I checked the two sides of the chart with a light meter and verified that I was getting the same values. No other sources of light were used. I set the White Balance and verified that it was at 3200K (Fig.2). I connected the camera to a laptop running DVRack with the HD pack. This is to get a calibrated monitor with WFM and Vectorscope. The connection was directly via FireWire cable.

I framed the chart as accurately as possible and then opened the iris, per DSC instructions, until I got the white chip at 100 IRE on the WFM. Then I used the Master Black and Black settings to stretch the opposite side and get the black chip to register at 0 IRE (Fig. 3)
Finally I adjusted the Gamma value to give an even distribution of the steps across the chart (Fig. 3, again).

That was the easy part. The next step was to change the Color Matrix to place every color in their VectorScope squares. Keep in mind that boosting or reducing a color will “pull” part of the chart in other directions. For example, boosting the Red signal will affect the angle of the Blue channel, even without touching the Rotation dials. You have to change each value slowly, check for “side effects” and compensate if necessary.

 

According to the DSC documentation, their colors are set to bring each color signal at about 75% of the vectorscope. Using DVRack you should increase the scale until the Red and Yellow fall in their boxes. These are the skin color components and need to be adjusted first. After they were in their places I adjusted the values to get the desired effect (Fig. 4).

I took a clip to verify skin tones and I’m quite satisfied. To verify how my settings differed from others I then loaded the “Cinelike24pHD” configuration and the popular “CineWide” posted by Tim Dashwood. Tim’s file is derived from the “Cinelike” one so it inherits the same color matrix. In my opinion this kind of manipulation is better left to postproduction work with tools like “Magic Bullet.” The purpose of this calibration is to find the most accurate way to reproduce color-accurate images.

Given the limited latitude of video, compared to film, we want to preserve every bit of data. All shots have been taken with the exact same configuration. The aperture has been left the same, the lamps and camera have not been moved or turned off. Also, the samples were taken one immediately after the other.
Figures 3 to 6 show the readings from the WFM and VectorScope for each scene.

Finally, my settings are at the bottom of this page. Hope this helps you, there can be differences between cameras and these settings might not be optimal for your specific HD100. I encourage you to get a DSC chart and calibrate your camera. By shooting a few seconds of the chart in front of every shot it can also be used for color correction in post.

 

Factory Scene 1 Factory Cinelike24p Cinewide DSC-based

Here are snapshots taken from the clips. You can look at each snapshot and check the corresponding WFM and Vectorscope value from above. Click on any image to see a full resolution version (1280×720). One thing to keep in mind while looking at my settings is that I disabled the camera’s sharpening. That generated a softer image that looks slightly blurry compared to the other samples. Another reason to shoot back-to-back samples and compare them on the screen. Without this kind of objective comparison it’s pretty hard to determine the effect of a given feature.

 

Factory scene 1 Cinelike24pHD
Cinewide DSC-based

Lastly, since this is a digital format, I compared the histograms of frames taken from each sample. This gives us a easy-to-read chart that shows the latitude of each settings. Notice how the DSC-based histogram shows a very nice and even distribution.
To get the frame grabs I loaded the original .m2t files into VLC and used the snapshot feature to export a frame from each clip. Figure 13 shows all the histograms as presented by the PhotoShop’s Levels window.

Factory scene 1 Cinelike24pHD
/td>
Cinewide DSC-based

And here is the scene configuration:

Master black -2 Color Gain Normal
Black Normal Color Matrix Standard
Detail Min R Gain 3
White Clip 108% R Rotation 4
Knee Manual G Gain 2
Level 100% G Rotation Normal
Cinelike Off B Gain 3
Gamma Std B Rotation -3
Gamma level -4

I’m curious to know your considerations about all this. Please, if you have the time, verify my settings and let me know if you find anything wrong with them. The point of all this is to get to know our camera as much as possible 🙂

Ciao!