The ColorChecker Video Workflow – Premiere, FCPX and Resolve

     

by Dave Mobbs – July 12th, 2016

By Coloratti Ollie Kenchington

OllieKenchington

On-Shoot Video Workflow

The X-Rite Colorchecker Video is an essential part of my video workflow and kit and it goes out with me on every shoot. This article should help you get the best out of these charts and, thereby, drastically improve the quality of your rushes.

ColorCheckerVideoWorkflow1

The first thing to get right is the placement of the chart. If you are shooting an interview, for example, do not move the chart right over to the camera, as the lighting that is falling on your subject is what should be measured, not the lighting near your camera. The other thing to bear in mind is that you want the glossy black strip at the bottom of the chart to be free of any reflections, as these will artificially lighten the black strip and render it useless when gauging contrast in post. Finally, it is a really good idea to leave the scene while the chart is being recorded. This could be by having your interviewee or actor hold the chart, by propping it up next to whatever product you might be filming or by attaching it to a stand so it can be freestanding in your scene anytime you need it. By leaving the frame while you record it, you guarantee that the light hitting the chart is not being altered or obstructed in any way by yourself, which would not be desirable.

Once your chart is placed properly, the next thing to do is use the spectrally flat grey side of your chart. Because it reflects all frequencies of light that hit it equally, without absorbing any, you can get a precise colour temperature reading from it, unlike if you simply pointed it at a “white” wall or piece of paper.

ColorCheckerVideoWorkflow2

If your camera doesn’t support spot white balancing, then you will have to bring the camera in closer to the chart, rather than the other way round. Alternatively, if you are using an external monitor that supports RGB Waveform scopes, then you can leave the camera where it is, find the part of your traces that relate to your chart and then manually adjust the white balance in camera until the red, green and blue channels overlap precisely.

ColorCheckerVideoWorkflow3

Once your white balance is sorted, flip your chart over and set your exposure correctly. If you are using a standard BT 709 gamma, then simply set your cameras zebras to 90% and adjust your exposure until you see the familiar zebra pattern appear on the upper most, 90% reflectance, white strip. This should leave you a little extra room for bright highlights and get you pretty close to a correctly exposed image.

ColorCheckerVideoWorkflow4

If your camera is set to apply a log gamma curve, then you will need to know the corresponding diffuse white level to set your zebras to. S-Log 2 is 59%, S-Log 3 is 61%, C-Log is 62% and C-Log 2 is 58%. The reason these values are so much lower is because these gamma curves are designed to compress highlight information, that would normally be clipped, into the same size “container”. This throws up an interesting side point; if you shoot a scene in log, exposing the 90% reflectance strip as above, but then find yourself in post, pulling your gain up to place this same strip at 100 IRE (because it was the brightest thing in your scene), then you shouldn’t have shot that scene in log in the first place! Log gamma curves are optimised for dynamic range, often at the expense of overall image quality. If you do not have anything in your scene that is much brighter than the 90% reflectance strip on your chart, then use a standard gamma, as you will not lose any dynamic range and have a much cleaner and less compressed image.

So those tips should help you record images that are as close in camera to neutral as possible. That doesn’t mean you’re not going to have to do any work in post, it just means you’ll have a lot less to do and the less you do to your rushes in post, the cleaner they’ll look and the less artefacts you’ll introduce.

Editing Phase

DaVinci Resolve 12.5

Below is an example of using DaVinci Resolve 12.5’s Color Match feature, in which you line up a chart overlay with your actual chart in the footage and Resolve will analyse the differences in values and correct for them. As you can see below, because I used the chart effectively during the shoot, there is only a very minor difference between the recorded values and the correct values. One thing I have found with this feature is that, although it is very good at shifting the hue and saturation around to line it up correctly, the way it affects the highlights is usually undesirable. For this reason, you may prefer, like me, to do a manual balance from your chart.

ColorCheckerVideoDavinciresolve

To begin my manual adjustment, I use a power window to isolate the four contrast strips on the chart. With ‘highlight’ mode enabled, it will effectively hide everything else in your scopes apart from the chips inside the mask. My 90% reflectance reference chip is more or less where I want it to be, though it needs to be raised slightly. In S-Log 2, middle grey drops to 32%, to make more headroom for the compressed highlights but I can now bring that up a bit, to between 40% and 45% IRE, so my skin tones are properly exposed. Finally, the glossy black is elevated so I will bring that down to between 0% and 3%. This is, don’t forget, a slightly arbitrary contrast adjustment, simply to get it in the realms of neutral. Once I have a neutral image, I can then start to apply secondary adjustments to push in whatever look I’m going for.

DaVinci1

DaVinci2

Now that my contrast is set properly, I can turn my attention to the colour. Swap your Waveform for the Vectorscope and move your power window so that it now masks everything but the 50% saturation colour chips. These chips correspond exactly with the boxes on the Vectorscope, when it is set to 2 X zoom. By adding controls points along the Hue vs Hue curves (there are shortcut buttons along the bottom to do this on each of the primary and secondary hues), you can then push and pull the hues around until they line up with their corresponding boxes.

DaVinci3

DaVinci4

Next, you can do the same again in the Hue vs Sat curves, this time moving the traces outwards, towards their boxes. It is almost always the case that you will find magenta, cyan and green a long way from their boxes. This is down to the quality with which your camera de-bayers the raw data in camera and subsequently infers magenta and cyan by processing the opponent colours in the signal. If you have a good, strong codec or have shot raw, you can go ahead and add saturation to these hues until they fall in their boxes. If you have shot in to a weaker codec, then push them only as much as the data can bear to be stretched. A good indicator of this is when the traces on the Vectorscope begin to appear thin and less cohesive. It doesn’t matter so much if the saturation isn’t spot on; it’s far more important that the hues line up, so get that right if nothing else.

DaVinci5

DaVinci6

If everything has gone to plan up until now, I should find everything in my image is looking good and well balanced. A great acid test for your image is skin tones, so as a final measure, I will just double check that my skin tones match up with the skin tone chips on the chart. To do this, I add a new node (Alt-S) and then use the HSL Qualifier to sample my skin. Because I am still in highlight mode, only my skin tones should appear in the Vectorscope. You’ll know if everything is good if two things happen – firstly the traces should mainly fall along the flesh tone line, or slightly to the right of it (particularly if you sample your subjects lips too) and secondly, the skin tone chips on the chart should be the only other thing visible on screen. If they are, then you know your skin tones are within the same narrow band as the charts skin tone references which, incidentally, will be the same no matter the race of your subject.

DaVinci7

So there you have it, a few simple steps in production, followed by another few simple steps in post and you’ve got yourself a perfectly balanced image.

Final Cut Pro

Below is an example of using Final Cut Pro X’s Color Correction effect to adjust the colour balance of your footage, keeping an eye on the RGB Parade to make sure the grey balance side of the chart appears in the exact same position in each of my three colour channels. If one were lower/higher than another, I’d know there was a colour cast in my image.

ColorChecker VideoFCPX

Setting contrast is also easy. In the Waveform, I can see that my 90% reflectance reference chip is more or less where I want it to be, though it needs to be raised slightly. In S-Log 2, middle grey drops to 32%, to make more headroom for the compressed highlights but I can now bring that up a bit, to between 40% and 45% IRE, so my skin tones are properly exposed. Finally, the glossy black is elevated so I will bring that down to between 0% and 3%.

ColorChecker VideoFCPX2

Now that my contrast is set properly, I can turn my attention to the colour. Swap your Waveform for the Vectorscope and you can check your overall saturation levels. You are looking to add saturation until the saturated colour chips on the chart fall at roughly 50%.

ColorChecker VideoFCPX3

If everything has gone to plan up until now, I should find everything in my image is looking good and well balanced. A great acid test for your image is skin tones, so as a final measure, I will just double check that my skin tones match up with the skin tone chips on the chart. To do this, I add a new Color Correction and then use a Color Mask to sample my skin. You’ll know if everything is good if, when you click the View Masks button, both your subject’s skin AND the skin tone reference chips on the chart go white. If they do, then you know your skin tones are within the same narrow band as the charts skin tone references which, incidentally, will be the same no matter the race of your subject.

ColorChecker VideoFCPX4

So there you have it, a few simple steps in production, followed by another few simple steps in post and you’ve got yourself a perfectly balanced image.

Adobe Premiere

Below is an example of using Premiere Pro’s Lumetri Color effect to adjust the temperature and tint of your footage, keeping an eye on the RGB Parade to make sure the grey balance side of the chart appears in the exact same position in each of my three colour channels. If one were lower/higher than another, I’d know there was a colour cast in my image.

ColorCheckerVideoPremiere

Setting contrast is also easy. In the Waveform, I can see that my 90% reflectance reference chip is more or less where I want it to be, though it needs to be raised slightly. In S-Log 2, middle grey drops to 32%, to make more headroom for the compressed highlights but I can now bring that up a bit, to between 40% and 45% IRE, so my skin tones are properly exposed. Finally, the glossy black is elevated so I will bring that down to between 0% and 3%. You can also tweak the Tone controls at this stage to make sure you are happy with the overall spread of brightness across your image, which is particularly important if you have shot using a logarithmic gamma curve.

ColorCheckerVideoPremiere2

Now that my contrast is set properly, I can turn my attention to the colour. Swap your Waveform for the Vectorscope and you can check your overall saturation levels. You are looking to add saturation until the saturated colour chips on the chart fall at roughly 50%.

ColorCheckerVideoPremiere3

There are 3rd Party software companies that have created plug-ins that natively incorporate the ColorChecker Video targets for quicker results, however this workflow has been designed to demonstrate the workflow without the use of plug-ins to give you that choice.

Plug-ins
3D LUT Creator – For Adobe Premiere
Color Finale – For Final Cut Pro

 

About Ollie Kenchington

Ollie set up Korro Limited, a UK based film production company, in 2008. Korro produces commercials, short films, documentaries and promotional content for major global brands. Between projects, Ollie also hosts film production and post-production masterclasses for Google at their flagship film studio, YouTube Space London, as well as acting as an industry ambassador, event speaker and blogger for the likes of X-Rite in Europe and Creative Video Productions in the UK.
Prior to setting up Korro, Ollie worked as a certified master trainer for Apple, via their Authorised Training Centre programme in the UK. This followed three years working his way up the ladder within Apple’s Premium Reseller initiative, consulting for various organisations on pro-apps deployment within post-production studios and the use of video analysis tools in sports and athletics.
As well as Apple and Google, throughout his career, Ollie has had the pleasure of working with Gary Barlow, Disney, Warner Bros, BBC, Aardman, Dyson, Rolls Royce, Toyota, Nissan, Peugeot, Barclays, Bank of England, American Express and many others.

www.korro.co.uk

To view the main video demonstrating the ColorChecker Video workflow in all 3 software applications then please view here –

Or to view Ollie’s recent webinar discussing “Achieving Perfect Exposure and Profiles in Video” click here

 

Categories: Uncategorized | Tags:

One response to “The ColorChecker Video Workflow – Premiere, FCPX and Resolve”

  1. This is an awesome post thank you. I’ve been trying to figure out how to get the ‘auto match’ ability in premiere and you supplied the info for 3D LUT Creator which I had no idea about. FINALLY I can get that DaVinci resolve ability in premiere. I’ve tested both Resolve, Premiere, and FCX and they each handle the compression of the color spaces of cameras differently after tweaking. Resolve tends to blend the colors and mux them heavily, FCX blends them smoothly but will use the same color tonality for doubledup pixels almost defeating the 4:2:2 usefulness, but the only one that doesnt ‘hurt’ the image in the end as much as Resolve or FCX isn’t just premiere…. (because the lumetri color panel does a terrible job even worse than Resolve since it uses an invisible ‘adjustment layer’ rather than manipulating the actual color information in the file itself) but Premiere with Synthetic Aperture’s Color Finesse. Now it doesn’t come standard in Premiere (it does for AFE) but the pro version for premiere is the ONLY program I’ve found that uses the actual color information to make your adjustments when grading vs using an invisible adjustment layer. this is beneficial and matters because when you get down to 1:1 and 2:1 you can’t tell that you manipulated the image vs the other programs where you CAN tell it’s been altered introducing blocking, noise, and other weird color artifacts occasionally. Plus it doesn’t make the footage look ‘mushy’ like Resolve. So FINALLY I can use Finesse in conjunction with 3D LUT Creator to save me even more time in Premiere to instantly get the white balance perfect. Thank you so much and I’m so stoked to pick up the lut creator. I know it sounds like a bunch of bs, but I’ve tested multiple file outputs, at the same compression levels, the same uncompressed levels, and the same file itself and the results are always the same. Premiere with Finesse just seems to produce the best-looking results in the finer detailed areas.

Leave a Reply