Help calibrating camera with DSC chart
 

I am addicted to making my Sony cameras look better by putting them on a vectorscope using my little $130 DSC Camette chart. I have gotten good results, so good in fact, I find myself having more vector adjustments than my chart has color panels!

Regardless, my main problem is lack of technical resources for understanding how to properly set exposure and camera base settings prior to calibrating. The vectorscope I use has "100", "75", and "3X" settings. DSC charts require 2x to get values properly in the box. I typically use the 3x settings, then switch back to "100" (I assume that's 1x?) and see the points around 1/3 of the way on line to the boxes. When using 3X, the test clips I shoot and load into FCP-X look about the same with saturation in NLE set to 100% and vectorscope set to 133%.

Now a few questions. With my ITU709 gammas, I typically turn off the knee, and try to place the black patches at just above 10IRE, and the white patches at 100IRE, as per the little paper that came with my chart indicates. It also gives steps for the 5 greyscale, and 50% grey wants to be at 60% or so. Is this correct? When I put the white at 100IRE, the resulting clip looks way too bright and washed out. Obviously, when I open or close the iris, the vector colors expand and contract accordingly, so setting proper exposure is very important.

As for not having a 2X setting, would increasing in camera matrix saturation 100% bring the 50% saturation levels of the DSC chart to proper levels on a 1x scope? I wouldn't leave the camera like that obviously, but any help getting things in the right spot would help.

If anyone knows some links to setting up cameras on a DSC chart, I would appreciate the assistance! I've been through everything on DSC's website, but require more details. I've heard that blacks should be set at 0IRE, and that whites in ITU709 are better set at 90IRE. These counter what the DSC paperwork states, so I am a bit confused. Below is a typical "calibration" I get (before/after) with my little chart. The reason I said I need a better chart is that all my Sony cameras have greens that sit half way to yellow, and require adjusting matrix values for colors that are not visible as scope boxes. The below example the camera (PXW-X180) identifies and adjusts YL- instead of green. Ugh. :)

Re: Help calibrating camera with DSC chart
 

Hi Paul-

I've consulted for DSC and helped design some charts, so I can help you out.

The key thing to understand is that these charts are designed with the actual Rec 709 spec in mind. This spec only calls for about 2.4 stops of overexposure latitude and three and half stops of underexposure latitude. Cameras have been way better than that for a long time, but the spec is quite old and once something like that is established it's not a good idea to change it. (This is partially because this spec wasn't official until recently even though it was devised in the 90s, and partially because every piece of equipment from the camera through post is set up to work within this spec.)

There are two kinds of charts. The ones where the gray background is set for a filmic 18% gray should be set up so the gray is at 40%. If you have the traditional Rec 709-spec chart the gray falls at 57.5%.

Beyond that, where white and black fall almost aren't relevant unless you are matching cameras. What the camera can hold as white is likely beyond the dynamic range seen in the chart, and the same with black. I'd focus more on putting gray in the right spot (either 40% for the dark gray or 57.5% for the light gray) and look at color from there.

I'd zero out everything in the camera and turn off knee before you tweak as knee circuits can cause colors to shift in highlights. Then look at color and play with the matrix. What kind of settings are available in the camera? You said something about the camera adjusting a YL control... I'm not familiar with that. The ones I know look like R-G, R-B, G-R,G-B, B-R, B-G.

Adjusting the matrix is always a tradeoff. The one thing you're missing on that chart is flesh tone, and it's the most important color you can look at. Adjusting green and red can throw off flesh tones so you always want to have a way of looking at that when adjusting the matrix. You may find that green is placed where it is because placing it somewhere else destroys flesh tone, and Sony has found that yellowish greens are more attractive than nasty flesh tone. These compromises are made in cameras all the time.

The thing to keep in mind is that the chart is correct. Cameras rarely show the charts perfectly, but that's the fault of the cameras. The charts meet the specs perfectly, as defined by the physics that define what the red, green and blue primaries are, but the cameras rarely do.

I'm not sure if increasing the saturation in the camera will help with the 2x zoom issue. It's better not to do that as color saturation changes the closer it gets to clipping. You can see hue shifts if you increase saturation too much. Give it a try and see what happens.

Also, don't worry so much about putting the colors in the boxes as the spec is VERY saturated and modern sensibilities call for "filmic" color that's much less saturated. Make sure the colors are on the vectors (line between center and their boxes) for accuracy but don't obsess about getting them in the boxes.

Also, get a flesh tone reference and work on making that perfect... then see what other colors you can make perfect without screwing up flesh tone. Red and cyan tend to be the hard ones to get right, but it may be less disturbing if a Coke can is a little orange than if flesh tone is a little green.

Let me know if you have any other questions.

Re: Help calibrating camera with DSC chart
 

Thank you for the response Art. I read somewhere that the RED vector should be the first and most important color to get accurate, for fleshtones. I have a non DSC chart with a lot of colors including multiple fleshtone patches, but I have always had very good luck with those the way I have been calibrating. The cheap Sony cameras seem to want to have yellow/green fleshtones, it is very irritating! I like when I vectorscope an interview clip, to see the little fleshtone line occupied by a vector. :-)

As for saturation, the scope from FCP-X below are set to 133% with a maximum (100 Apples) saturation boost, but still not in the boxes. The resulting video was VERY COLORFUL indeed! See below for a set of clips shot with those colors. I love them, but back to back with the cheap JVC's at my station, the look black and white by comparison. I am also pushing the limits of the capability of my X180's 1/3" chips to handle that level of color without getting some pretty noticeable video noise. I am retaining the settings, but doing a 15% matrix-wide saturation decrease. My concern currently is that I have adjusted 6 primary colors, but the other 12 adjustments available in my camera are unaltered. Worried I am pumping up certain colors (primaries) while leaving the in betweens at unmatching levels.

And YL meant Yellow. My camera has a matrix menu for those R-G, R-B etc, then a multi-matrix for each primary plus a + and - of each primary (Green+, Red -, etc) on each side of the vector box, so 18 in total for that secondary menu. I had poor results trying to use the R-G, R-B matrix, the multi-matrix was much easier.

Re: Help calibrating camera with DSC chart
 

Paul - you really scored in this post with getting Art to come by and comment!

To chime in with one of the areas Art hit on, was the skin tone and how it looks. Just a few hours ago I was looking at a video that was referenced in another thread "Big doubt - Canon x Nikon" by Rafael Lopes. In Post #9 by Chris Joy, there is a link to a camera comparison at https://www.youtube.com/watch?featur...&v=EsZtUZvoeO0

The reviewer, MichaelTheMentor, compared numerous aspects of the two cams but when he got to one part, how skin looks, [ @ 19m:13s ] he had an interesting approach. Rather than state his opinion of which is the best cam for the purpose he said to get out a pen and paper because he had a test, or quiz. With 12 sample pictures under various conditions, bottom line, it would be up to the viewer to determine which cam was best

The test shuffled the pictures and there was no indication of which cam took the picture, a very interesting approach, I think, to eliminate bias. The preverbal "beauty is in the eye of the beholder" approach.

Art - thanks very much for the informative post as that was a lot of interesting information.

Re: Help calibrating camera with DSC chart
 

Since there is a dearth of calibrating references online, I will take some time and compile my list of favorite links on the topic and place them in this post for future Google searches.

Anyone have an old used expired ChromaDuMonda chart they would be willing to sell cheap? :-)

Paul

Re: Help calibrating camera with DSC chart
 

Actually Paul, I have an expired Chroma DuMonde in excellent condition. If you're serious about wanting one, I can probably make you a good deal. But, it might have to wait until after NAB.

Re: Help calibrating camera with DSC chart
 

Perfect Chris! :-D

My Hamlet MicroFlex vectorscope has a "Choma DuMonde" screen, says that on it, and has all the extra boxes. In the past I paid $300-$500 to align my 3 chip broadcast cameras, simply due to the hours involved. Now I have the scope, and the time, plus the settings, just need the colors for my boxes. Most people don't appreciate or realize the difference calibrating a camera can make. Most expensive broadcast cameras are really good out of the box, but cheaper cameras are often intentionally "off" to avoid things like noticeable noise on certain colors, like blue. These cheaper cameras can have a night and day difference from a few key adjustments! The "stock" X180 scope on the left up top is a good example. That green was a shade of neon "lime" before being swung into the correct position.

Paul

Re: Help calibrating camera with DSC chart
 

Found a Tektronix link with some good information. I liked this part! :-O

"Today’s modern camera has advanced multi-matrix settings
and it is possible to line-up each primary to be in the
vectorscope boxes, but inadvertently reduce the overall
color gamut."

www.tek.com/dl/25W_21309_0.pdf

Re: Help calibrating camera with DSC chart
 

Mine is a standard (21 x 13) Chroma DuMonde 28R with the cavi-black option and resolution trumpets & bowties. It's going to be 10 years old soon, but it's in pristine condition and hasn't seen the sun in years (and has only rarely ever been outdoors). Still in the original ziplock with paperwork.

This chart is now known as the CDM 24+4R and sells for more than $1500 new with the cavi-black. I also have a standard (21 x 13) Multi-Burst SquareWave, same age, same condition, that sells for $587 new. I can let go of one or both for a fair but very generous price.

I need to take photos and get them posted in our classifieds board but I'm absolutely swamped right now with NAB prep.

Re: Help calibrating camera with DSC chart
 

>I read somewhere that the RED vector should be the first and most important color to get accurate, for fleshtones.

Not always the case. If you look at Canon cameras they sacrifice red in order to make flesh tones look better. Canon reds are orange, which helps blend red blemishes into flesh tone. I don't really like this approach but it's part of the Canon "look." It certainly doesn't work with every camera and has to do not only with the dyes they use on the sensor but the spectral filters on the OLPF package, the matrix setup and possibly a LUT.

I wouldn't intentionally do that on a camera that wasn't from Canon; just point out that if you try to place a Canon camera's red on that vector you're going to screw up flesh tone in a big way. Anything red on a face will shift toward magenta and look purple, like lips.

So... that's the long way of saying "It's not that simple." :)

>I have a non DSC chart with a lot of colors including multiple fleshtone patches, but I have always had very good luck with those the way I have been calibrating.

Hopefully not the Macbeth chart. :) If you don't have the money to go out and by a different chart (the next step up would probably be DSC's OneShot Plus, which does have flesh tone patches) cover over all the patches on the Macbeth that aren't flesh tone and put it next to the DSC chart, in the same light. That's not a perfect solution but it may work. (Not sure how accurate Macbeth colors are... it's a $70 chart printed on cardboard that's meant as a still photography reference from the 1970s.)

>The cheap Sony cameras seem to want to have yellow/green fleshtones, it is very irritating! I like when I vectorscope an interview clip, to see the little fleshtone line occupied by a vector. :-)

Is that before or after your tweaks? It looked like the default mode was for green to pull toward yellow, which may be their way of sacrificing green to get decent flesh tones. You're less likely to notice that tree leaves are a little yellow than flesh tone being a bit green.

>See below for a set of clips shot with those colors.

I get a message that says "This movie is private."

Once again, saturation isn't really in style now. I'd pull it back a bit. No camera ever puts all the colors in the boxes at the same time, and when they do they rarely look good. Nearly every camera has some sort of color issues and the more saturated the image the more those show up.

I wouldn't worry about the secondary adjustments too much. The primaries are the main ones to worry about. Secondary adjustments are just more math. At some point you'll stress the matrix (seen when the color dots smear like crazy) and that starts compromising color on its own.

>I am also pushing the limits of the capability of my X180's 1/3" chips to handle that level of color without getting some pretty noticeable video noise.

That's the other issue with matrix adjustments: they increase noise. All the color channels overlap somewhat, which is where secondary colors come from. The only way to make a primary pure is to subtract the other color channels from it (which is why R-B, for example, literally means "red minus blue.") A certain amount of this is already happening just to create basic color science, but the user matrix allows you to do more. The problem is that subtracting one channel from another increases noise, because subtracting a signal from another signal also means subtracting random noise from that signal. Essentially, you're leaving holes in the second signal due to noise in the first signal... which increases noise.

Extreme matrix adjustments are never good. Those controls are meant for finessing.

>I had poor results trying to use the R-G, R-B matrix, the multi-matrix was much easier.

Yeah, I'm not sure how those work--they're set up a bit different. Alister Chapman has a great video showing how standard matrix settings work. If I recall correctly, using G-R as an example, increasing that value makes green more saturation but shifts red's hue toward green. Going the other direction decreases green's saturation and shifts red the other way. The first letter changes only in saturation, the second in hue.

Re: Help calibrating camera with DSC chart
 

If you're going to be at NAB DSC Labs will often sell charts at a discount on the last day so they don't have to ship them back to Canada.

In the past I've worked their booth but I'm going to be at Sound Devices this year showing off their new PIX-E line of recorders. Adam Wilt will be at DSC Labs, though.

Re: Help calibrating camera with DSC chart
 

Yup. They cheat on colors that they hope you won't notice in order to preserve the ones that you do. Yellow or green flesh tone is pretty obvious, but blueish or yellowish grass or orangeish reds often skate by. They'll compromise on the latter to get the former.

Re: Help calibrating camera with DSC chart
 

Tektronics has a ton of great information on their website. It's aimed toward the broadcast market but still relevant to us.

DSC Labs make the best charts in the world but their expertise is in making charts, not teaching people how to use them. They're a little behind the curve when it comes to how cameras are being used outside of broadcast. Still, they're awfully nice people and answer questions as best they can.

That's how Adam and I came to work with them: we can explain the stuff they can't and make it relevant to the digital cinema crowd.

Re: Help calibrating camera with DSC chart
 

They have expiration dates on them but they tend to last a lot longer than they should. DSC has a new program where, for a subscription rate, you get a new chart every few years at a significant discount and they send you color chips to check and see if your chart is off or not.

I know a lot of people who put a label with their name on it across the expiration date so no one notices. Not ideal, but... I've never seen a DSC chart shift a crazy amount. The shifts do happen, and they happen with every chart, but DSC is honest enough to say so.

Re: Help calibrating camera with DSC chart
 

If I'm not mistaken, it's sunlight and intense studio lighting that can damage the color chips over time... my impression was that charts stored in the dark tend to last a lot longer. Hope that's an accurate assumption. Mine stay in a portfolio stored in the closet.

I've always wondered, though, if there's any off-gassing degradation involved from the ziplocks that I keep mine in all the time.