HD100 New Calibration with ChromaDuMonde DSC chart

[Paolo Ciccone]

Stephan Ahonen: Yes, the chip in the middle is the cavity black and it's supposed to be a 100%. The procedure to calibrate the camera via the DSC chart is to get the black at 100% after you expose the chart to get the whites at 100%. In this way you get the maximum latitude of the camera. The midpoint can be adjusted via the gamma and that is based on the capability of the camera and personal preference. For example, I just did the TC3 for the Canon XHA1 and its gamma adjustment is nowhere as fine as the one on the HD100. And that explains why the HD100 is about $1200 more :).

TC3 moves the gamma a bit so that the midpoint is not exactly at 50%. I feel that that works well for cinematic work. A couple of notes about TC3. First, it's not a video format. This means that the latitude provided exceeds the limits of NTSC. This is by design and it's A Good Thing(tm). Commercials and high-end drama shows are shot in film even though they are targeted for TV. This is because film gives the shooters the best colors and latitude. The footage is then converted to "legal" TV values in post. TC3 is meant, in the limits of the HD100, for the same use.

Second, I saw comments, from time to time, about the blacks being crushed. That's not the case. It is true that TC3 has a high contrast and gives you blacks that are at 100%, something that is not always common in the world of video. There is a simple test that you can do. Shoot a scene with TC3 and be sure to have dark areas in the picture, for example and dark suite or dress. Assuming that the scene was well lit and exposed, bring the footage into your NLE. In FCP, what I know and use, drag the footage in the timeline, apply "Color Corrector 3 way", double click on the clip and click on the CC3W tab. Click on the black level arrow 2 or 3 times and check the image on a good monitor. If you see more detail showing in the dark areas then you did not crush the blacks. If there is no data then the color corrector cannot do anything. But, as it always happens to me, if you see more detail coming up in the dark, detail that you could not see before in the LCD panel or on your computer monitor, then the camera correctly recorded the grades of black.
And it could not be different because in clibrating the HD100 with the chart I verified all the shades of black being present so if you loose detail in the black it's because of insufficient light or wrong exposure.
Regarding the Knee function, I would not pay much attention to it with the HD100. When I calibrated the Sony F350 I played with its knee adjustment a bit and I was able to see the diference from, say, 100% to 85% knee by naked eye. That camera can pick up detail in the highlights like I never saw in anothe camer at that level. Encourage by that I switched to the HD10 and tried the same test on the same scene. The result was minimal. Keep the knee at 85% but don't expect much from it. As confirmation of this I checked the impact of the knee by using the WFM and the difference is very small. I'm curious to see if it's any better with the HD200/250.

Hope this helps.

[Marc Colemont]

Hah! He's back Mr. Paolo. Tonight I will load your V1 settings to compare our camera's with your vectorscope readouts. Because I compared your V1 vectorscope apparently with the V3 loaded in my HD100 (Tim brought it to my attention).
Mean while I was in touch with a technical guy at JVC, who said that the camera's do have a calibration in factory on the DSP side (matrix @ normal). So the camera's should perform pretty equal on settings done afterwards in the Matrix.

[Paolo Ciccone]

Marc, why are you going to use V1? V3 is better and I would stay away from the early version. I did more testing with V3 and it's more accurate.

[Marc Colemont]

Hi Paolo, Do you have vectorscope screenshots available of V3 loaded? Because the one I find on your website seems to be V1, correct? That's why I wanted to load and compare the V1 vectorscope screenshots tonight to have a cross reference.

[Paolo Ciccone]

Originally Posted by Marc Colemont

Hi Paolo, Do you have vectorscope screenshots available of V3 loaded? Because the one I find on your website seems to be V1, correct? That's why I wanted to load and compare the V1 vectorscope screenshots tonight to have a cross reference.

I don't but I might get them later, I'm just swamped with "El Papel" but I plan on updating the site asap.

[Teodor Miljevic]

I think that Marc is on to something about reducing the reds. I've shot several projec's usind Paulos settings (V3) and I find it to be in a class of its own compared to the factory settings. The only problem I have is skintone which i feel gets an artificial tone especially in condition that wander away from situations with 5500K (either varmer or colder conditions). If some of you users that are more versed in the art of calibrating the hd100 could dive into this issue. I would gladly loose some lattitude if i could get some better skintones (since it takes more time to correct them then adjusting lattitude when I shoot(I almost always shoot in situation where I have ways to modify the lighting situation).
It would be interesting if someone could compare the different recipies looking at the skintones on a calibrated monitor. There is something "wrong" with the skintones. At first I thought it was the red channel that maxxed out but this wasn't the case. When I have highlights in caucasian skin it tends to emphasize them in unnatural way. When I check them in a waveform monitor on my NLE neither of the channels clip som I'm lost in whjat the problem is. Does anyone else share my problem?

[Paolo Ciccone]

Teodor, you might want to double check if your settings are correct because one of the most frequent comments that I get is that the skin tones are dead-on. I shot more than 20 interviews with people of all ethnicities and the skin tones with V3 were just perfect.
Latitude and color matrix are not conected so you can alter the level of colors without loosing range.

[Teodor Miljevic]

I will check my camera but I´m fairly sure that they are correct. Maybe my camera varies a little from the one that you used for your calibration. If this is the case what values in the colormatrix would you recommend adjusting(to enhance skintone)? The feeling I have about the colors is the same I had tovards a negative I fixed in photoshop for a professional photografer. I needed to go in and rotate every colors hue separately. I can not mention some kind of special shift because my camera seemes to emphasize all shifts (either to orange or to blue) I have never had this much problems correcting skintones from any camera (and I've used about everything under 10k). Most other cameras I have used handle whitebalance differences in an easier manner. does anyone have this experience or am I alone?

[Paolo Ciccone]

Teodor, cannot give you any suggestion about ajusting the camera. That has to be done when looking at the image on a calibrated monitor.

[Michael Wiegand]

Originally Posted by Paolo Ciccone

Stephan Ahonen: Yes, the chip in the middle is the cavity black and it's supposed to be a 100%.

Hi folks, unless I'm really missing the boat here - the chart is a CDM FrontBox. Frontbox charts are not available with the CaviBlack option, therefore the center chip is not a cavity and does have reflectance.

While it is a very good black (especially compared to old matte charts), it is not "zero black" or "superblack". That being said, we tend to find that there's only about a .5 to 1 IRE difference between the Caviblack & the chart black.

Michael
DSC Labs

[Paolo Ciccone]

Michael, thank you for the correction. Yes, I got the CaviBlack confused but, as you pointed out, the black chip is very close to the non-reflectance version and, as per directions with the chart, it was set to 0%.

Thanks again.

[Marc Colemont]

Hi Paolo and others.
Interesting stuff :-) I loaded the True Color V1 settings, and compared them to the DVrack screenshots from Paolo. On my camera the Reds were again out of the boxes. It does show my camera likes the reds more from start.
With the color matrix is perfectly adjustable. It shows the color charts surely do have their purpose to calibrate cameras.
I added the screenshot to the original posted link.

[Chad Terpstra]

Greetings fellow image engineers. I'm going to throw in some of my experiences with my newly-acquired "Camette" chart.

First off, to those of you frustrated by having your colors still "off" using Paulo's TC3, I really recommend you get one of these. Don't let the girly name scare you, the Camette is still very accurate and will help you better understand what your camera is doing. Also they're half the price of the larger charts, have a great white balance on the back (trust me, it makes a big difference; I used to WB to anything that looked white but this gives oh so much better results) and they're the perfect size for use in the field (which has already saved my butt). -Worth every penny to me.

To Marc I have to ask what you're white-balancing to. I mention it because I noticed your first parade (the one with just white) was zeroed at neutral but the second parade (with the grayscale steps) showed the red channel lower than green & blue. That's opposite what you'd expect from your problem but does make you wonder if something is going on. I've gotten into the habit of stopping down so white is middle gray before balancing (using auto iris). Seems most reliable to me. Also I guess it goes without saying that you have to use something engineered to be pure white.

Having said all that, it is probably the case that each camera has it's own quirks and should be calibrated by the user. However when I brought up my own camera on the chart and went from scratch, I found that I could not get all the colors where they needed to go even after hours of work and head-scratching. The matrix on this camera is like a circular Rubik's cube. Move one color up and the three across from it move too. Going back to TC3 got me closer than my own efforts.

So I wanted to ask Paulo if he could offer any advice as it seems that he managed to fit all the colors in the right places. For me the problem is Green and Cyan. Red is opposite Cyan and they effect each other equally so there's no way to pull one without the other. I've attached a picture to show you where I'm at, but bear with me as I'm a Mac person with no access to DVRack. I had to improvise a target system using Final Cut Pro's live scopes and tape on the monitor where the targets should be (based on 2x gain as required by the Camette). This is a blurry pic taken with my phone. Can you think of any way to pull out the green and cyan without messing everything up? Any tricks I should know about?

Apart from that I'd suggest that others frustrated with skin-tone problems try using Standard gamma (-2) as opposed to CineLike. I found CineLike to be too high in contrast for low-key or dark scenes and sometimes makes red mud of dark skin-tones. I know the issue of latitude is a hot debate but that's just my personal taste and could be helpful to other readers looking for a place to experiment. Just be sure to reset your light meter ASA to match the new gamma as it shifts things a bit.

Thanks!

[Paolo Ciccone]

Originally Posted by Chad Terpstra

I used to WB to anything that looked white but this gives oh so much better results) and they're the perfect size for use in the field (which has already saved my butt). -Worth every penny to me.

Chad,
you are making great point here, WB is often overlooked. I usually WB on the GS of the DSC chart so if you use TC3 and WB to something not as accurate your colors will be off. Sometimes this is a desired effect.

So I wanted to ask Paulo if he could offer any advice as it seems that he managed to fit all the colors in the right places.

Hard to give you advice. Since we have 6 Vscope points and just 3 adjusters, R,G & B, the effect of moving one has side affects on the opposite color. That is what made the HD200 so hard to configure. The original color matrix is so skewed that the HD200 could not be aligned as precisely as I wanted to the target. In fact I consider my TC3 for the HD100 far more accurate than the version that I made for the HD200. In talking with other people, some of them true experts of color calibration, my findings were confirmed. The HD200 stock matrix shows as a elongated hexagon in the Vscope. As you noticed, the influence of one dial is two-fold. In fact you will notice that if you raise the gain of, say, Red, the Blue will raise too and after a given point, it will start to rotate as well. The way I was able to get TC3 so close to the target has been via isolation and observation. You work with only one dial and keep turning it until you burn in your memory the effect of it on the whole system. Gain is, IMHO, the harder to master because rotation is fairly predictable.

Start with TC3 and the pick one color and try to bring it closer to your target. Don't try to fix the red by simply moving the red. The system is a like a group of rubber bands. Put some tension on one side and then release some on another side. The combined effect should result in moving the color to the target. Small increments and huge amount of time are the only secrets ;)

BTW, has you rightly pointed out, two cameras of the same model can have different reactions to the same settings. I know that a few hundred dollars on a chart seem a lot but if you are bothered by the color rendition of your camera than your eyes are good enough to appreciate the advantages of precise adjustments. At the professional level the cost of the chart is recovered pretty much with one job and, IMHO, can be the one thing that will bring you more work.

[John Vincent]

Guess you're saying we shouldn't WB off of a t-shirt, huh?

:)

Any thing available cheaply (say form board or even paper) than is true white?

My experience is that unless it's expensive (like the camera), it will eventuallu be lost, dropped, destroyed, or gotten dirty on a set.

Thanks....

john

evilgeniusentertainment.com