PXW-X70 + DSC Chart + Vectorscope

[Josh Bass]

I guess that begs the question. . .what kind of LEDs do you mean? I have a client who has a set of 1x1 Ikans that are bi color (have a pot to switch, in 100K increments, anywhere between 3200 and 5600) and dimmable. A couple of the higher end DPs in town really like them as well. We use them and white balance as we would with tungsten, no issues. See what I'm getting at? We don't really have different settings or methodologies for those LEDs vs tungsten.

[Paul Anderegg]

Cheap 50CRI eBay $30 LED. :-D

I have a separate thread seeking accurate 3200k LED lights,

I am much happier with the 3200k setting, the cheap bad CRI LED's skew the vectors all over as you would expect. The 3200-3300k scoped settings should be perfect for daylight as well, I just wanted to hash out exactly how the settings apply themselves before finalizing them. As long as I can get push to white to match the manual kelvin, I thin that is a success.

Paul

[David Dixon]

Originally Posted by Paul Anderegg

If anyone out there is sufficiently experienced with correcting color in FCP X, please drop me a private message. There is only so much that can be done with the in camera settings, but having the DSC chart means you can actually create a color correction preset that you can apply to EVERYTHING you shoot prior to grading. I would love some help manipulating the final vector that is always tilted. :-)

Paul

Although I do some simple grading in FCPX, this is almost certainly beyond my expertise. I assume you mean you want to use the Color Board to set things like you want and then save that as a Preset. But as for actually helping do the CB settings I'm not really qualified. My suggestion is if you don't get help here, post this over at fcp.co

[Craig Seeman]

Reading this thread. What I think you might need is actually Davinci Resolve 11 Auto Color Match feature. You'd shot the chart and you use the color match feature to properly calibrate. Now if only there were a way to send that info to FCPX.

Here's an example of how that works though.

Davinci Resolve 11 Color Match: Automate Color Correction - YouTube

[Craig Seeman]

BTW my own understand of these issues is that I suspect one shouldn't assume these problems are specific to the X70. My understanding is that different cameras have a bit different "color science" so they may be off in different ways. At issue is whether you can perfectly compensate.

[Paul Anderegg]

David, what I am looking for is to be able to correct IN CAMERA as best as possible, then provide those settings, as well as a sample 1080 4:2:2 video file with those settings shooting the DSC chart. Would be nice if someone could load said file into FCP X, Vegas Pro, Adobe etc, and provide color correction settings for the NLE that would bring any uncameracompensatable variables into line. The hope would be that the in camera color correction settings as well as the NLE companion settings would allow someone to bulk align all of their raw footage prior to grading or making other changes.

Any further suggested in camera settings will include a link to a ProRes LT 4:2:2 1080 30p video file of the 3300k (The Sony says my ID Light is 3300k) lit DSC Camette chart, so that users can have a calibrated scene to align to.

Paul

[Terence Morris]

The X-70 could potentially make a great second camera, provided I can cross-edit X-70 footage easily with that from a PMW-200. At least without too much hand wringing (or vectorscope twisting) for out-of-the-bag stuff. I was making the assumption that as part of the Sony EX series it would have the same overall "look" on canned settings. But is this true? Can anyone with a deeper insight of this pipe in?

[Paul Anderegg]

I was shooting a film festival right next to a shooter with an EX1. The colors were way different between his camera and my CX900 on stock settings. On playback at 10Mbps h264, he said my video looked better than his.....heh.

Paul

[Robert Young]

Originally Posted by Paul Anderegg

I highly recommend simply setting color mode to STILL and black level to +10, even if you do no other adjustments.

Good call on the PP setting.
I had already settled on STILL as the most pleasing color rendition, but looking at the footage in Premiere I could see that the blacks were really crushed.
With the black level at +10, it seems to open the blacks and extend the dynamic range a bit.
I used to agonize over the complex PP settings when I was shooting with an EX-1.
I'm hoping to keep it way more simple with the X70.
Outside of the WB issues, are you using PP "STILL, BL +10" for both daylight and low light shooting??

[Paul Anderegg]

What's daylight? I work the overnight shift, I still think the ND filter is to assist the camera in floating when dropped in the water. :-)

Seriously though, daylight W/B should be fine with the 3200k set-up, as the camera actually has a +/- W/B shift for 5600k outdoor preset. Don't know why they didn't include that option in the indoor preset. Once I get the camera dialed in for 3200k, I will see what works well in the day. Day is different than night. All the blacks and highlights and high gain really push certain elements of the picture. I loved Cinema mode on the CX900 in the day, but it was horrible indoors under LED lighting or at night.

Paul

[Alister Chapman]

Paul: One thing I found over the years doing picture profiles is that while a DSC chart is a great starting point, in practice trying to precisely match one can often be detrimental to the cameras overall colorimetry. There are several things to consider.

The DSC chart is using very narrow and precise colors. Your vector adjustments will only be targeting these individual narrow band hues and you may not be aware of what is going on with all the in-between hues.

No video camera has perfect uniform color response, especially bayer sensors which will have some large peaks and troughs in the color response. The relationship between how these peaks and troughs line up with the 6 primary vectors used by a DSC chart will greatly effect the overall colorimetry. Getting a perfect single hue may result in an overall color response that is skewed.

Often the more you try to precisely put a vector in the right box on the scope the more issues you will see in in between hues.

So, while I will often refer back to a DSC CDM chart for a sense of the overall matrix line up. I would never use one to create a picture profile where the overall colorimetry is more important than putting vectors in boxes. Also, the best pictures are rarely "perfect" pictures, often a little warmth or some other adjustment away from perfect will give a more pleasing image. One of my preferred ways of creating a picture profile is to use a photograph or picture of a real world scene. It's much easier to spot rogue colors in a "real" scene than by looking at a test chart.

Having said all of the above the last frame grabs you posted look pretty good with the exception of the blood which looks to be a very strange hue.

[Paul Anderegg]

OK, finished up the calibrations. Lit the chart with 55watt IDLight. Did a push to white, got 3400k, set manual kelvin to 3400k, and began adjusting. I am providing calibrated settings for both ITU709 MATRIX and STILL color modes. STILL was able to be corrected much more successfully than the ITU709 MATRIX. As promised, I am providing ProRes LT 4:2:2 1080 30p files of the charts shot with the settings listed below as well as stock ITU709 PP4. I am also providing pictures of the actual vectorscope monitor for each set of calibrations, as well as the stock ITU709 MATRIX PP4. These settings are good for setting manual kelvin, and only an ever so slight full matrix shift to red was noticeable when push to white was engaged at the same kelvin. The white blob still stays in the center, and all color vectors stay on the correct lines. For reference, this DCS chart is 50% saturation visually, so it looks dull, and requires saturation boost in post if you want to see the colors stretched to the boxes. These settings are as close to the boxes as I could get, I spent several hours adjusting everything a hundred times over. I adjusted RED on target first, then adjusted BLUE on target. Like I said previously, you only get two adjustments, but I feel pretty good about these calibrations. Magenta is slightly off, but look at the stock calibration and you can see that every color except magenta is off by at least a full shade.

The next post will include the vectorscope photos, and the ProRes files. All files will be in the following order: STOCK ITU709 PP4, CALIBRATED ITU709 MATRIX, CALIBRATED STILL COLOR MODE.


GOOD: ITU709 MATRIX CALIBRATION
Black Level +7
Gamma ITU709
Black Gamma not adjusted
Color Mode ITU709 MATRIX
Saturation 0
Color Phase -3
Color Depth not adjusted
Color Correction
Color Revision
Memory Selection 1&2
Memory 1 Color
Phase 28
Phase Range 31
Saturation 25
Memory 1 Revision
Phase +15
Saturation -14
Memory 2 Color
Phase 10
Phase Range 14
Saturation 0
Memory 2 Revision
Phase +8
Saturation -3
WB Shift not adjusted
Detail +3

BEST: STILL COLOR MODE CALIBRATION
Black Level +7
Gamma ITU709
Black Gamma not adjusted
Color Mode STILL
Saturation 0
Color Phase -1
Color Depth not adjusted
Color Correction
Color Revision
Memory Selection 1&2
Memory 1 Color
Phase 28
Phase Range 31
Saturation 24
Memory 1 Revision
Phase +8
Saturation -14
Memory 2 Color
Phase 10
Phase Range 11
Saturation 0
Memory 2 Revision
Phase +8
Saturation -3
WB Shift not adjusted
Detail +3

[Paul Anderegg]

STOCK PICTURE PROFILE 4 ITU709 MATRIX SETTINGS

VECTORSCOPE PICTURE

http://1drv.ms/1scS6Ub

ProRes DSC chart file

http://1drv.ms/1scNM7t

Select DOWNLOAD to access the raw ProRes file.

[Paul Anderegg]

CUSTOM CALIBRATED ITU709 MATRIX SETTINGS

VECTORSCOPE PICTURE

http://1drv.ms/1oQrWAK

ProRes DSC chart file

http://1drv.ms/1oQpj1O

Select DOWNLOAD to access the raw ProRes file.

[Paul Anderegg]

CUSTOM CALIBRATED STILL COLOR MODE SETTINGS

VECTORSCOPE PICTURE

http://1drv.ms/1scSylx

ProRes DSC chart file

http://1drv.ms/1scPaHj

Select DOWNLOAD to access the raw ProRes file.