X70 Full PP Color Correction settings

[Paul Anderegg]

I spent 3 full hours on the vectorscope aligning all the PP options (except gamma) using my DSC Pocket Camette chart. These are BASIC COLOR CORRECTION values for the COLOR REVISION and base menu PHASE adjustments.

METHOD: I started by selecting a COLOR MODE. Gamma for was left at ITU709 for all adjustments. First adjustment made for each PP was PHASE. Phase was rotated until the majority of the primary colors were aligned as best as possible with their vectorscope boxes. If possible, blue, as well as it’s corresponding white balance line to yellow, were prioritized. Once basic phase was adjusted, the color correction menu was entered. The ONE PUSH SET function was used to target RED for color correction 1. Once the correct values for RED were set by this function, the next step was to move down to the color 1 revision, and adjust the phase (rotation) for the individual RED arm on the vectorscope. RED is the most important color to adjust first, as it is the primary color that makes up skin tone. After red was properly aligned, the next most offending and inaccurate color on the vectorscope was selected for adjustment. In most cases, this was the GREEN value, with the exception of Cinematone 2, which required aligning of the YELLOW vector, and STILL which required BLUE adjustment. The same process as was used on RED was used on GREEN, and YELLOW in the case of Cinematone 2.

These are BASIC color alignments, meant to bring the stock PP’s as close to accurate colorimitry as possible. The ONE PUSH SET values were sightly different for each COLOR MODE selected, as each color mode slightly varies j how it presents it’s color pallet. As reference, the Sony manual gives base values for each color, but these do not match the values that are obtained by pointing at the actual calibrated DSC chart. The DSC chart is meant to accurately place vectors in the boxes meant for their specific color, so I consider these values more reliable. If you wish to alter these PP color adjustments, it would be advisable to only modify the revision settings for phase. Be advised that adjusting saturation can and will alter phase. For instance, if you increase or reduce saturation for the green revision, it will rotate the vector (color) of green towards yellow or cyan, and without a live vectorscope picture, you will not be able to see how far off the mark your final colorimitry will be. There are two adjustments for saturation, the above reference is for the revision menu, the one in the primary correction menu will have less effect on the phase if adjusted.

Because the color correction menu only allows 2 colors to be altered, this means that there will be some compromises in colorimitry. BLUE will in most of these settings should be on the mark, without undue magenta/purple being introduced into your true blue pallet. YELLOWS in many instances will be slightly into the orange spectrum on the scope, but this is not something you should notice. MAGENTAs will in many instances be pulled towards red or blue, but again, this is not something you should notice. In a few instances, CYAN will be slightly towards green, and this will be noticeable in many instances such as florescent lighting, and very light blue dawn/dusk skies, as a green hue. Of all these settings, Cinematone 1 has the best ability to resist this tendency to green cast on cyan.

VALIDATION: Once you enter these color corrections into your X70, you can toggle them on and off in the color correction menu. This will allow you to point the camera at something colorful, and toggle the custom setting off and on so you can see if you agree with them. Due to the limitations in adjustability, certain shades of yellow may have a very slight Nerf/Highlighter/Neon yellow/green hue. This issue, as well as orange or slightly umbra reds can be adjusted for as necessary by the basic PHASE menu selection. I would love to get feedback from various shooting scenarios of how these settings work, and how you altered them to correct for colors or lighting situations you encounter. I would also appreciate any feedback and input on COLOR DEPTH adjustments, which I did not go into, as those would require extensive access to colorful daytime scenes, which I do not have the time to seek out on my graveyard shift. If you are able to nail down a COLOR DEPTH value that works for a certain color with these listed color correction adjustments in any and all scenes you try them on, I will add them to a master calibration list for future posting.

I apologize for the length of this post, I wanted to run through as much as possible in the hope that some of you would be able to go through the menu as you read this, and it would not seem as discombobulated as trying to ingest it off-hand. I don’t think it is possible to modify these basic color corrections any further, as I spent hours throttling each value up and down in an attempt to find any further improvement possible.

Paul

[Paul Anderegg]

Cinematone 1
color phase: +3
memory 1 color: 9, 4, 0
memory 1 revision: +8, 0
memory 2 color: 18, 7, 0
memory 2 revision: -15, 0

Cinematone 2
color phase: 0
memory 1 color: 10, 5, 0
memory 1 revision: +5, 0
memory 2 color: 14, 7, 0
memory 2 revision: -5, 0

Standard
color phase: -1
memory 1 color: 10, 5, 0
memory 1 revision: +8, 0
memory 2 color: 19, 7, 0
memory 2 revision: -6, 0

ITU709
color phase: -2
memory 1 color: 10, 4, 0
memory 1 revision: +8, 0
memory 2 color: 19, 7, 0
memory 2 revision: -14, +5

Pro
color phase: -1
memory 1 color: 10, 4, 0
memory 1 revision: +10, 0
memory 2 color: 19, 7, 0
memory 2 revision: -2, 0

Still
color phase: +2
memory 1 color: 9, 5, 0
memory 1 revision: +12, 0
memory 2 color: 30, 7, 0
memory 2 revision: -9, 0

[Christopher Young]

Paul I guess you are aware that these Color Correction setting changes will only hold when using a manual white balance and then leaving it at that balance. The process being to do your white balance and then use the CC settings to fine tune your vectors as you suggest to match the colors to your subject matter under the specific lighting conditions that the white balance was taken under. A practice that has been used for years under fixed studio lighting situations where nothing alters in lighting or CCU white balance.

Under varied shooting and lighting conditions as soon as you do another manual white balance those CC settings will not reflect the settings you stored as the 'Phase Range' and 'Phase' of the adjusted CC memories will change.

The problem here is many users of this little camera will use 'Auto White Balance.' A few posters on the forum say they run auto white even when shooting concerts etc. As soon as you run auto white balance the CC settings will constantly alter as the camera tries to constantly track white.

For those who want to understand the relationship between the white balance and CC interaction the manual gives a pretty accurate description on pages 57 & 58 of the camera's Operating Guide. One extract of which is shown below.

Chris Young
CYV Productions
Sydney

[Paul Anderegg]

The "color correction" settings should not be confused with white balance. Color correction affects colorimity, otherwise known as calibration. White balance merely shifts the entire color palate as set in any given direction until white/black/grey is centered on a vectorscope. Hue, or "rotation" of the color wheel (each color moves towards the next color in the rainbow), is not adjusted by white balancing. 3200k studio incandescent lighting is the standard by which camera calibrations are done, and they are good for 5600k daylight as well. This is why LED lights have a CRI index. You would NEVER want to calibrate a camera to a poor CRI LED or even florescent lighting, as they have spectrum color spikes that affect particular colors, and not the others. If you calibrate using florescent, and "corrected" the color for GREEN which is spiked, anything not shot under florescent lighting would have a tremendous green issue, which would throw off all you other colors.

Just to be clear here, WHITE BALNCE and COLORIMITY (CALIBRATION) are totally separate things. You can rest assured your push to white with any of the above settings will produce more accurate colors than the stock settings. Also, there is an option for OFF in the PP screen. When this is set, you are basically shooting on a stock CX900/AX100. :-)

Paul

PS: For anyone wanting to argue semantics, what I refer to things as is how the camera calls it, and how people setting up professional camera call stuff. :-P

[John Nantz]

Paul - you've become one hell'ava resource on the X70 cam! In fact, all your posts, along with comments by others, would be a good reason for someone to consider getting one. Tweaking the settings will make the difference between getting OK color vs Good color.

Speaking of CRI, a few days ago there was a post by Craig Seeman titled "Light weight portable "500" LED Panel" http://www.dvinfo.net/forum/photon-m...ml#post1871866 and CRI kept cropping up as an important factor. I thought I'd dig into how it is measured and found an informative article in wikipedia at Color rendering index - Wikipedia, the free encyclopedia. Bottom line: This doesn't seem to be as hard a number as one would think because it depends on ...... a lot of things. Not the least of which is what/which test is being used. Manufacturers don't indicate how they measure it, at least none that I've seen.

There almost seems to be "committee wars" over which things that should be considered to determine the number.

" ~ The CRI has been criticized for not always correlating well with subjective color rendering quality in practice, particularly for light sources with spiky emission spectra such as fluorescent lamps or white LEDs." Since tungsten light sources for general use have all but gone by the wayside, the more difficult to deal with light sources are becoming more prevalent, such as LEDs. We've got one building that we'll most likely be retrofitting the fluorsent bulb fixtures out with the new LED fixtures. Many big-box stores such as Macy's, for example, have already made the change.

For the future, it would be nice to somehow program a histogram for light balance that can deal with LED light spikes. Maybe some kind of feedback between the vector-scope in the editing application that can be saved in the camera. Wouldn't that be neat, especially for correction with one's own lighting equipment.

For now, though, I'll be taking the manufacturers CRI figures for their lights with a little grain of salt. After reading the article, it will depend a lot on how I think their light looks.

Trying to correct for color spikes in post is a lot of work and what has helped me is to use some "good light" to try and cover up the "bad light", f.e., sunlight over LED, or good LED over bad LED. This isn't Hollywood and I don't have unlimited time to seek perfection.

The good news is that Oren Arieli's post about his Aputure Amaran 672c LED light (post #6) shows that the LED technology has improved and the price/performance bar has been raised.

[Paul Anderegg]

That is a good read John. The primary thing to keep in mind with 3200k calibrations, is that they work with DAYLIGHT. You could never calibrate a camera using 3200k florescent lighting (do they come in 3200k?) then simply shift the WB kelvin to 5600k and expect realistic daylight colors.

In the night news I shoot, with the LED on the little X70, I am dealing with as many varied light sources as you can imagine. You've got your 3000k auto headlights and police spotlights, then you have white LED scene lighting such as fire truck floods, and newer police take-downs and spotlights. Street lights range from blue all the way to old school astronomy friendly sodium vapor 2300K, and some florescent. Mix this with the new alwaysonecolor active red and blue police LED lightbars, and it is a lighting mess! Most studios or professional shooters who have light kits with specific CRI shifts sometimes will have a scene file calibrated for their specific light kits. the other workaround often employed, and seen on clap-sticks, is a color bars chart. Not the most practical in news, run and gun, or when you don't have the time to have an assistant walk out into the shot at a distance to hold up a chart or white card for each shot.

A suggestion to fellow shooters. I actually run my PP's in PAIRS. I keep one PP1 calibrated with incandescent WB's 0'd, and then PP2 would be the same calibration, but with the WB settings LL or RB to compensate for my LED magenta shift. I have the same "sets" for multiple different calibrations. If I shoot on my LED PP, then anything past the lights flood will have a GREEN SHIFT. This is problematic as it turns the black sky and background blacks GREEN. Honestly, unless I shooting a close in SOT, I would rather correct close in highlights with a bit of green than correct the entire scene and shadows by shooting LED calibrated.

Paul

[Christopher Young]

That's what I love about democracy everyone can have their say. I have to disagree with what you say Paul that the white balance has no effect on the stored CC settings because in my opinion it does and it's measurable and the manual even states it as a fact.

Paul my comments are based on many years of experience in the broadcast industry from my work with many different types and brands of cameras in multi-cam studio shoots and multi-cam OBs. For twenty-five years I was a registered engineer responsible for the installation and TD'ing of hundreds of TV outside broadcasts, like fifty a year. Be it a night time national football game or a live daytime OB to Europe the camera alignments had to be spot on to meet EBU or SMPTE broadcast specs depending on the shoots. Part of the role as a technical director is to integrate all camera chains into the OB system then ensure the color alignment of all cameras met those broadcast specs for colorimetry and that all cameras matched. Then finally to oversea each broadcast went through with the minimum of technical and transmission problems. So I guess I know my way around test and measurement equipment by now. Well I would like to think I do :) In fact we still maintain a full six camera HD-SDI fly away OB kit when clients require it.

The general basis for color alignment of cameras today is based on what is referred to as the Daylight D series one of which is Illuminant D65 with a color temperature of 6504K. To fully understand this a bit better without this becoming a laboured technical post anyone interested can look at this PDF with particular reference to pages 57 thru 61 which gives one a basic understanding on the subject matter of 'colorimetry.'

http://web.itu.edu.tr/~pazarci/rtv/T...5W_14700_3.pdf

Sad to say as great as the little X70 is, I love mine, is that the camera is way short of having the degree of matrix adjustment that a full broadcast spec camera has. It lacks the full broadcast Matrix for colorimetry adjustments and has no broadcast standard Preset Matrix, many cameras have a number of Preset Matrix selections available for storage of settings. Add to that no User Matrix or Multi Matrix with their sophisticated sixteen axis modes of color adjustments. It will forever be a struggle to get an X70 to be 100% color accurate. In fact it isn't possible with the range of adjustment the camera has.

Point in case look at the attached pic which I just quickly created on the office floor with the X70 on the left and a three CCD XDCam on the right. I'm not saying theses cameras are matched by any means but check the chart and check the file folder color. Basically most of the colors on the chart are pretty closely replicated by both cameras but check the third chip down on the left in each case. This shade is very accurately recreated in the XDCam image on the right and the file folder on the XDCam image is almost identical to what the human eye sees on a broadcast reference CRT. The X70 does not have the degree of adjustment to correct for this 'purple hue' without affecting other hues. It can only represent this shade of purple as a shade of blue and this isn't the only shade in the purple spectrum it cannot reproduce correctly. I wish it could but we can't get there with the limited controls the camera has. Not that I am complaining because for its price the camera is quite amazing.

But sure as anything I couldn't use a camera like this on a job where these purple hues were required to be reproduced as accurately as possible. Be it someone's corporate colors, or probably worse still imagine what a bride would say if her bridesmaids were dressed in this colour purple and they all appeared to be dressed in blue. We have also seen this problem with the X70s and on a number of other cameras for that matter on stage shows with purple LED lighting. On many shoots the disadvantages of 100% color reproduction can be overlooked in the case of the X70 because of its advantages in size, mobility and speed of deployment.

Happy shooting!

Chris Young
CYV Productions
Sydney

[John Nantz]

Wow! That color chart picture left an impression!

The University of Washington (in Seattle) has the school colors of purple and gold so there are, naturally, a lot of purple colors around here. Everything from clothing, to cars (including my son-in-law's car), signage, and even to store fronts. To have the color rendition of something come out looking blue (University of Southern California) might not be so good. And yes, I know there are a lot of brides maids wearing purple, too.

Originally Posted by Christopher Young

Point in case look at the attached pic which I just quickly created on the office floor with the X70 on the left and a three CCD XDCam on the right. I'm not saying theses cameras are matched by any means but check the chart and check the file folder color. Basically most of the colors on the chart are pretty closely replicated by both cameras but check the third chip down on the left in each case. This shade is very accurately recreated in the XDCam image on the right and the file folder on the XDCam image is almost identical to what the human eye sees on a broadcast reference CRT. The X70 does not have the degree of adjustment to correct for this 'purple hue' without affecting other hues. It can only represent this shade of purple as a shade of blue and this isn't the only shade in the purple spectrum it cannot reproduce correctly. I wish it could but we can't get there with the limited controls the camera has. Not that I am complaining because for its price the camera is quite amazing.

But sure as anything I couldn't use a camera like this on a job where these purple hues were required to be reproduced as accurately as possible.

What really amazes me is how well the color charts look, left and right, and yet how much different the folder looks. Unbelievable!

Fortunately, (for the videographer) the Seattle Seahawks main color is blue!

Because I'm not exactly a fan of the color purple, I'm still interested in the X70 so this is good to know. At this rate though, getting close to Vegas, maybe Sony will have a model update. Also keeping an eye on the JVC HM170/200 coming out in February that is FCPX friendly.

As an aside, doing nighttime news coverage has got to be about the worst video situation there is with the contrast between light and dark. On top of that, by the time one arrives on scene with all the action there isn't the time to sit and fiddle with all the camera settings 'cause it's time to roll or miss the action. Fortunately for the X70, there aren't too many flashing purple lights!

[Paul Anderegg]

Chris, you input is appreciated and respected as a reliable and for the most part superior source. I totally agree, and am saddened by the lack of colorimetry adjustments in the X70. It's nice to be able to better it, but it can never be perfected. As for purple, my stations logo and news vehicles include a purple stripe, which I always point the camera at after aligning to see how blue it turns out. The X70, unless you wish to have all your true blues pull towards magenta, will almost always render the DSC blue (slightly purple in real life) true blue.

I have posted previous color correction settings, which were done similar to how you would typically expect to align a broadcast camera on the vectorscope with a chart. These provided certain benefits, but I could never realize a quality I was happy with, and this turned out to be due to how Sony designed the alignment corrections on this camera. This latest set was done with the end user, non chart/scope owning shooter in mind. Previously, I had simply chosen a color correction number based on where and how it moved a branch on the vectorscope, I used Sony's method and function built into the camera to point and push select a panel on the DSC chart. As the saturation levels for each arm on the vectorscope were randomly stretched, and you can only adjust for the two that you are correcting, I mostly left these unchanged. For those unfamiliar with how alignment settings work in most broadcast cameras, the basic idea is to point each prime color vector at it's box on the scope, then increase saturation until the tip is in the center of it's box on the scope. That may be an oversimplification, but that is the basic idea.

As for YUV and RGB, I ran into that (learning experience) when I switched to P2 on Sony Vegas. All my ingested footage was coming out ink black, and I discovered Vegas handled P2 as RGB, and I had to apply a YUV (studio levels conversion) filter to make things right. I am running into this again with Sony's Catalyst Browse, as XAVC footage transcode in the other direction, and my blacks are super grey.

Basically, I want this post to provide BASIS color correction for each color profile the camera has to offer. Think of this post as giving you the corner pieces to the jigsaw puzzle, and the replies as helping you put all the center pieces in the right places. These are much better STARTING values to attempt adjustments from than the stock values which are not balanced properly.

Paul

[Ron Evans]

I do not yet have a X70 but have NX5U, NX30U, FDR-AX1, CX700, XR500. I can attest that all have this issue with purple but to different degrees. I shoot theatre multicam with at least 4 of these cameras and set all to preset indoor WB and fine tune when I edit with Edius. With the mix of lighting from LED and tungsten or spots by scene setting a WB I found to be a waste of time. It is more important that the cameras do not continually change their WB. I have no problem this way getting the purple I want with the 3 way colour corrector. The NX5U does have matrix parameters but I have never ventured in that direction !!! If the X70 does not have what is needed then spend twice as much on the PXW-X160 /180 with the fine matrix controls needed.

I had a demo unit X70 a week ago and compared with the NX30 and NX5U. Very similar to operate and nice clean picture. Not quite as sensitive as the NX30 but really close. Lens seems to ramp more than the NX30 and with the lens difference starting point is the main difference in sensitivity I think. Framing the same scene the X70 showed F2.8 24db and the NX30 F1.7 18db. So close once lens and gain difference taken into account.

Ron Evans

[David Dixon]

I've been out of touch for over a week and am just now getting back to some testing.

I did some shooting over the holidays, using some flat settings I'm experimenting with. I did a simple white balance, but per Paul's earlier instructions rather than a full custom white balance I just did a read of a white card then manually set that Color Temp. However, I did not use the color settings Paul posted earlier in this thread - I'm going to try with those next. I'm pretty pleased with the results in general. Colors are good once I did some minor "grading" but as expected, the biggest problems are blues and purples. I'm having the opposite problem from Chris's example. Purples are OK, but blues all skew a bit purpleish.

I'll post some examples in the next day or so.

I'm finding the camera is not that great in low light - but it's saved by the relatively clean gain. I'm finding that +9 and even +12 are actually useable, unlike my previous Canon XF100. However, I really miss the waveform monitor on the XF100. The histogram is just not the same and I'm struggling with exposure. I am usually erring on the underexposure side.

[Paul Anderegg]

David, if you only adjust ONE setting by eyeball to wrangle your blues away from magenta (purple), try moving the main PHASE setting. And for white balance, I find that I can be a few hundred K off the mark in either direction and still get very acceptable results, although my lighting is all over the place. You can experiment with manually adjusting the kelvin (and phase/color corrections) by eyeball, but it is recommended you use the EVF for this, as the LCD panel has an ever so slight yellow/green hue to it. The EVF actually has a color correction setting and color panel chart. I played with it, but it always looked the same to me.

Paul

[Mike McKay]

It's fantastic to have you guys on here testing and discussing this situation with color balance. I think I'm still going to get one of these as the features/value for money seems so good. Without the camera in hand, reading these threads and the manual it's a little difficult wrapping my head around exactly how one should approach getting the best white balance possible.

I've been a 'push to balance' guy for years, then use my eye and a couple monitors and experience to tell me if I'm close enough.....then tweak in post. Are you guys saying that if you make picture profile adjustments then push to WB the camera will then alter those pp adjustments just made? If true, it sounds really fiddly to push and then go and change pp settings. Is it as bad as it sounds?

[Paul Anderegg]

Color Correction settings SHOULD NOT alter the way you white balance. They improve push to white as well as preset white balance settings.

Paul

[David Dixon]

Paul, here's a quote from you in another thread. Maybe I'm dense, but the following quote is either contradicting what you just posted here, or I don't fully understand either post. Could you clarify??? All I want to do is a manual white balance to a white card (is that the same thing as Push to White?) to save time in post, with the color corrections listed in this quote helping get a more accurate color matrix. I've already found that they help the problem I had last week with not enough differentiation between blues and purples. ---Thanks

Quote from Paul:
<<<I use push to white merely to attain the KELVIN of the light, I then quickly toggle into manual kelvin set, set to whatever push to white said the K was, and I am set to shoot. The PP WB adjustments should be sued to "center" your whiting after you select the correct kelvin for your lighting, but still see your whites a bit off-color. You can push to white 5 times and get the same color temperature reading, but vastly different hues and colors. When you dial the camera manually, you are getting solid and repeatable colors, like an actually broadcast camera. High end broadcast gear changes kelvin, and nothing else when you push to white.

Here are my current Cinematone 1 PP settings that I got off the scope and with a bit of "eyeball" calibration as well. You can try them out to see how you like them. NO color depth adjustments made on this. These were adjusted with the camera seeing and set to 3300k on a 3200k incandescent light source.

Color phase: +3
Memory 1: 12, 4, 0/ +13, +15
Memory 2: 13, 7, 0/ +7, 0

Try using those color revisions as a base, and simply point them at a colorful scene and ramp them up and down to see changes visually. It is sort of like an eyeglass test at a optometrist....better, or worse, better or worse. :-)>>>