"TrueColor" Scene File Tests
 

I had the opportunity last weekend to try out Paolo Ciccone's excellent "TrueColor" scene file for the HD100. I had previously been testing a file I named "WIDELAT2," based on one of Tim Dashwood's scene files, and wanted to see how they compared. I also tried a third variation, which combined the "TrueColor" matrix settings with Tim's wide latitude settings.

The TrueColor variations certainly display much more vibrant and natural colour than the HD100's un-adjusted matrix. Highly recommended. But I also noticed that the third (combined) variation, while looking a slight bit more washed, also exhibited more detail in the shadows.

Paolo measured his latitude at about 7 stops, but I think that the combined setting may eke out just a little bit more.

Here are some comparisons for one particular example scene (I did about 10 scenes, which I will post on my site a bit later):

Original "WIDELAT2" settings:
http://ethur.stargate.ca/neopics/QEPark_07_widelat2.jpg

Paolo Ciccone's "TrueColor" scene file:
http://ethur.stargate.ca/neopics/QEPark_07_truclrpc.jpg

Combined "TrueColor" with "widelat" variations:
http://ethur.stargate.ca/neopics/QEPark_07_truclret.jpg

Crop showing difference in the shadows:
http://ethur.stargate.ca/neopics/QEP...r_compared.jpg

And for reference, here are the settings:

WIDELAT2
Master Black -1
Detail MIN
Black STRETCH3
White Clip 108%
Knee MANUAL
Level 80%
Cinelike CINELIKE
Color Matrix OFF
Adjust (all NORMAL)
Gamma STANDARD
Level NORMAL
Color Gain NORMAL

TRUCLRPC (Paolo Ciccone)
Master Black -2
Detail MIN
Black STRETCH1
White Clip 108%
Knee MANUAL
Level 90%
Cinelike OFF
Color Matrix STANDARD
Adjust
- R Gain 3
- R Rotation 4
- G Gain 2
- G Rotation NORMAL
- B Gain 3
- B Rotation -3
Gamma CINELIKE
Level -1
Color Gain NORMAL

TRUCLRET ("widelat" variation)
Master Black -1
Detail MIN
Black STRETCH3
White Clip 108%
Knee MANUAL
Level 80%
Cinelike OFF
Color Matrix STANDARD
Adjust
- R Gain 3
- R Rotation 4
- G Gain 2
- G Rotation NORMAL
- B Gain 3
- B Rotation -3
Gamma CINELIKE
Level NORMAL
Color Gain NORMAL

(By the way, if anyone read the above posting before 6:10pm tonight [April 30], I had some of the settings backwards. They've been corrected above.)

Earl, the links doesnt work!

Indeed. -Thrilled to see them, but the links don't work.

Please send me a screen shot of the exact error message you are getting. The links work fine for me, so I need to know more in order to fix it.

I'm getting a basic "page cannot be displayed" message. Very generic.

Hi again. My tech guys think they found the problem. Please try again and let me know if the links are working now. If not, I'll move them to a different server.

(I work for an ISP, BTW.)

Thanks for posting these, Earl. I was wondering how one scene might be rendered differently with different settings. I personally much prefer having the widest latitude possible, so I'm with you on the combination of the two. The native True color version seems to crush the blacks a touch (or at least doesn't see into them). I prefer to bring the blacks down in post if need be.

Very good study with positive results.

Great job Earl, thanks for posting this, I liked your combo (TRUE/WIDE) setting alot, best of both I have to say!

I like your variation, the original Truecolor looks a tad too saturated and contrasty, stretching the blacks more helped quite a bit.

Hi Earl.

It's 12:57am after a full day spent tweaking the Truecolor config on the set of 3 commercials and I have some more information about color matrix. This shoot gives us the oppurtunity to test the colors with good monitoring gear and the set has a lot of good saturated colors. Long story short, here is TrueColor v.3, for your testing pleasure ;)

Set Master Black to Normal (originally -2), no black stretch. This, IMHO, gives you more detail in the dark areas without affecting the midtones. If you really need to, use the lowest stretch but I wouldn't do it with controlloed lighting, the dramatic effect, without stretch, is stronger with good definition in the dark areas. Also, I had reports of people been able to affect those dark tones in post, a confirmation that TrueColor preserves the information even when, at first, it seems that parts of the picture are too dark.

Set Blue Gain to 1 (originally 3). The Blue gain will actually shift the red toward Magenta a little bit. It's a good combination if the camera seems to be too "red happy".
Dropping the blue to lower value will affetct the skin tones too much, a value of 1 or 2 seems to be a great adjustment to bring the reds a little more "under control" without sacrifice.
I'll be interested to know the Forum's opinions on this.

Thanks for the update Paolo (and all the initial hard work as well!) It almost seems counter-intuitive that lowering the blue would make the red go more magenta. You'd think it should be the other way around (i.e. more blue = more magenta). I'll have to give it a try.

By the way, "red happy" seems to be the only major concern from people using the "TrueColor" file. Since red gain is set to 3, what have you found happens/goes wrong when that is lowered to 1 or 2?

I though you were going to write something like
"if anyone read the above posting before 6:10pm tonight" You get a prize.

Nah, for that I would've hidden a secret message in a pixel.

Perhaps the name "True..." creates a naming problem.

I watch almost nothing but HD and it's clear that many productions, like the 2005 Grammies, have reds far more intense than anything seen in NTSC video. It looks very much like the HiVision I was seeing in Tokyo in the late `80's that I called "Hyper-Real."

So rather than modify the setting, I'd simply rename it as "HyperColor." It will be great at 60p!

I too wonder why one couldn't simply lower the R to 1 to get "TrueColor."

And, you are correct, lowering B should not move toward Mg but increase yellow.

Yes, I was too expecting the red gain to be directly responsible fot tha but after spending a few minutes with the red gain it was obvious that that was not going to yeld the expected result. The blue does it and I think that you'll be able to verify it easily. The reason why the red is at that level is that it places it at the right spot in the Vscope. Initially I gave more emphasis to correct position o f red and yellow as they are the major components of skintone. In fact pretty much everybody reacted postively to the skin tones of my configuration. Version 3 is an attempt to adjust it a little bit based on "real world" situations.

Gimme the prize! ;)

Can't claim I have come up with the name. It was part of the subject line that Tim D posted as a reference to my article. Then the name stuck. I'm not complaining :)

This post is really late in getting to the discussion but... have any of these tests been performed using a person as a subject to look at skin tones?

dynamic range issue
 

Hi i ve been using Paolo's TCV2 and now on TCv3 but i didn't set the color as he suggested, from my montior it's a bit too cold/bluish.

may be we separate the color from tone. as a matter of fact i never do color balance , but just set the either day light or tungsten. this is easier for me to make a plug in( i m using vegas) to correct the color in one shot.

my main concern is dynamic range as i realy want to creat a Digital "negative" to squeeze in more detail on both highlight and shadow.

both gurus are great , it will be even better if can work on a super dynamic raneg to go beyond 8 stops. i m from old school of Great Ansel Adams. so i guess 8 stops is just enough for me to work. -4 stops for shadows and +3 stop for high light.

does anyone try to use ultra contrast /digi-con filter to make acheive this?

JY

tested 3 scenes set up
 

Hi after i test all 3 , i tend to like "widlat2" becos the black seems to be smoother insetad of some darkess back is become very dark compare to deep gray
and i like the shadow area has more detail.

anyways this will give more room for em to do tone adjust at post.
color wise i think set as u like is OK, as i think different camera may be different.

JY