C300: Cowpunk's Profiles

cowpunk52

Veteran
These CP’s are designed on a Canon C300 with firmware version 1.0.7.1.00. As of 02/7/13, this is the most recent firmware available only through Canon Service. It seems to change colorimetry of the camera slightly to more closely match that of the C100; as such, these profiles should work similarly on the C100 as they do on the C300. If you have older firmware installed on your C300, you will likely get unexpected and unintended results.

These CP’s are designed to work best in daylight white balance. I’ll be updating this sheet with tungsten versions ASAP. These profiles are a work in progress, and may change at any time. If I do update them, I’ll make an announcement on Twitter. Feedback is always welcome and appreciated!

Download C_PICT zip file for C300's with Daylight Balanced CP's here: http://www.mediafire.com/download.php?pm7gc3rzqt9kwoh
Download C_PICT zip file for C300's with Tungsten Balanced CP's here: http://www.mediafire.com/download.php?v7uglkrripd9vf4

***A note on the Tungsten C_PICT Folder***
In adapting the profiles for tungsten light, I found the V.Weed and C.Weed files were the only ones that needed adjustment. L.Weed did not have a noticeable color shift in tungsten light, so that file is good to go for daylight or tungsten light shooting. The file has been renamed WEED_log since it doesn't need a color temperature designation and it is still included in the tungsten download folder, but it is the exact same file as the L.WEED_5 profile in the daylight download folder; because, well, I like to make things as confusing as possible apparently ;) For those of you manually entering the settings into your C100, the only difference between the day and tungsten CP's is that Phase is set to +14 on the tungsten CP's.

There is also a bonus CP in the tungsten folder, called EOSnutrl, and it's an attempt to match the Neutral setting of Canon DSLR's. It's not exact, as there are some colors that just don't want to match, but it's close. Having a correct WB is paramount to getting the most similar color possible.


V.Weed_5: Profile with video gamma, natural saturated colors, & small amount of sharpening. Suitable for delivery without color grading.

Gamma: Normal 3
Knee:
Enable On
Slope +2
Point 90
Saturation 0
Sharpness:
Level -6
Color Matrix:
Select Normal 4
Gain +30
Phase +3 (+14 for Tungsten)
R-G -20
R-B +11
G-R -23
G-B -14
B-R -10
B-G +1
Color Correction:
Select Area Area A
A Setting:
Phase 15
Chroma 31
Area 15
Y Level 31
A Revision:
Level 0
Phase -1

C.Weed_5: Profile with cine gamma suitable for a filmlook, slightly muted colors & smooth, soft skin tones. Handles color correction nicely, but has a great look without grading

Gamma: Cine 1
Low Key Satur.
Enable On
Level -10
Sharpness:
Level -10
Colo Matrix:
Select Normal 4
Gain +8
Phase +3 (+14 for tungsten)
R-G -20
R-B +11
G-R -23
G-B -14
B-R -10
B-G +1
Color Correction:
Select Area Area A
A Setting:
Phase 15
Chroma 31
Area 15
Y Level 31
A Revision:
Level 0
Phase -1

WEED_Log, aka L.Weed_5: Profile for achieving maximum dynamic range from sensor while avoiding color skews of the installed CINEMA preset, such as the greenish yellows and blue-ish greens. The profile has an accurate color profile, muted color saturation, and flattering skin tones.

***IMPORTANT NOTE ON MANUALLY INSTALLING LOG CUSTOM PROFILES***
When installing LOG custom profiles, start with a COPY of the original CINEMA profile. Unprotect that CP, then make the custom changes below. The gamma of Canon Log is slightly different when starting from a standard or default CP.

Gamma: Canon Log
Black:
Master Ped -4
Low Key Satur.
Enable On
Level -10
Sharpness:
Level -10
Color Matrix:
Select Canon Log
Gain +8
Phase +18
R-G -11
R-B +6
G-R -9
G-B +16
B-R -1
B-G -6

EOSnutrl: A profile for emulating the look of the Neutral setting on Canon DSLR’s. Proper WB is very important for getting the most similar look possible because the cameras react slightly differently to WB variation.

Gamma: EOS Std
Sharpness:
Level -10
Color Matrix:
Select EOS Std
Gain +41
Phase +16
R-G -32
R-B +4
G-R -21
G-B -10
B-R -3
B-G -15


Log Matrix Based CP's:


For this batch, particularly, I wanted to base a set of custom profiles on the slightly remapped Canon Log profile that the new firmware gives the C300. In my opinion, the Canon Log color matrix gives the prettiest colors and best looking skin tone available on the C-series cameras, and these profiles are meant to take advantage of that. At the same time, there are definitely some inaccuracies in the color representation of certain primaries, and I tried to compensate for that as much as possible in these CP's as well.


Download here: LINK

The profiles are described below:

709LOG - video gamma profile that uses adjusted Canon Log color matrix. Looks quite similar to shooting Log + View Assist, but without the increased dynamic range that the log gamma provides.

cineLOG - Similar to above but with the cine1 gamma profile; more muted colors and flatter curve with a bit more dynamic range.

CINEMA.2 - A subtly adjusted CINEMA profile that corrects some of the color skews inherent in the native profile. Yellows are less green, greens are less cyan. Skin is just very slightly warmer. Takes full advantage of Canon's log gamma dynamic range, and looks great in View Assist and with Canon's LUT applied.

709LOG
Gamma: Normal 3
Black / Master Pedestal: -2
Knee / Enable: On
Slope: -2
Point: 90
Sharpness: -10
Color Matrix / Select: Canon Log
Gain: +30
Phase: +11
R-G: +3
R-B: +1
G-R: -8 or -6
G-B: -2
B-R: +3
B-G: -1

cineLOG
Same as above, but
Gamma: Cine 1
Black / Master Pedestal: 0
Color Matrix / Gain: 0


CINEMA.2 (make sure you create adjustment on an unprotected copy of the original CINEMA profile)
Same as above, but
Gamma: Canon Log
Black / Master Pedestal: -3
Color Matrix / Gain: 0


Link to online Google doc w/ same info: https://docs.google.com/document/d/1hU4LNgTUjqw6hPR9UvlDzmo7CRUDjWISGqLj4-EVdtw/edit?usp=sharing
 
Last edited:
Brian, did you get the Excel spreadsheet I sent you?

BTW, thanks for the advice about the Canon Log picture style.
 
Good work Brian! Has Canon admitted that the color science changed with the new firmware, or is it you that had to do all the detective work? I'm also intrigued by the need to do different CP's for 3200K and 5600K. I would have thought that, assuming a proper white balance, the matrix offsets from the preamp gain levels would be very close, but clearly Art Adams found it necessary to do as well.
 
Good work Brian! Has Canon admitted that the color science changed with the new firmware, or is it you that had to do all the detective work? I'm also intrigued by the need to do different CP's for 3200K and 5600K. I would have thought that, assuming a proper white balance, the matrix offsets from the preamp gain levels would be very close, but clearly Art Adams found it necessary to do as well.

I called Canon and spoke with a few tech guys, but no one knew much about exactly what the new firmware does. They thought my experiences were interesting, and promised to send the report up the chain to see what's happening. We'll see... but don't hold your breathe for any official word. Other than that, it was my own sleuthing - which is why I'm still reticent to make the claim with absolute certainty, but all the evidence I've gathered so far supports the theory.

The need to do different CP's isn't an absolutely necessary one - but the camera does see red a little differently in tungsten light. Mostly it just looks a little more orange. While most people probably wouldn't notice, I think it still is a worthy idea to compensate for the differences since it won't be terribly hard.
 
Last edited:
Brian, I just shot a 5600K scene with my wife as a model and various CP's, modified Art's Alexa709 by changing Matrix B-R from 0 to 7 for less magenta in the reds. Of all the CP's, including Alistair Chapman's, another rental house's various CP's, my own EOS color(EOS Std. with Cine gamma) and saturation down to -15, I like your C.Weed_5 best. Definitely works for WYSIWYG. I think I'll try your CP on the next shoot I do with our C300! Thank you.
 
What's the deal with the temp profiles as far as white balancing goes? Should I leave my WB hard set at 3200/5500 or can I fudge a little? Such as a manual WB that suggests 5700k for example?
 
Glad you're enjoying the images, Jeff! Gary, there's no hard and fast rule regarding white balance, but I would probably use the daylight version on anything above 4000k and then a tungsten version when shooting below that. Speaking of which, I'll have the tungsten versions finalized very soon. They're just slightly adjusted variations.
 
The original post has been updated with tungsten versions, and also a bonus CP - EOSnutrl, which is my attempt to emulate the Neutral setting of Canon DSLR's. Have fun!
 
i'm wondering, if you did an equivalent of your fabulous TruLog Profile?
(It was Log Gamma and like Normal 4 Matrix ?)

2 weeks ago we used it again on a rental C300 with the new firmware installed, while i wasn't aware of possible colorimetry changes due to the new firmware.
It looked great as always. At least as i remember.

Would you recommend still using it? Or better your WEED_Log instead (which has the log Matrix)?

thanx so far!
 
i'm wondering, if you did an equivalent of your fabulous TruLog Profile?
(It was Log Gamma and like Normal 4 Matrix ?)

2 weeks ago we used it again on a rental C300 with the new firmware installed, while i wasn't aware of possible colorimetry changes due to the new firmware.
It looked great as always. At least as i remember.

Would you recommend still using it? Or better your WEED_Log instead (which has the log Matrix)?

thanx so far!

Definitely use what looks best to your eye, but I think the TruLog profile may skew reds slightly orange and skin goes a tiny bit green with new firmware. When I tested it on the new firmware, TruLog and TruCine (built on same adjustments) didn't look bad, but just different from what I remember setting it at. What I think I'm noticing with the new firmware is that color is not so much completely changed, but a little subdued and subtly different. This likely has the effect of causing color adjustments made to compensate for older colorimetry issues to be a bit overdone on the newer FW. But again, I'm just speculating here, and trying to figure out what's going on!

I sent my C300 out on a week long rental today, or I would post some comparisons of the TruLog and WEED_Log CP's. Either way, I think that WEED_Log is going to have a little bit nicer skin tones.
 
Hey Brian
I had a chance to try out your newest settings with the C300 with 1.07 firmware.
The profiles all look great. I'm noticing that on red objects there's a slight shift towards orange. Do you notice that too?

I wonder if its just a canon tendency to not capture red as naturally as the other main colors. I guess it could also be my lens (24-70 L and 24-105 L).
 
Last edited:
Definitely use what looks best to your eye, but I think the TruLog profile may skew reds slightly orange and skin goes a tiny bit green with new firmware. When I tested it on the new firmware, TruLog and TruCine (built on same adjustments) didn't look bad, but just different from what I remember setting it at. What I think I'm noticing with the new firmware is that color is not so much completely changed, but a little subdued and subtly different. This likely has the effect of causing color adjustments made to compensate for older colorimetry issues to be a bit overdone on the newer FW. But again, I'm just speculating here, and trying to figure out what's going on!

I sent my C300 out on a week long rental today, or I would post some comparisons of the TruLog and WEED_Log CP's. Either way, I think that WEED_Log is going to have a little bit nicer skin tones.

With the old firmware, i had the feeling, the Normal-Matrix was superior to Log-Matrix (not only skin tone) wise. In combination with Log-Gamma and your adjustments it was imo the best picture one could get out of the C300. (I did some adjustments on your TruLog CP as well).

Damn, looks like i have to start for myself testing again.
 
I wonder if its just a canon tendency to not capture red as naturally as the other main colors.

Interesting that you bring this up - I had a brief discussion with Art Adams via twitter about the same subject yesterday. Here's the transcript:

"Art Adams ‏@artadams
@briancweed Why my C300/Alexa settings don't work: http://ow.ly/hzCMj Red in Canon's CFA isn't red but orange-red, with green crossover.


Brian C. Weed ‏@BrianCWeed
@artadams That makes sense; but what's the real-world benefit to the application?


Art Adams ‏@artadams
@briancweed The good news is flesh tones always look great, and blemishes that have red in them never stand out, because you can't make red.
@briancweed The bad news is that if you ever need to shoot flesh tone and a Coke can at the same time, you can't get Coke w/out secondaries.


Brian C. Weed ‏@BrianCWeed
@artadams And what happens when we try to compensate for this in color matrix settings?


Art Adams ‏@artadams
@briancweed You can't. As soon as you pull that "red" value onto red you push flesh tones into red, and "red" objects become blue.
@briancweed The color space of the sensor is truncated as it can't reach the Rec 709 primary. It falls short on the red/green line.
@briancweed Also, green is pulled toward blue b/c they are subtracting a LOT of the green signal from red to make red as pure as possible.
@briancweed The red dye isn't pure enough so you can't ever get there. Any real reds must be created in post artificially.
@briancweed You can't access colors outside the color space defined by the primaries. Matrix only works inside space defined by primaries...
@briancweed ...because all the matrix can do is add/subtract primaries from each other to make pure color.
@briancweed Bottom line: people will look great, colors will look pretty, but the only really accurate color is blue and red doesn't exist."

Unlike practically every other subject we shoot, everyone has a very concrete notion of what human skin looks like, and I think that arguably the toughest skin color to reproduce accurately and pleasingly is the complicated coloring of Caucasian skin. After all, skin has a broad range of colors within it, and natural skin tone prevails with pail colors. In painter's terms, it's made up of a combination of pigments like burnt sienna, burnt umber, cadmium red, permanent rose, Ultramarine blue and Pthalo blue.

As Art said in a follow-up article to his Alexa Custom Profile (written before the discovery outlined above): "I played around with the C300 matrices yesterday during lunch on a shoot and was reminded just how hard it is to get both red and flesh tone dead-on accurate at the same time. Also, Canon cameras really emphasize red, probably because they put flesh tone ahead of all else, and there may be a very good reason why they tend to cheat red a little toward orange in most of their matrices: If your camera's colorimetry really punches up red, it's always better to err on the side of orange-red than blue red so as to avoid compromising flesh tone as well as dulling bright reds. To my eye, anyway, I find orange-biased reds less offensive than blue-biased reds, and I suspect Canon feels the same." source: http://provideocoalition.com/aadams/story/back-to-the-drawing-board

It's clear to me that Canon puts skin tone above all else in it's color matrices, and having true red in it's CFA might make skin tones less pleasing and more red/blue than the soft warmth that the camera maker is known for. I have a feeling this may be the secret sauce that makes 'Canon color' so pleasing. At the cost, of course, of 100% true red.
 
Last edited:
Unlike practically every other subject we shoot, everyone has a very concrete notion of what human skin looks like, and I think that arguably the toughest skin color to reproduce accurately and pleasingly is the complicated coloring of Caucasian skin. After all, skin has a broad range of colors within it, and natural skin tone prevails with pail colors. In painter's terms, it's made up of a combination of pigments like burnt sienna, burnt umber, cadmium red, permanent rose, Ultramarine blue and Pthalo blue.

I'd have a difficult time painting skin variety with those pigments alone :D Just the way Canon does, the reds would need a tiny dash of yellow to get the slight orange bias to mix convincing caucasian hues...
 
I'd have a difficult time painting skin variety with those pigments alone :D Just the way Canon does, the reds would need a tiny dash of yellow to get the slight orange bias to mix convincing caucasian hues...

i heard from an oil painter who taught students in china that he had to get used to painting east asian skin tones because they contain much fewer cool hues than caucasian ones.

and on that note, when you guys say that canon's colorimetry is optimized for accurate caucasian skin tones, what does this do to non-caucasian skin tones?
 
i heard from an oil painter who taught students in china that he had to get used to painting east asian skin tones because they contain much fewer cool hues than caucasian ones.

and on that note, when you guys say that canon's colorimetry is optimized for accurate caucasian skin tones, what does this do to non-caucasian skin tones?

The colorimetry is certainly not optimized for caucasian skin at the expense of other skin tones - it's just optimized to give a very pleasing look to caucasian skin because it's the most difficult to get right. As far as I can see, other skin tones don't suffer.
 
I'm out shooting some doco stuff today, and have made some adjustments to the profiles. It's amazing how different the camera can act on a chart vs being out in the real world. Anyway, I'm fine tuning the color response to primary colors, and things are looking really nice. Red looks far more natural and skin tones are still great. Will post the updates tonight.
 
Back
Top