skintone

[Cliff Hokanson]

YEA my clients want the alexa look for better skin tones/film look. I feel Jim is thinking about this!!!!!!!

 

[paulherrin]

well, it certainly never hurts to shoot a chart, especially in a studio environment. i hardly ever do, though... for better or for worse.

that would be a really cool feature: take an r3d snapshot of the chart that gets stored straight into the metadata, not as a part of the actual take (keeps it out of the way).

 

[Björn Benckert]

i mismatched 2... right in the green. could be my eyes, could be my old, uncalibrated laptop's rendition, or just sloppiness.

besides color perception/tolerance... there is another factor. as you mentioned, there is a range of skintones... and a range of light filtration. i know this is a huge no-no for some people, but i'm okay with letting the light affect the skin color. or the mood of the shot for that matter (within creative reasoning). [heck, look to the left at my 'avatar'... 5 year-old canon jpg (i'll grow up sometime) and a little intense... but that's besides the point. the light bends a certain way at sunset and fluctuates based on time (season) and space (geo)... and 'artistic intent'.] i understand, at least a little bit, the idea of 'memory colors' - but perhaps i have a little bit of artistic leeway in my philosophy. i'm still just learning all this stuff, though - you guys each have much more experience than i do, so i'm willing to listen and learn about this stuff and really appreciate this discussion. sorry to throw a tangent out there, but it is a factor... (and a balancing out of technical minutia)

so, what does 'accurate' skintone rendition look like, and is archetypical 'accuracy' via plato the ultimate form? how do you guys think about these things?

How do you mean missmatched two? what was your score? If yours was 2 then I will keep mine a secret

 

[Patrick Tresch]

Well, we owe you a beer. That's for sure. Awesome test!

I will download the originals since I can't realy see 100% what is going on in the color. The look is very glamorous and awesome but personally I'd like to try pushing a more saturated look and see what's there.

Only further 2 questions would be: light source (5600K is stated but is it HMI?) and makeup used.

Oh - and how much do we need to bribe you to do another test just like this with tungsten lighting?

Thanks again!

Bruce Allen
www.boacinema.com

Hello Bruce,

I didn t do the test myself. Thanks for the beer! But It was HMI, IMHO the key light had a bit green. They filtered the Delta as on screen it looked green, and then they took out the filter as the results wheren t good.

The main point I see in today's camera technology, that contributes to the quality of the picture, is the human factor, not the technology one.


Pat

 

[David Battistella]

So the reason to never shoot a chart?

It's like saying, "don't record a tone for the audio for this scene because the actors are whispering and I don't want it to be loud." There is no good argument for not getting a chart on a scene. I agree that it is not necessary in the instances you describe, but as an example, what if you have an Alexa an EPIC and a 5D for a scene? Pick one and match the rest?

There is no harm in shooting a chart.

The DP can always do screengrabs for the colorist at wrap, but shooting a chart is a good practice.

David

Shooting a reference chart and white balancing based on it may be less than optimal strategy, in cases such as:

- when shooting a scene with varying light throughout the shot (camera movement, uncontrolled light, mixed light etc.)
- when scene's actual captured color values are very far from perfectly white balanced color, and typical WB creates an undesired starting point by negatively affecting either texture, range or tonality
- when it creates a longer route to the desired look in any way
- when challenging and important shots lack the reference chart but are chosen as a reference for the look creation
- when aditional fiddling with additional variable on a fast moving shoot with 100+ shots starts presenting a distraction or limitation of any sort, being less helpful than a adequate colorist.
...etc...

 

[paulherrin]

How do you mean missmatched two? what was your score? If yours was 2 then I will keep mine a secret

well, i'm just guessing the score is based on which ones were 'out of place' apparently there were just a couple right next to each other in the wrong arrangement (i guess). and don't worry, tests scores are lame anyways (hardly a medical-grade test anyways). what was revealing about that test (and this thread, and our current brainstate) is that we're wired to more readily pick out 'what's wrong'.

anyways, don't be so sensitive story of my life...

 

[Björn Benckert]

well, i'm just guessing the score is based on which ones were 'out of place' apparently there were just a couple right next to each other in the wrong arrangement (i guess). and don't worry, tests scores are lame anyways (hardly a medical-grade test anyways). don't be so sensitive story of my life...

I though I was not to sloppy and that I had some fine gradients but I scored 53. Maybe I should take my hands off the color wheels from now on, LOL.

 

[Björn Benckert]

regarding the 3dl luts posted herre before I forgot to say if you want them in resolve they need to be converted trough photoshop according to the following (found in another thread)

If you have Photoshop CS6:
Import Resolve's Trim_LUT0.DPX file into Photoshop
Add a LUT Layer
Choose the .3DL LUT
Flatten the image
Save changes to the file
Import to Resolve
Choose the option to Generate Lut from Current Grade.

edit: make sure to turn off your Photoshop color management settings so the file isn't converted on import.

 

[paulherrin]

I though I was not to sloppy and that I had some fine gradients but I scored 53. Maybe I should take my hands off the color wheels from now on, LOL.

van gogh wasn't an artist because he saw the world like everyone else... do what you do. your stuff is good.

and thanks for posting these resources, guys. super cool thread.

 

[Scott Crawley]

How do you mean missmatched two? what was your score? If yours was 2 then I will keep mine a secret

Since we are comparing dicks... I got a 7. That's pretty high for my age. I'm OK with that. ;-)

 

[Ramesh Jai]

regarding the 3dl luts posted herre before I forgot to say if you want them in resolve they need to be converted trough photoshop according to the following (found in another thread)

If you have Photoshop CS6:
Import Resolve's Trim_LUT0.DPX file into Photoshop
Add a LUT Layer
Choose the .3DL LUT
Flatten the image
Save changes to the file
Import to Resolve
Choose the option to Generate Lut from Current Grade.

edit: make sure to turn off your Photoshop color management settings so the file isn't converted on import.

Thank you!!!! Was wondering how to use them.

 

[paulherrin]

haha... oh no. i opened the gates. and i'm stuck with the smallest. it's how you use it, play the game.

if we're going to skew our topic here, tell me what you guys think about allowing light/atmosphere/mood to affect the 'accuracy' of skintones (either in the grade or as a tandem between dp/colorist). i feel the emotional impact of a color that is slightly 'off' or immersed in a greater source of light/color can be huge. to me, color symbolism/resonance is more important than scientific accuracy... but it's all comparative, you've got to have the natural subtleties to start from.

my whole point is, we're gonna grade it anyways... it's just a matter of education/tools getting out there, no?

[see a page or two back for my question about this, may have gotten buried]

 

[Scott Crawley]

Well Paul in this case, smallest is best. Finally it is working to your advantage, eh? ;-)

I wonder how well we would have done with a limited number of moves.

It's a funny thing, context. The first time I came upon this test I was following another link from their website. The referring article (If I remember correctly) was making the point that our perception of color is relative and that is why we need a reliable reference. Sliding those chips around really drove the point home.

 

[Graeme Nattress]

It might lean to the green because there are two green pixels and one red and one blue in the bayer pattern.

the bayer pattern sensor captures more green pixels.

but as you said it's very easily corrected with white balance and temperature shifts.

FWIW I am finding that daylight comes in at around 5200 consistently over 5600 which is always a touch warm, but it could also be the time of year in this part of the world.

david

Nothing to do with green pixel ratio. It's everything to do with the variability of light and that standard tools for measuring colour temperature are woefully inaccurate.

Graeme

 

[Hrvoje Simic]

So the reason to never shoot a chart?

Premise 1: This technique is really useful
Premise 2: This technique may also be not really useful or even redundant
Conclusion: This technique is never useful.... ?

It's like saying, "don't record a tone for the audio for this scene because the actors are whispering and I don't want it to be loud."

No, it isn't. No need to generalize.

It's like saying: "No need to do 10 steps if you can do 6."
Or: "All techniques do not work in all situations".

There are six examples mentioned where this approach "may be less than optimal strategy".
No words like "never" or "don't" in that post.

what if you have an Alexa an EPIC and a 5D for a scene? Pick one and match the rest?

Too many variables to provide a single, fool-proof answer.

- If shots from all three cameras have the same priority in the edited sequence, it may be advisable to pick the least manipulable material from 5D, go as far as one can with the look, and adapt other two camera looks based on it. (if one wants to avoid 5D material falling apart and unnecessary complications to fix it)
- If Alexa's color rendition is the priority, it would be advisable to adjust the look for 5D in camera (not in post), closest to desired look as possible (before it gets burned in the 8 bit h.264) and just do the necessary fine tuning in post.
- If it is a very challenging scene regarding light variations or light deficiency, WB route + later look creation/application with 5D will feel like trying to paint a canvas made of butterfly's wings.
- If recording Prores on Alexa and final desired look is drastic and less "accurate", it may be advisable avoiding the typical WB step for Alexa. This particular statement is also pretty simplified, with both pros and cons, depending on more detailed circumstances.
- If the DoP crafts the main "look" on set, filling the range and painting the tone precisely as he wants it (many do), or just gets very close to the desired look, post work requires just some secondaries and fine tuning and typical WB route and charts may be completely redundant.
- Charts + WB route may also be the best route in other scenarios.

It also depends on whether the chart is being used with a purpose of WB or just for comparing the color differences and offsets between cameras or between different shots of the same camera, and treating the material accordingly, without the typical WB step. (notice the repeating term "typical", not generally dismissive)

If cameras responded exactly the same way throughout all shooting conditions (with just different chromaticity between different cameras but no offsets throughout all shooting conditions), did that in same range, with same precision, and same digital equivalent of "density", the answer for this example would be more straightforward. Suggesting simply: "It's advisable and foolproof always depending on : Shooting charts > Doing the WB > Applying the camera matching LUT > Crafting the look for one camera > Applying the crafted look on all shots > Done. "

Since chromaticity, DR and recording format are not the only variables, with additional inter-dependent variables such as: consistency/"thickness"/precision/flexibility/gamut range/type and amount of recorded data, etc...and since camera choices, shooting conditions, workflows and subjective routes introduce a large amount of variables, which are in many cases over-generalized or over-simplified, the answer is not that straightforward, color chart or grayscale chart will not be always necessary or even helpful, and generic LUTs and plugins will not provide optimal or sufficient solutions in all cases.

Of course, that does not negate usefulness of these approaches in many appropriate scenarios.

The DP can always do screengrabs for the colorist at wrap, but shooting a chart is a good practice.

Of course.

In addition (not negation), screen grabs introduce additional CMS variables in the pipeline.
Which may not be necessary. In a bunch of cases. Some of which being mentioned.

 

[paulherrin]

Nothing to do with green pixel ratio. It's everything to do with the variability of light and that standard tools for measuring colour temperature are woefully inaccurate.

Graeme

do you have any general recommendations for getting proper color temperature (on set and/or in post) that we might not be thinking about? i tend to go by eye (and scopes).

 

[David Battistella]

To summarize.

Its not a bad idea to shoot a chart, it's just not always useful.


peace
david

There are six examples mentioned where this approach "may be less than optimal strategy".
No words like "never" or "don't" in that post.

Too many variables to provide a single, fool-proof answer.

- If shots from all three cameras have the same priority in the edited sequence, it may be advisable to pick the least manipulable material from 5D, go as far as one can with the look, and adapt other two camera looks based on it. (if one wants to avoid 5D material falling apart and unnecessary complications to fix it)
- If Alexa's color rendition is the priority, it would be advisable to adjust the look for 5D in camera (not in post), closest to desired look as possible (before it gets burned in the 8 bit h.264) and just do the necessary fine tuning in post.
- If it is a very challenging scene regarding light variations or light deficiency, WB route + later look creation/application with 5D will feel like trying to paint a canvas made of butterfly's wings.
- If recording Prores on Alexa and final desired look is drastic and less "accurate", it may be advisable avoiding the typical WB step for Alexa. This particular statement is also pretty simplified, with both pros and cons, depending on more detailed circumstances.
- If the DoP crafts the main "look" on set, filling the range and painting the tone precisely as he wants it (many do), or just gets very close to the desired look, post work requires just some secondaries and fine tuning and typical WB route and charts may be completely redundant.
- Charts + WB route may also be the best route in other scenarios.

It also depends on whether the chart is being used with a purpose of WB or just for comparing the color differences and offsets between cameras or between different shots of the same camera, and treating the material accordingly, without the typical WB step. (notice the repeating term "typical", not generally dismissive)

If cameras responded exactly the same way throughout all shooting conditions (with just different chromaticity between different cameras but no offsets throughout all shooting conditions), did that in same range, with same precision, and same digital equivalent of "density" & and insameamount of data, the answer for this example would be more straightforward. Suggesting simply: "It's advisable and foolproof always depending on : Shooting charts > Doing the WB > Applying the camera matching LUT > Crafting the look for one camera > Applying the crafted look on all shots > Done. "

Since chromaticity, DR and recording format are not the only variables, with additional inter-dependent variables such as: consistency/"thickness"/precision/flexibility/gamut range/type and amount of recorded data, etc...and since camera choices, shooting conditions, workflows and subjective routes introduce a large amount of variables, which are in many cases over-generalized or over-simplified, the answer is not that straightforward, color chart or grayscale chart will not be always necessary or even helpful, and generic LUTs and plugins will not provide optimal or sufficient solutions in all cases.

Of course, that does not negate usefulness of these approaches in many appropriate scenarios.




Of course.

In addition (not negation), screen grabs introduce additional CMS variables in the pipeline.
Which may not be necessary. In a bunch of cases. Some of which being mentioned.

 

[David Battistella]

Graeme,

I knew I could draw you into this thread. What are you noticing with current color temperature meters/tools.

David

Nothing to do with green pixel ratio. It's everything to do with the variability of light and that standard tools for measuring colour temperature are woefully inaccurate.

Graeme

 

[Graeme Nattress]

do you have any general recommendations for getting proper color temperature (on set and/or in post) that we might not be thinking about? i tend to go by eye (and scopes).

Shoot a chart! All the colour temp meters I've used just don't get accurate enough. And bringing a NIST certified spectroradiometer on set is overkill.

Graeme

 

[paulherrin]

haha, okay. i'll grab a chart next time.