skintone

[Jason Rudd]

Lukas Ettlin, the DP of Battle Los Angeles says good things about his experience with RED but he does note that he did have some trouble with skin tones:

Good: "The RED has come closest to emulating film of any digital camera I've used".

Skin Tone: "Our biggest technical challenge with RED on the Lincoln Lawyer was dealing with yellow/green artifacts in skin tones when using Kino Flo lights, low color temperatures, and certain types of make-up. We ended up using daylight balanced fixtures for our stage work, because the camera performed more cleanly in 5600K."

-ICG April 2011

So does RED perform poorly with skin tones in tungsten vs other cameras such as the Alexa in tungsten? I know the M-X chip is a daylight sensitive chip, what about the Alexa?

 

[Adam Eden]

In the days of 35mm film you would aways shoot a grey scale on set so the colourist had a point of reference on a first light and or primary colour balance.
I still shoot a grey scale on talent for everything i shoot.

If i have mixed lighting IE tungsten mixed with daylight then i always shoot a couple of frames of a Macbeth chart next to the talents face. This really helps when you are shooting with 2 cameras so you can easily match them
later in whatever package you are finishing in.

 

[Blair S. Paulsen]

Love the RED Cambook chart that Graeme worked with DSC to develop - don't leave home without it ;-)

On that note, I wonder if it would be possible to use that chart for alignment in a more structured way. What if you could frame up the chart using a special frame guide selectable in the frame guide menu that would allow you to place the swatches in known positions. Then you do the equivalent of a white balance setting (only with far more information) by simply pushing a button to create a unique LUT. That LUT can be applied to the monitor tap for on set fidelity (within the capabilities of the monitor of course) and perhaps put into the metadata header of the next shot.

In theory it seems like a great way to reconcile the known colors on the chart to the finished image by providing a way to null out variations introduced by lighting, electro-optical path, etc. Sort of like a grey card on steroids. In practice it may be worthless, but thought the idea merited discussion.

Cheers - #19

 

[Björn Benckert]

Love the RED Cambook chart that Graeme worked with DSC to develop - don't leave home without it ;-)

On that note, I wonder if it would be possible to use that chart for alignment in a more structured way. What if you could frame up the chart using a special frame guide selectable in the frame guide menu that would allow you to place the swatches in known positions. Then you do the equivalent of a white balance setting (only with far more information) by simply pushing a button to create a unique LUT. That LUT can be applied to the monitor tap for on set fidelity (within the capabilities of the monitor of course) and perhaps put into the metadata header of the next shot.

In theory it seems like a great way to reconcile the known colors on the chart to the finished image by providing a way to null out variations introduced by lighting, electro-optical path, etc. Sort of like a grey card on steroids. In practice it may be worthless, but thought the idea merited discussion.

Cheers - #19

That would be cool. frame it and then the camera color picks in each color and grayscale box and leaves a super balanced setting in the side car if wanted.

 

[paulherrin]

i'm pretty sure i remember jarred mentioning something about working on recognition of the red/dsc chart... i guess it's basically like facial recognition, or you could just tell it to 'look for' the chart. but that could happen on set and/or in rcx... you could have controls to dial in colors on a more individual level from the chart, plus you can always move any of the settings around that you want to.

 

[David Battistella]

Love the RED Cambook chart that Graeme worked with DSC to develop - don't leave home without it ;-)

i second that. I reviewed that chart a while back and I have it in the lens case. It's a great chart.


david

 

[Graeme Nattress]

Lukas Ettlin, the DP of Battle Los Angeles says good things about his experience with RED but he does note that he did have some trouble with skin tones:

Good: "The RED has come closest to emulating film of any digital camera I've used".

Skin Tone: "Our biggest technical challenge with RED on the Lincoln Lawyer was dealing with yellow/green artifacts in skin tones when using Kino Flo lights, low color temperatures, and certain types of make-up. We ended up using daylight balanced fixtures for our stage work, because the camera performed more cleanly in 5600K."

-ICG April 2011

So does RED perform poorly with skin tones in tungsten vs other cameras such as the Alexa in tungsten? I know the M-X chip is a daylight sensitive chip, what about the Alexa?

So from what I remember there was an issue with a certain combination of light, makeup and camera. I've seen makeup that looks very very different under different lights to my eye, never mind a camera. This type of complex interaction can and will happen with any camera.

Graeme

 

[Hrvoje Simic]

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

In addition (not negation), to avoid any generalization or misconceptions: the term "proper" may be perceived in two ways regarding this topic, which often seems to be perceived as the same thing:

As 1) Technically accurate and balanced
or 2) "The way it looks good to me"

1) WB is unbiased, vision-stripped and emotionless. 2) Eyes and aesthetics are biased and vision & emotion led.
Technically precise corrections and aesthetics in many cases can oppose one another. What is desired as "white" is not technically white, what is desired as "correct" is not technically correct, what is desired as "natural looking" is not photorealistic.
They can work together and work against each other.

In some cases a good starting point is slightly more "off" and subjectively "better" results appear if adequately balanced (if needed) later in the chain.
In some cases it's smart to balance at a starting point.

The key premise is that although the material is RAW and although in high sampling, and although one might use 32 bit float for shaping it, the starting point still affects the final outcome. It can affect it to a minor extent or to a quite relevant extent. Taking this into consideration allows routes and "looks" unavailable with previous typical approaches.

 

[Björn Benckert]

Lukas Ettlin, the DP of Battle Los Angeles says good things about his experience with RED but he does note that he did have some trouble with skin tones:

Good: "The RED has come closest to emulating film of any digital camera I've used".

Skin Tone: "Our biggest technical challenge with RED on the Lincoln Lawyer was dealing with yellow/green artifacts in skin tones when using Kino Flo lights, low color temperatures, and certain types of make-up. We ended up using daylight balanced fixtures for our stage work, because the camera performed more cleanly in 5600K."

-ICG April 2011

So does RED perform poorly with skin tones in tungsten vs other cameras such as the Alexa in tungsten? I know the M-X chip is a daylight sensitive chip, what about the Alexa?

I would say so, pretty much as further away from a balanced light at 5600K you are the greater are your chance's that your colors skew of in different directions if you then try to balance with a color picker on gray the oddities are quite likely to be more prominent.

 

[Matt Ryan]

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.

Thanks Bjorn!

Atempted this twice with no luck. Got them in resolve and generated 3D lut's from the dpx images, yet when I apply the LUT's to footage, no transform takes place. Anyone have these in Resolve format already?

 

[Mark Toia]

Good to see a positive end to this thread. :yesnod:

 

[Carlos Reyes]

Thanks Bjorn!

Atempted this twice with no luck. Got them in resolve and generated 3D lut's from the dpx images, yet when I apply the LUT's to footage, no transform takes place. Anyone have these in Resolve format already?

+1 That would be greatly appreciated!!!!

 

[Eirik Tyrihjel]

I grade from the R3D RAW files only... not colour baked DPX, not colour baked Pro res, not colour baked H264... I can grade exactly and quickly any skin tone I desire.
People saying they can't get perfect skin tomes from RED is something of a mystery to me.
Not sure why this subject comes up all the time. I can only assume people are trying to grade from compressed color baked codecs and not from the actual RAW files.

I skipped 10 pages of this thread, this is how I feel as well!

 

[paulherrin]

does RED perform poorly with skin tones in tungsten vs other cameras such as the Alexa in tungsten? I know the M-X chip is a daylight sensitive chip, what about the Alexa?

our planet revolves around the sun. i'd say it's a safe bet to shoot or filter for daylight. other spectrums are pale in comparison. that being said, i haven't had any technical problems with tungsten lighting and i shoot it alot, but sunlight cannot be beat. light = life, same is true when creating an image. just don't ignore the shadows, may even want to peer into them and see if you find a dragon.

Good to see a positive end to this thread. :yesnod:

do you guys filter in mixed/tungsten lighting these days?

 

[Matt Ryan]

+1 That would be greatly appreciated!!!!

yes it would! I'm banging my head on the table haha. Spent 2 hours trying several methods and can't seem to get it to work. Tried in photoshop and with LUT buddy. Both export fine, but they do nothing in resolve, and no transform happens...

 

[Hrvoje Simic]

I wonder if it would be possible to use that chart for alignment in a more structured way. What if you could frame up the chart using a special frame guide selectable in the frame guide menu that would allow you to place the swatches in known positions. Then you do the equivalent of a white balance setting (only with far more information) by simply pushing a button to create a unique LUT. That LUT can be applied to the monitor tap for on set fidelity (within the capabilities of the monitor of course) and perhaps put into the metadata header of the next shot.

In theory it seems like a great way to reconcile the known colors on the chart to the finished image by providing a way to null out variations introduced by lighting, electro-optical path, etc. Sort of like a grey card on steroids. In practice it may be worthless, but thought the idea merited discussion.

Cheers - #19

With that in mind, it would be interesting to know if the camera colorimetry is fixed, exclusive, and solely hardware based, or if it could actually be changeable/tunable to some extent, in such ways like using colorimetry presets dependent on - and optimized for - particular color temperatures, or assigning color values/range interpretation based on the chart example you mentioned. Not thinking about just a LUT here, as the camera is probably tuned to a particular color temperature having the most optimal results in it, and having more offset in others.

Variable camera colorimetry would be the equivalent of choosing different stocks which behave optimally in desired conditions, with drawbacks in other conditions.
Just an idea as well...

 

[paulherrin]

With that in mind, it would be interesting to know if the camera colorimetry is fixed, exclusive, and solely hardware based, or if it could actually be changeable/tunable to some extent, in such ways like using presets dependent on - and optimized for - particular color temperatures, or assigning color values/range based on the chart example you mentioned. Not thinking about just a LUT here, as the camera is probably tuned to a particular color temperature having the most optimal results in it, and having more offset in others.

Variable camera colorimetry would be the equivalent of choosing different stocks which behave optimally in desired conditions, with drawbacks in other conditions.
Just an idea as well...

i don't know anything about it like graeme does, but from what he's said and what i've read... i think it has to do with the silicon tech - every digital sensor (that i know about) responds naturally to daylight (could fluctuate around 5-6k?) - so after that it's just the color science to try and interpret that. that's about all i know. i'd say it's really great and has come a long way. try reprocessing some old R1 footage.

 

[Matt Ryan]

yes it would! I'm banging my head on the table haha. Spent 2 hours trying several methods and can't seem to get it to work. Tried in photoshop and with LUT buddy. Both export fine, but they do nothing in resolve, and no transform happens...

Anyone have these converted to .cube? Sorry but I've given up after 2 hours worth of attempts. Not getting it right apparently.

 

[Björn Benckert]

yes it would! I'm banging my head on the table haha. Spent 2 hours trying several methods and can't seem to get it to work. Tried in photoshop and with LUT buddy. Both export fine, but they do nothing in resolve, and no transform happens...

Sorry for all this I'm only on my little macbook air at the moment no photoshop no resolve... as I wrote, I posted this here before in this thread: http://www.reduser.net/forum/showth...t-Rec-709-LUT-Anyone-have-one-Made-one-Graeme

Some where in there they use AF to translate the 3dl into something usable for resolve. But if nothing else somebody at my office can punch them out next week so do not bang your head to much...

What it looks like that both Kodak and Fuji does on these print luts is that they quite heavily expand the midrange and bring up the tones skin colored tones in the middle. But it's quite far from the regular curves stuff more like like a 3Dcolorwarper. They also make something really nice with the highlights a rollof that looks very much like a well made film print.

Here is a little test clip I do not have the before sadly. It's just shot with some poor tungsten lamps and the light from the telly but I find it to look more than film would have done in the same condition and the guy's skin tone is not the most common still it looks very natural in a good way I think.

http://www.syndicate.se/Files/~usr/callboypilot/HAWK_100mmgradedsmall.mov

 

[Matt Ryan]

Sorry for all this I'm only on my little macbook air at the moment no photoshop no resolve... as I wrote, I posted this here before in this thread: http://www.reduser.net/forum/showth...t-Rec-709-LUT-Anyone-have-one-Made-one-Graeme

Som where in there they use AF to translate the 3dl into something usable for resolve. But if nothing else somebody at my office can punch them out next week so do not bang your head to much...

hahaha I even read that thread as well. Tried photoshop and fcp with LUT Buddy and no luck. I must be an idiot, but in both instances the lut's created did nothing haha. I felt like an idiot so I tried it over and over for a couple hours and then finally gave up. Thanks Bjorn!