PDA

View Full Version : IPP2 + Helium and SKINTONES



Brett_M
02-14-2018, 12:01 PM
I don't know why this isn't mentioned anywhere when I search, but I'm finding now with my new Epic-W (vs my EPIC DRAGON) I no longer have smooth faces/skintones. I used to never see in the Dragon workflow what I see now... muddy green patches of skin mixed with clay plastic looking patches of skin, zombie skin! The higher I go with the IPP2 contrast the more the color in the skin tones separates. I'm using unscientific language but if you are seeing this too you get it.

This could evolve into a highly technical conversation but I'd be interested in just knowing if other people are seeing this too? I guess I will have to graduate from my basic color correction using REDCINE and RED Raw settings and start exploring my own LUTS? I really loved being able to just gently tweak the RED settings!

Šabović Adis
02-14-2018, 12:07 PM
Any R3Ds to play with? :biggrin:

Brett_M
02-14-2018, 12:49 PM
Exactly what I'm not interested in, looking at R3Ds, I'm just looking for confirmation from other people that they're seeing this and that maybe the new color pipeline is not without its flaws.

Šabović Adis
02-14-2018, 01:09 PM
Got it.

Phil Holland
02-14-2018, 01:24 PM
I've oddly found exactly the opposite in relationship to IPP2. But yeah, without .R3Ds it is rather hard to know what you are seeing or experiencing.

Adam Montville
02-14-2018, 02:32 PM
I don't have much experience with Resolve, like at all, so learning new tricks is awesome. I was seeing greens in the blacks with the Epic-W on IPP2, but then I learned the Lum vs Sat curve, and turning down the left side down to 0. That removed all the nasty green! Give that a try on the footage. I did it for the entire timeline.

PatrickFaith
02-14-2018, 07:16 PM
IPP2 I think shows color much more correctly, I had some glass that I though was perfect, but with IPP2 I realized it had a bunch of depth chromatic aberration (i.e. colors were bent slightly differently when they were on a point slightly out of focus). Super quick way to test this, is shoot it at like t2, then shoot it at t6, the t6 will have reduced chromatic aberration if that is the problem.

Ryan Sauve
02-14-2018, 09:44 PM
I'm seeing the green pretty bad. I'm at -14 or more with the tint on Epic-W where I never had to go past -5 on Dragon. I also Spectrometered my lights and there's no significant green spike in them. But yeah, not very impressed with Helium's green bias so far.

Brett_M
02-15-2018, 09:26 AM
Again I apologize for not having .r3d's or keeping the language subjective and abstract. I don't like hearsay about camera image quality, but people do use natural language and metaphor to describe and make decisions around these things.

Its as if the pipeline is exaggerating yellow jaundice (bilirubin) around eyes or other areas? Other times I feel like when I use the contrast curves it begins to get that plastic overly noise corrected look, I keep wanting to use the term "clay"

There was a night that I pulled out the camera to show a friend and I realized that the cheap Ikea LED bulbs in my lights were likely causing all kinds of patchiness on the face. I know there are issues with new LED artificial lighting all over the city here, street lamps to subways that might be causing issues as well.

I will try to find an example. the 8k .r3d samples on RED's website with the caucasian model's face by Mark Toia has this yellow around her mouth and eyes, however that could be makeup.

Does anyone have links to tutorials of how to smooth skintones?

Björn Benckert
02-15-2018, 10:17 AM
Have not used helium lately. But in the begining it colorfringed quite heavily in the highlights. Depending on lens but also exposure so if not watching it the colors could go a bit wack in the top of the color cube. So the thing not yo do is to overexpose, especially skin tones as you loose accuracy.

In this film you see it in the highlights of the waves, purple colors that simply does not belong there.

https://vimeo.com/225964349


https://vimeo.com/225964349

Most visable in 4k. Dont remember what version of the color science this was but likly not the final ones and not developed in redcine but with luts in resolve so could be better. But to me those out of gamut fringing colors are still there on helium. Om monstro I see nothing of it. And if exposing correctly it´s not much of a problem on helium. I blame this one on my wife. I set the exposure and went ot kiting and she did the filming, ofcourse the sky cleared up and things got a little overexposed, so messed up the highlights, but the little sun made the kite day much nicer. :)


If having high exposure in the face the little skin bumps highlights picks up that kind of fringing thats, my theory as if you stop down just a bit colors are very accurate. But then ofcourse you also get inaccurate in the other end of the spectrum important to keep skin in the sweet spot if its going to look right.

Adrian Jebef
02-15-2018, 08:06 PM
Poor spectrum lights (like older / cheaper LEDs) will exaggerate the problem. As will lack of exposure. Test the same model under your LEDs in a low-light shot and then take them outside under daylight. Also be wary of old / poor NDs and Polas.

Björn Benckert
02-16-2018, 06:00 AM
Poor spectrum lights (like older / cheaper LEDs) will exaggerate the problem. As will lack of exposure. Test the same model under your LEDs in a low-light shot and then take them outside under daylight. Also be wary of old / poor NDs and Polas.

Yes, shit lights is 90% the fault when skin looks poor. The camera is made for clean full spectrum k5600 everything else is dragging down the performance. Tungsten is not optimal if you dont filter for it. And sheep or household leds are clamping colors for sure. A good way to see how good / bad your leds are is to look at their reflection trough the read surface of a CD, if you see a full rainbow, they are usually quite good. But normally you will see the rainbow being jagged and color fields missing in it. That means some colors are simply not there and will make skin and other things look clamped.

Zeb B
02-16-2018, 08:17 AM
Yes, shit lights is 90% the fault when skin looks poor. The camera is made for clean full spectrum k5600 everything else is dragging down the performance. Tungsten is not optimal if you dont filter for it. And sheep or household leds are clamping colors for sure. A good way to see how good / bad your leds are is to look at their reflection trough the read surface of a CD, if you see a full rainbow, they are usually quite good. But normally you will see the rainbow being jagged and color fields missing in it. That means some colors are simply not there and will make skin and other things look clamped.

Cool tip the CD is

Björn Benckert
02-16-2018, 08:25 AM
Cool tip the CD is


A Spectroscope probably works tons better. But who has such. :)

Marcus Friedlander
02-16-2018, 09:01 AM
Yes, shit lights is 90% the fault when skin looks poor. The camera is made for clean full spectrum k5600 everything else is dragging down the performance. Tungsten is not optimal if you dont filter for it. And sheep or household leds are clamping colors for sure. A good way to see how good / bad your leds are is to look at their reflection trough the read surface of a CD, if you see a full rainbow, they are usually quite good. But normally you will see the rainbow being jagged and color fields missing in it. That means some colors are simply not there and will make skin and other things look clamped.

Woah!!! That's an awesome trick! Thanks for sharing!