Look Transforms

[jedsmith]

Just a quick post to say I’ve added two more tools for look creation:

NotoriousSix Value : Nuke | Resolve
Adjusts brightness of color by hue angle.

Could be used for similar purposes as the HK compensation I started the thread with, or for something else entirely.

NotoriousSix Vibrance : Nuke | Resolve
Vibrance emulates what happens to hue and chroma in the bottom end of a per-channel contrast increase: Chroma is increased and secondary hues are bent towards primary hues. Image “richness” is increased without slamming into the gamut boundary as with a traditional saturation adjustment.

This one could be useful for getting back some of that per-channel look in the midtones and shadows.

Happy testing…

[Derek]

Regarding the desire to have swappable Rendering Transforms along-side the default ACES look (for example K1S1, IPP2, etc) I note that T-CAM has these as looks (which it calls “camouflage looks”). Would it not be possible to do the same with looks for openDRT? That is, openDRT could come with a number of analytical LMT‘s which emulate popular Rendering Transforms like K1S1 in addition to the the default ACES look. This seems to me to be in line with how many studios (don’t) use ACES Output Transform currently, bypassing like so:

IDT > ACES > logAWG > K1S1 > invODT > invRRT > RRT > ODT

This would have the advantage of empowering artists, without them needing to invert the RRT+ ODT which is… um… suboptimal. Instead you would just do:

IDT > ACES > K1S1 look > OpenDRT

[daniele (Daniele Siragusano)]

If you make OpenDRT just an exact copy of TCAM than you can also take an exact copy of the camouflage looks.
Great.
I hope we can formulate a vision which goes beyond the status quo.

[jmgilbert (Jean-Michel Gilbert)]

I think that is part of the fundamental conflict that we were trying to resolve yesterday isn’t it. I would vote for having both swappable renderings and default LMTs : have your cake and eat it Also, I would vote for OpenDRT being its own thing while being flexible enough to support the kind of looks that you have in BaseLight without infringing on your proprietary IP

[daniele (Daniele Siragusano)]

I think my last comment sounds a bit rude.

I wanted to express that I find it a waste of time to reimplement already existing solutions instead of opening the architecture so that existing solutions can coexist with new ones.
My apologies.

[Derek]

Here’s some more experiments with Jed’s Look tools.

Zone Grade

I’ve been playing around with the zoneGrade and it’s pretty amazing. It makes it possible to dial in the desired tone mapping with granular precision. Very happy that I’ve been able to get that “skin sparkle” I’ve been missing in previous looks. Going for a “fall off the truck” nice look that I’d want for doing CG lighting work, rather than a neutral starting place that a colorist might want for beginning a grade. I did however pull back the contrast, and lift the shadows to avoid the crushing of the shadows in the current ACES tone scale. YMMV.

Zone Grade Settings

global
exposure: 0.76, offset 0.006, contrast 1.05
high
exposure -1, pivot -2
low
exposure 0, pivot 0.7
higher
exposure 2, pivot 0.65
lower
exposure -1.25, pivot 1.55

Here’s OpenDRT before

and here’s with the Zone grade:

Notorious6 Vibrance

The ZoneGrade desaturates which gives it a kind of film emulation look which in kinda cool. This tool makes it possible to get saturation in the shadows which really helps faces to not look flat and “colorized.” Previously I was attempting to do this with Zone Saturation, but I’m finding it’s much easier to have precise control with the No6 Vibrance.

Notorious6 Vibrance Settings:
red 0.5, blue 0.5, green 0.5
zoned low, zone range -2

Additionally I added global saturation with Jed’s Saturation tool in ACEScg luminance weights.

Saturation Settings:
weighted ACEScg
saturation 1.055

Here it is added to the above image (OpenDRT + ZoneGrade + saturation + No6 Vibrance ):

Notorious6 Hue Shift

This is for everyone’s favorite golden sunshine glow! This is zoned to effect bright stuff like lights or fire, without effecting faces. The goal being to capture the golden halo glow that my eye sees when first glancing at luminous objects like fire or light bulbs. CAM meets Impressionism if you will.

Notorious6 Hue Shift Settings
yellow -0.2
zoned, zone range 1

Switching images as this has virtually no effect on the last image. Here’s the image in ACES

sky is lemon

With the above look (OpenDRT + ZoneGrade + saturation + No6 Vibrance ):
![image|690x389]

sky is pink

And with the vibrance and hue shift: [OpenDRT + ZoneGrade + saturation + No6 Vibrance + No6 hue shift ]:

sky is golden

Here’s the same with Legos

ACES

pink coffee cups (OpenDRT look without hue shift)

golden coffee cups (OpenDRT look with hue shift)

The hue shift only effects yellow so there is no impact on red:

OpenDRT

OpenDRT + Look

Here’s some fire. In ACES

OpenDRT

OpenDRT + Look

CG skin in ACES

OpenDRT

OpenDRT + Look

If there’s a TL;DR it’s that these Look Tools from @jedsmith make it possible to add pretty much whatever look you want to OpenDRT, making something possible which was not possible with ACES. Pretty excited about that

[gray (Gray Olson)]

@Derek

I’m using IPP-2 as a reference for an example of what appealing skin looks like. I also looked at ALF and K1S1 which also render skin nicely. My intention is not to attempt to match any of these, but rather to identify what is nice about them in how they display skin and to attempt to get the same qualities in a look for OpenDRT without any unwanted “side-effects” such as skews or changes in color.

Hi, I realize I am sorta necro-posting in this thread and especially to your post here which was back in July. But, having just read through a lot of posts in this forum, I just want to highlight something that’s important which I seem to find repeated by many of the more artist-side people and I’m just using this as an example/vehicle to do so

With that disclaimer, the point of my post:

I think the comparisons made here to other “”“DRTs”“” are completely ill-formed in the first place because, for example, IPP-2 uses an entirely different philosophy for building a “”“DRT”“” than what we are attempting to create here/what OpenDRT is attempting to be.

What I mean is that, we don’t care or even want to care about whether a DRT makes skin look pleasing because the DRT (as we are defining it) is not meant to do that… in fact, that IPP-2 makes skin look good is in some ways a bad thing because it implies that it is modifying the “look” instead of being a faithful, transparent, reproduction of that look.

A more fair comparison would be to extract whatever it is that is modifying the skin tones in IPP-2 into a separate LMT which is applied before OpenDRT, and ensure that OpenDRT isn’t doing anything to destroy or otherwise change that “look”.

The fact is that any comparisons or evaluations of OpenDRT (or indeed the newer CAM-model-based DRTs that are being experimented with) which is based on a visual “does this look good on this input footage” test are ill-formed in my opinion. It is an essential part of the new DRT workflow to use an LMT to create a Look which you then want to be retained by the DRT.