ACEScc Color Space

Comments

RogerS wrote on 10/6/2020, 7:55 AM

I don't think you can use S-log 2/sgamut3.cine without the proper IDT if you want to have it be accurate. Maybe there are paid 3rd party services that can create IDTs on demand? Or just use the standard s-log 2 IDT that exists in Vegas and accept it will be somewhat wrong.

Color space transform is yet another different workflow but there's no Vegas equivalent I'm aware of.

Gerald Undone did some useful tutorials as of late showing different ways to color correct in Resolve using charts. Without a proper IDT or LUT to exactly match your settings I'd do it by hand with a Colorchecker (or other chart.)
is one. He also has a deeper live grading video published after this.

(This might be helpful to someone as an overview to ACES https://www.provideocoalition.com/is-aces-right-for-you/)

Musicvid wrote on 10/6/2020, 9:17 AM

@RogerS Thanks for sharing your in-depth experience. I'm learning from it without having an unlimited equipment budget to play with.

alifftudm95 wrote on 10/6/2020, 10:25 AM

I don't think you can use S-log 2/sgamut3.cine without the proper IDT if you want to have it be accurate. Maybe there are paid 3rd party services that can create IDTs on demand? Or just use the standard s-log 2 IDT that exists in Vegas and accept it will be somewhat wrong.

Color space transform is yet another different workflow but there's no Vegas equivalent I'm aware of.

Gerald Undone did some useful tutorials as of late showing different ways to color correct in Resolve using charts. Without a proper IDT or LUT to exactly match your settings I'd do it by hand with a Colorchecker (or other chart.)
is one. He also has a deeper live grading video published after this.

(This might be helpful to someone as an overview to ACES https://www.provideocoalition.com/is-aces-right-for-you/)

The color chart he is using cost 1K+ USD, damn.

 

PC DEKSTOP

CPU: Ryzen 5 3600 X

GPU: RTX2060 6GB

RAM: 32GB 3200MHZ

Storage: 480GB SSD & 1TB HDD

Monitor: BenQ PD2700U 4K HDR

 

LAPTOP

Dell Inspiron 15 7577 4K 8Bit Screen

CPU: Intel i7 7700HQ

GPU: GTX 1060 6GB Max-Q

RAM: 16GB RAM

Storage: 128GB SSD & 1TB HDD

 

 

Musicvid wrote on 10/6/2020, 8:44 PM

@RogerS

What LUT do you recommend for 2020 to 709?

RogerS wrote on 10/6/2020, 10:31 PM

Also, for ACES here's where I started reading about it:
https://mixinglight.com/color-grading-tutorials/getting-know-aces/
It appears the Input Display Transforms are provided by manufacturers so that might explain why less common combinations aren't included in Vegas (or any other ACES supporting software).

@Musicvid Thanks, my budget is far from unlimited as well : ) There are cheaper workarounds- let me explain more.

@alifftudm95 For charts, yes he's using an excellent Chroma du Monde chart which is individually measured and where the RGBCMY primaries line up *exactly* on the vectorscope. That said, in previous videos Gerald used the X-Rite Colorchecker Passport Video and got good results. That chart is close enough to be useful and perfect for camera/lens matching, even if it deviates somewhat from an absolute standard.

For workarounds... well I don't own that chart and have the cheaper X-Rite Colorchecker Passport Photo (v1). That has the advantage of being the standard Colorchecker chart. The fine folks at BabelColor took the reference data from X-Rite for the current formulation of it and created a synthetic chart (scroll down to "From X-Rite L*a*b* D50 (formulations AFTER Nov. 2014)" )
https://www.babelcolor.com/colorchecker-2.htm#xl_CCP2_data

While the Passport Photo has a RGBCMY row, it does *not* line up exactly with video vectorscope graticals (and sadly can't customize the position of the graticals). So I used the sRGB reference chart (convert to HD Rec 709 in Photoshop to correct the gamma) to compare a clip of my chart with to match up the primary colors. Babel's comparison of measured data from pre-2014 charts shows pretty good agreement so I suspect recent charts also don't vary *that* much (and if they did, they'd be useless for creating DNG camera profiles which is their main selling point. )

Step one is to match luminance using the grayscale step wedge row.
Step 2 is matching colors with the vectorscope and selective hue (I switched to Graide Color Curves for speed and for smoother transitions).

For a sanity check, I compared this to a PNG Paul Leeming posted in his Facebook group of a shot perfectly balanced for the Chroma Du Monde. I recommend joining that group. He helpfully included both the Colorchecker video and photo in the same shot. I borrowed his corrected shot to see where my primaries should line up. Not a big difference from the synthetic reference data.

Here's an example matching 3 Sony cameras shooting Cine 2:

I then adjusted the saturation/hues to precisely match and saved them as Graide presets with the camera name. In theory you could create a correction LUT from this Fx in Vegas but in practice found the hues were different between the Graide edit and the LUT made from Graide. I don't now why.

I use it on top of the Leeming Pro II Sony Cine 2 LUT to get rid of minor differences between his A7III and my a6600 and the somewhat larger generational difference between the a6600 and earlier a6500 and RX100IV. They are all similar but I'm after precision in my camera matching.

RogerS wrote on 10/6/2020, 11:03 PM

For Rec 2020 to Rec 709, Paul Leeming has a LUT for Sony HLG Rec 2020 cameras that converts it to Rec 709. Note that while the primaries will match the gamuts are different so the actual colors may vary if you attempt to match with a camera shooting Rec 709 (I saw an example where the yellows/greens were more saturated coming off the Rec 2020 camera post conversion).

I'm not really engaged with HDR filmmaking as I don't have a HDR enabled TV or editing monitor and don't need to be on the cutting edge. Paul Leeming did get one recently and once he masters the workflow will put out technical correction LUTs for Rec 2020 as well.

Disclosure: I don't have any business relationship with Paul Leeming/Leeming LUTs. I was a tester for the Pro II Sony profiles and am a happy user as the LUTs save time (esp. matching cameras like when I was asked to shoot Panasonic for work). Gerald (Undone) and others use them for the same reason.

alifftudm95 wrote on 10/17/2020, 12:53 PM

Also, for ACES here's where I started reading about it:
https://mixinglight.com/color-grading-tutorials/getting-know-aces/
It appears the Input Display Transforms are provided by manufacturers so that might explain why less common combinations aren't included in Vegas (or any other ACES supporting software).

@Musicvid Thanks, my budget is far from unlimited as well : ) There are cheaper workarounds- let me explain more.

@alifftudm95 For charts, yes he's using an excellent Chroma du Monde chart which is individually measured and where the RGBCMY primaries line up *exactly* on the vectorscope. That said, in previous videos Gerald used the X-Rite Colorchecker Passport Video and got good results. That chart is close enough to be useful and perfect for camera/lens matching, even if it deviates somewhat from an absolute standard.

For workarounds... well I don't own that chart and have the cheaper X-Rite Colorchecker Passport Photo (v1). That has the advantage of being the standard Colorchecker chart. The fine folks at BabelColor took the reference data from X-Rite for the current formulation of it and created a synthetic chart (scroll down to "From X-Rite L*a*b* D50 (formulations AFTER Nov. 2014)" )
https://www.babelcolor.com/colorchecker-2.htm#xl_CCP2_data

While the Passport Photo has a RGBCMY row, it does *not* line up exactly with video vectorscope graticals (and sadly can't customize the position of the graticals). So I used the sRGB reference chart (convert to HD Rec 709 in Photoshop to correct the gamma) to compare a clip of my chart with to match up the primary colors. Babel's comparison of measured data from pre-2014 charts shows pretty good agreement so I suspect recent charts also don't vary *that* much (and if they did, they'd be useless for creating DNG camera profiles which is their main selling point. )

Step one is to match luminance using the grayscale step wedge row.
Step 2 is matching colors with the vectorscope and selective hue (I switched to Graide Color Curves for speed and for smoother transitions).

For a sanity check, I compared this to a PNG Paul Leeming posted in his Facebook group of a shot perfectly balanced for the Chroma Du Monde. I recommend joining that group. He helpfully included both the Colorchecker video and photo in the same shot. I borrowed his corrected shot to see where my primaries should line up. Not a big difference from the synthetic reference data.

Here's an example matching 3 Sony cameras shooting Cine 2:

I then adjusted the saturation/hues to precisely match and saved them as Graide presets with the camera name. In theory you could create a correction LUT from this Fx in Vegas but in practice found the hues were different between the Graide edit and the LUT made from Graide. I don't now why.

I use it on top of the Leeming Pro II Sony Cine 2 LUT to get rid of minor differences between his A7III and my a6600 and the somewhat larger generational difference between the a6600 and earlier a6500 and RX100IV. They are all similar but I'm after precision in my camera matching.

I Join ACES forum recently, and I found this, seems like it is possible to create IDT

http://acesidtdctl.tcolorscience.com/

PC DEKSTOP

CPU: Ryzen 5 3600 X

GPU: RTX2060 6GB

RAM: 32GB 3200MHZ

Storage: 480GB SSD & 1TB HDD

Monitor: BenQ PD2700U 4K HDR

 

LAPTOP

Dell Inspiron 15 7577 4K 8Bit Screen

CPU: Intel i7 7700HQ

GPU: GTX 1060 6GB Max-Q

RAM: 16GB RAM

Storage: 128GB SSD & 1TB HDD

 

 

RogerS wrote on 10/18/2020, 12:47 AM

Very interesting, thanks for sharing. It looks like it lets you mix and match some common gammas and gamuts. It appears to be based on publish manufacturer data. I'd be curious to see how accurate it is using a chart.

I don't think I'll try, though as for Sony it's focused on log, which I have no desire to use with my 8-bit cameras.