ARRI Christmas LUT

[rand thompson]

One interesting thing is that if you "pregrade" enough, you'll find you've corrected the scene without a LUT. Try two nodes in a row and match the gamma and colorimetry that way. It can be done most of the time, particularly for just a straight camera image without a heavy stylized look.

Anton and Marc,

I love building S-curves. So I thought I would build an S-curve for Anton's Dropbox image that would give it more dynamic range. I didn't use any qualifiers or power windows. The S-curve is actually two curves in two separate nodes working together. The 1st to bring down the highlights and the 2nd to give contrast to the shadows/mids. This is just a Lut( If you can call it that) with a S-curve only in it. It doesn't have any color matrix to properly transform the RGB values into the REC709 color space or does it do the job of an actual Lut.



The 1st to bring down the highlights

[/url]1st curve by rand thompson, on Flickr[/IMG]

The 2nd to give contrast to the shadows/mids

[/url]2nd curve by rand thompson, on Flickr[/IMG]


Here is what they look like combined

[/url]A001_C011_1227DQ by rand thompson, on Flickr[/IMG]

Here is the Parade for that image.

[/url]parade for curves combined by rand thompson, on Flickr[/IMG]

and Here is the "Pseudo Lut" with the curves combined in it.


https://www.dropbox.com/s/igqu8njukyjoaet/lut with curve for anton_1.A001_C011_1227DQ.cube?dl=0

Requires REDwidegamutRGB-Log3g10

 

[Marc Wielage]

Anton and Marc, I love building S-curves. So I thought I would build an S-curve for Anton's Dropbox image that would give it more dynamic range. I didn't use any qualifiers or power windows. The S-curve is actually two curves in two separate nodes working together. The 1st to bring down the highlights and the 2nd to give contrast to the shadows/mids. This is just a Lut( If you can call it that) with a S-curve only in it. It doesn't have any color matrix to properly transform the RGB values into the REC709 color space or does it do the job of an actual Lut.

Yes, that will work essentially as a 1D LUT to change the gamma curve. But you don't need to use two curves -- you can use contrast, pivot, the Log controls, and the Primary controls to mimic the same things. One curve is really enough.

Again, if you had a full-frame DSC chart, you could see exactly where the grayscale should go and where each of the primary colors and hues should go in terms of level.

 

[Anton Esteban]

Marc and Anton, I have a question for you. When making a 3D LUT what are the best tools that get translated into a LUT and what are the practices that you should avoid? I'm going to post some interesting results soon that I've been making with Christoph's tests. I was quickly able to get a correction going that got the Helium footage very close to the Alexa. The highlight response of the Alexa is still untouchable but even there I feel I've improved on my original LUT especially in regards to skin tones and overall nuance in tonality which I've verified by testing on a number of other Helium clips. At this point, I want some expert advice on how to best translate those corrections into a 3D LUT especially to avoid banding issues.

 

[rand thompson]

Yes, that will work essentially as a 1D LUT to change the gamma curve. But you don't need to use two curves -- you can use contrast, pivot, the Log controls, and the Primary controls to mimic the same things. One curve is really enough.

Again, if you had a full-frame DSC chart, you could see exactly where the grayscale should go and where each of the primary colors and hues should go in terms of level.

Thanks again Marc, for your vast knowledge.

When I started using resolve(about 6 to 7 months ago) I would create a black and white "gradient" at resolve's free version's max resolution of 4kuhd. I would then import the "gradient" into resolve to see what "S-curve" the various combinations of "contrast" and "pivot" would make in the"waveform" scope.However, I always went back to using curves like I did in avid media composer's symphony color correction optional add on.



Marc,

Just tried using "Contrast" and "Pivot" again, and it does give a simpler as effective result. Thanks again

 

[rand thompson]

Redo with better "white point", less noise, better overall image.

[/url]Untitled00086407jjjj by rand thompson, on Flickr[/IMG]

 

[Marc Wielage]

Redo with better "white point", less noise, better overall image.

I have to say, once you stumble on Highlight and Shadow, those can be extremely effective in some situations.

But often, I'll struggle prior to the DP walking in to retain all the detail in windows like what you've shown, then they get in, sit down, and say, "oh man! I didn't want to see any of that -- blow it all out!" So you never know what the creative intent is. Even in a clipped/blown-out situation, I think there's a way to do it softly, maybe with a bit of a glow, that takes the harsh digital edge off it. (And I would say the same for Alexa, Blackmagic, Canon, Panasonic, and Sony, in addition to Red.)

I had a case not too long ago where the filmmakers had a specific reason not to want to quite see what was going on out the windows, and we wound up doing kind of a hazy/glow/defocus thing that I thought worked well. It wasn't a typical "burn the window straight to hell" look, and looked "real" enough that you could buy it for the scene.

Note that none of these things can be done with LUTs -- it's all knobs and experimentation. It also helps a great deal to use a panel for these adjustment; trying to do it with a mouse alone is for the birds.

 

[Stacey Spears]

then they get in, sit down, and say, "oh man! I didn't want to see any of that -- blow it all out!"

And here is a perfect example of where HDR can bite you in the butt if you wanted it blown out. Might blow out in SDR but can be all there in HDR and blowing it out might blind you.

 

[rand thompson]

I have to say, once you stumble on Highlight and Shadow, those can be extremely effective in some situations.

But often, I'll struggle prior to the DP walking in to retain all the detail in windows like what you've shown, then they get in, sit down, and say, "oh man! I didn't want to see any of that -- blow it all out!" So you never know what the creative intent is. Even in a clipped/blown-out situation, I think there's a way to do it softly, maybe with a bit of a glow, that takes the harsh digital edge off it. (And I would say the same for Alexa, Blackmagic, Canon, Panasonic, and Sony, in addition to Red.)

I had a case not too long ago where the filmmakers had a specific reason not to want to quite see what was going on out the windows, and we wound up doing kind of a hazy/glow/defocus thing that I thought worked well. It wasn't a typical "burn the window straight to hell" look, and looked "real" enough that you could buy it for the scene.

Note that none of these things can be done with LUTs -- it's all knobs and experimentation. It also helps a great deal to use a panel for these adjustment; trying to do it with a mouse alone is for the birds.

Marc,

I hear what you are saying. You really don't want to give your audience too much info in a scene. Sometimes things in the background can distract the audience's attention from what should be important in a scene. I still personally like grading manually without Luts, like the image above. Luts can be usefull for technical transforms of colorspace to colorspace. However, I think relying on them too much dulls your color corrections skills, Like not lifting weights in a long time and you get kind of flabby.

I'm trying to get a "good" panel that I can afford because like you stated, All of this "Mouse-clicking" is really for the birds.

I think one of my biggest problems, IMHO, is that I like the "technically accurate" look and I always go out of my way to keep it "Technically accurate" but with improvements. If you asked me to give you a creative/artistic original look, It would be like asking me to explain Einstein's "Thermodynamics"

 

[rand thompson]

And here is a perfect example of where HDR can bite you in the butt if you wanted it blown out. Might blow out in SDR but can be all there in HDR and blowing it out might blind you.

Stacey,

I have absolutely no experience at all with shooting in HDR. I didn't realize that it held that much info. in the Highlights.

 

[Björn Benckert]

Stacey,

I have absolutely no experience at all with shooting in HDR. I didn't realize that it held that much info. in the Highlights.

I think Stacey means not when shooting HDR but when viewing HDR screens. As they have such a wide range of luminance. So when showing pure white its a bit like watching a welding torch. Simply the eye dont like to view that high difference in light levels. So the iris of the eye has to adjust for the dark part of the screen or the white window but on a small screen it simply becomes difficult to watch.

 

[Hrvoje Simic]

I think Stacey means not when shooting HDR but when viewing HDR screens. As they have such a wide range of luminance. So when showing pure white its a bit like watching a welding torch. Simply the eye dont like to view that high difference in light levels. So the iris of the eye has to adjust for the dark part of the screen or the white window

Which is one of the reasons why generally grading in HDR makes zero sense.

 

[Björn Benckert]

Which is one of the reasons why grading in HDR makes zero sense.

Agree, there definitly are brightness and monitor DR levels that is simply to good to the human eye. we want to watch somthing that is a bit like log not like the real light levels. Film a sunset and show it so bright you burn your eyes when watching the screen does not make sense.

 

[rand thompson]

I think Stacey means not when shooting HDR but when viewing HDR screens. As they have such a wide range of luminance. So when showing pure white its a bit like watching a welding torch. Simply the eye dont like to view that high difference in light levels. So the iris of the eye has to adjust for the dark part of the screen or the white window but on a small screen it simply becomes difficult to watch.

Thanks for explanation Bjorn.

 

[Anton Esteban]

Hey everyone, here's v2 of my Helium to Alexa LUT. I have a good correction going on that I think is an improvement on the first ARRI LUT I posted. The problem is I hit the limit of what a 3D LUT can actually correct. One of the details I learned from Christoph's test is how Helium and ARRI read color’s Luminosity differently. For example, Alexa’s tend to read blue a lot darker than the Helium chip. But when you try and translate that into a LUT it causes banding issues. My solution around this was making a LUT that mimics the ARRI Rec 709 gamma response, color hues, and color saturation (all corrections that seem to translate well into a 3D LUT) and then uploading a resolve color grade that adds the color qualifiers that I used to get the luminosity of those colors closer to the Alexa (making luminosity changes to specific colors seems to cause the most problems in translating a correction into a 3D LUT from my testing). The beauty of this strategy is you can disable or refine the color nodes to taste.

In order to use this, all you have to do is put the Helium to Alexa LUT in your LUT folder in Resolve and then import the still that has the DRX file with the correction on it and then relink the LUT on the node that has a LUT on it. There is also a TINT/White Balance node before the LUT that you can key the green to taste. The better option would be to disable that node and use the Tint and White Balance controls in the RED Raw file. For best results you have to white balance your camera per lighting scenario (I left that in there for people who shoot PRO RES). Also, expose to save your highlights as much as possible.

As you can see from the PNG's I posted in the link, the LUT and correction is not a proper transform in any way. Look at how the colors on the xrite change between exposures of my correction compared to the official ARRI Rec 709 Transform or RED's IPP2 Transform. The beauty of the node structure I'm supplying is you can adjust the color nodes to help compensate for these shifts as long as you properly shoot with a color chart between cameras your trying to match.

I'm also including stills of Christoph's tests which I white balanced from the originals so you can quickly see the differences between corrections. I applied the same correction to all the exposures per correction. I also used the same tint and white balance settings between all the Helium Shots.

I continue to be surprised by this camera. I honestly didn't think I could get it close to the ALEXA but I'm happy with my results. I still think the color fidelity and dynamic range of the Alexa is still the one to beat but the price, resolution, RAW codec, form factor, and upgradability of RED's surpass that gap in my mind.

Enjoy!

https://www.dropbox.com/sh/k9mfatftdjroilz/AADwe6_sxVO80lgRNVJOIi7Ca?dl=0

 

[John Schick]

Marc and Anton, I have a question for you. When making a 3D LUT what are the best tools that get translated into a LUT and what are the practices that you should avoid? I'm going to post some interesting results soon that I've been making with Christoph's tests. I was quickly able to get a correction going that got the Helium footage very close to the Alexa. The highlight response of the Alexa is still untouchable but even there I feel I've improved on my original LUT especially in regards to skin tones and overall nuance in tonality which I've verified by testing on a number of other Helium clips. At this point, I want some expert advice on how to best translate those corrections into a 3D LUT especially to avoid banding issues.

Hey Anton, great effort so far in making this LUT. You had mentioned you were interested in learning more about LUT build quality. I’ve made a free LUT Stress Test Image that helps you see the most extreme areas that the LUT is affecting. I would recommend using an image like this when building or checking a LUT:
http://truecolor.us/downloads/lut-stress-test-image/

As an example, here’s v1 of the Christmas LUT against the stress test (Click for larger images):


Here’s v2:


And just for a comparison, here’s the stress test using a RED to Alexa Rec709 Preview LUT that I recently finished creating (it’s not a free package though):


There’s a lot of complexity involved in making a high-quality LUT. I’ve found that sometimes when you adjust one section, that creates problems elsewhere, so it’s a good idea to also check everything on highly saturated gradients to see if it still holds up. Also, increasing the viewing gamma (on top of the LUT) helps show if there are any artifacts hiding in the shadows.

 

[Anton Esteban]

John, where have you been all this time? You could have saved me so much trouble Had I known this was even out there, I would have just bought it. The good news for me is I learned how to simplify my correction process to get a correction in Resolve that helps me get an image out of the Helium sensor that I'm much happier with and that came about by trying to reverse engineer the ARRI rec 709 LUT. I've always admired how subtle an image comes out of the Alexa. I've never been happy with RED's color out of the box. I never liked its skintones, color contrast, or the way it handles highlights and shadows. But I have learned that there's plenty of information in the RAW file to get a more nuanced image that I like. Of course this is all my subjective taste but I'm happy to have gone through the process and learned so much from all the very intelligent people on this board who steered me to a much more precise and scientific methodology in color correction. I'm just sick of this whole RED vs ARRI thing. This journey has taught me that the new DSMC2 RED bodies are amazing and in the hands of people who know what they're doing, the sky's the limit. I'm definitely buying your LUT right now. Thanks for posting this!

 

[Anton Esteban]

Amazing work John, been putting it through its paces right now.

 

[Kevin Lang]

Thanks John! Just made the purchase myself and look forward to putting it to work!

 

[Craig Lees]

Hi John,

How does your Red to Arri preview LUT look with the new helium sensor?

Cheers

Craig Lees
UK

 

[Anton Esteban]

I've been trying to reverse engineer John's LUT to see if I could get a correction that would work with John's stress image in Da Vinci Resolve and I can't even get it in the ball park. This LUT is by far the best RED auto-correct I've ever seen. I still find you need to tweak it a little bit to get it closer to various side by situations but I've been able to do this in minutes with a few qualifiers and a good color chart. Amazing work John. Thank you for putting this out there!