F55: Split Screen comparison of Venice-like LUT vs. LC709TypeA LUT

Doug Jensen

Doug Jensen

Carbonite Member
I decided to do a split screen comparison between the new Venice-like LUT and the popular LC-709TypeA LUT that is built into DaVinci Resolve to see what the differences look like. I often use the TypeA LUT myself as a starting point when I'm grading S-Log3/S-gamut3.cine footage. Instead of shooting charts, I used some random clips I already had on my hard drive to see how the two LUTs compared on real-world footage in a variety of different situations. Most of the footage originated as RAW or XOCN-ST footage from my F55 but there are some interviews towards the end that were shot with my FS7 (XAVC-I) at a lighting workshop I taught last summer in Maine.

NOTE: NONE OF THIS FOOTAGE IS SUPPOSED TO BE A FINAL GRADE. This is just a comparison of applying two different LUTs as the primary grade. Nothing else has been done to the footage except to make an exposure adjustment on some clips prior to the LUT being applied. The exact same exposure adjustments have been made to the A and B grades and that is why one looks a little darker. I didn't want to apply different exposure adjustments to the clips because that would introduce another variable. One of the two LUTs is always A and the other LUT is always B. I do not mix them up. Which one do you think looks better in this double-blind test? I will reveal which LUT is which after a few people have shared their opinions.

LUT 1: Slog3-S-Gamut3.Cine_To_s709_V200
LUT 2: Slog3SGamut3.CineToLC-709TypeA

 
Hi Doug, i prefer lutB because i 'm fighting against sony's magenta orientation.
My favorite Lut for Fs7 (slog3 CineEI s-gamut cine3 internally recorded with xavci) stills Kojicolor 2383 85% with litle addition of adobe kodac 2383 15% to brigth and desaturate skintone.
I use colorista IV to Apply luts tweaked with 3dlutcreator into premiere pro.
 
The differences I see (besides the brightness difference) is the magenta tint on the A image and the green tint on the B image. The B image is closer to a natural grade, but the ideal is somewhere in between. I would say that image A is the LC709A, since the Venice colorimetery is supposedly less magenta. I also see a small difference in contrast, which I would expect with the Venice LUT (lower contrast).
 
Last edited:
Good question. I assume you are talking about ACES input/output transforms? Two reasons: 1) I feel that have more control over the grades of individual clips when using LUTs instead of blanket transforms over everything. I don't always use the same LUT on every clip. 2) ACES is a resources hog and slows my machine down too much. So slow that grading becomes a chore instead of fun.

104 views so far of this thread and only two comments about the differences between them. Interesting.
 
Last edited:
You could use ACES but also YRGB color managed by setting the input, timeline and output color spaces or use YRGB unmanaged and using color space transforms using the OFX Color Space Transform plugin.

You can set transforms globally while still able to set different transforms on individual clips or groups or even in single nodes on a single clip.

LUTs are inaccurate since they interpolate data, both trilinear and terahedral mode have issues. Also LUTs sometimes clip data.
Color space transforms on the other hand are not interpolated, they for all intents and purposes never clip data and can be performed as gamma only or gamut only transforms or both. Color space transforms also add great flexibility to edit sources from different cameras in the same color space.

In my opinion not using transforms in Resolve is not using some of its greatest strength.
 
Last edited:
Thanks for the suggestion. I've experimented with that setup and still prefer working with LUTs. But you're right, it does solve the playback speed issue.
 
Cary, you inspired me to take a fresh look at the workflow you describe above that doesn't involve LUTs just in case anything has changed for the better in Resolve 15 since I last played with it version 12.5. But I still run into the same issues that I didn't like before, and that is mainly how to deal with exposure. I prefer exposing about 1.5 - 2 stops higher than Sony recommends for S-LOG because it is a lot cleaner, safer, and very easy to compensate for in post. I normally adjust for exposure on node 1 and then apply the LUT I want to use on node 2. It works great because I have brought the exposure down to where the LUT expects it to be before the LUT sees the video. But with the workflow you recommend, if I choose S-LOG3/Cine for the Input Color Space I don't see where I can adjust for the over exposure (on a clip by clip basis) before that transform occurs because that seems to be the first change that hits the video. The exposure cannot be easily changed with a node as far as I can see. It looks nuclear on the highlights and I don't see anyway to compensate for it. What am I missing?

BTW, I'm talking about XAVC clips. I already know that if I have shot RAW then I can adjust for exposure in the Camera RAW settings.
 
Doug - thanks for the test. After a quick watch, it's interesting how close they are in many of the shots. Overall, I'd choose B... mostly for the skin tones...but interestingly not always, since I do occasionally find I prefer some A shots.
Also would love to know, what the "actual" color of the helmet around the 01:03:26 mark was - and which LUT renders it more accurate.
 
I wish I could help you with the color of that helmet but I only saw it briefly almost a year ago though an un-LUTed viewfinder and I have no idea what the real color is! Your guess is as good as mine. I'm not going to say yet which side is which LUT or which one I prefer but it does illustrate one thing quite clearly that I have said for years -- you can't just throw a LUT on your footage and walk away from it. In my opinion, neither A nor B looks good enough to be considered a final image -- it all needs more work. And whatever small differences there are between these two LUTs will just be washed out with further grading towards a final image.
 
Doug Jensen said:
In my opinion, neither A nor B looks good enough to be considered a final image -- it all needs more work. And whatever small differences there are between these two LUTs will just be washed out with further grading towards a final image.
Click to expand...

Agreed. I'm honestly still scratching my head a bit with the release of this new "Venice" Lut.
 
For my old eyes the difference at the F55 was more visible than at the FS7 samples. As Mark said, none of them is perfect as it is for my taste.
 
I had a quick look through.

I think the differences are inconsequential - ie, either would be acceptable.

Minor colour/gamma differences, but I suspect you could grade with either and people will like the end result.

It's the hifi issue again - 24k gold or nickel? They sound very slightly different - but is one qualitatively better? No, I don't think so.
 
Oh, and how did you make the test double-blinded? (ie, you yourself don't know which is which.)

Did you get somebody else to apply the LUTs?
 
bengiles said:
Oh, and how did you make the test double-blinded? (ie, you yourself don't know which is which.)

Did you get somebody else to apply the LUTs?
Click to expand...

Uh, it's double-blind for YOU. I know which one is which because I'm running the experiment, but it doesn't matter because i didn't grade either side, I just applied a different LUT to each side. My actions have not affected the outcome.
 
bengiles said:
I had a quick look through.

I think the differences are inconsequential - ie, either would be acceptable.

Minor colour/gamma differences, but I suspect you could grade with either and people will like the end result.

It's the hifi issue again - 24k gold or nickel? They sound very slightly different - but is one qualitatively better? No, I don't think so.
Click to expand...

I already said the same thing in an earlier post. I guess we can find something to agree on afterall.
 
For general color/scenics I preferred A side from the F55.
However, I disliked the color of the A side on the FS7 faces. I also felt that the gamma on the A side was too flat for the FS7 faces, but that the B side FS7 highlights lacked character. Probably would be easier adjustment to get a color correct I'd like for the FS7 from the B side.
 
Doug Jensen said:
I normally adjust for exposure on node 1 and then apply the LUT I want to use on node 2. It works great because I have brought the exposure down to where the LUT expects it to be before the LUT sees the video. But with the workflow you recommend, if I choose S-LOG3/Cine for the Input Color Space I don't see where I can adjust for the over exposure (on a clip by clip basis) before that transform occurs because that seems to be the first change that hits the video. The exposure cannot be easily changed with a node as far as I can see. It looks nuclear on the highlights and I don't see anyway to compensate for it. What am I missing?
Click to expand...
If you want to mimic what you currently do with a LUT you could:

1. Create a node to adjust the exposure.
2. Create a second node using the Color Transform OFX and select:

Input Color Space: S-Gamut3.Cine/S-Log3
Output Color Space: the color space you want to work in (presumably Rec.709/Gamma 2.4 - avoid sRGB!)
​


3. Create a node to make adjustments

The maximum output is likely higher than 1 and that means that you have more dynamic range to work with than Rec.709 (this is a good thing!). It is now up to you to map this onto Rec.709. There is a tone map feature in the Color Transform OFX but I would not use that, instead I would first try to keep the mid-tone contrast by rolling off the blacks and whites for Rec.709. You might want to use the gamut mapping or leave that under your own control.
​


I would be careful in adjusting the exposure in log form, since each manufacturer's log curve is different there is no uniform way to lower or raise exposure. The (technically) correct way to correct exposure it to convert the video to linear light, then use the LGG Gain control and then convert back to log or any other format.

If you try this with Resolve 15 I would:

1. Create a node to adjust the exposure, right click on the node and set the Gamma to linear, use the LGG Gain control for exposure adjustments. This way the transformation will take place in linear light.
2. Create a second node using the Color Transform OFX and select:

Input Color Space: S-Gamut3.Cine/S-Log3
Output Color Space: Rec.709/Gamma 2.4​


3. Create a third node for adjustments to fit the high dynamic range log footage into Rec.709 and optionally adjust out of gamut colors.
 
Last edited:
You must log in or register to reply here.
Back
Top