Other: LUTCalc v1.9RC3 for generating MLUTs

cameramanben

Well-known member
Hi All,


after a couple of days of trying out every combination of adjustments I could think of (picking off a few bugs along the way), I've just uploaded a new build of LUTCalc:
LUTCalc Main Page
LUTCalc Github Releases Page
LUTCalc For Mac Github Releases Page

It's mostly bugfixes, some code reorganisation to be easier to maintain and develop, and I've slipped in a new colour tool I've been experimenting with. I call it 'PSST-CDL', for (P)rimary, (S)econdary and (S)kin (T)one colour decision list. The idea is that it is based on a polar version of YPbPr (Y is luminance, then Pb and Pr, which are basically the x and y on a vectorscope, are represented as an angle and a magnitude).

The big trick is that the 'angle' parameter between each 'colour' (blue->magenta->red->skin tone->yellow->green->cyan->blue again) is constant, 1. +/- 7 takes you back to your original colour.

It's rather like the HSV or HSL controls in Photoshop, but more so. ;-)


You can define colour shifts, saturation and the S, O and P controls of ASC-CDL for each of the seven channels, then click 'Refinements' to lock down values in between. On that tab there is a row of checkboxes underneath which locks the values. I do still need to add instructions within LUTCalc, but I'm keen to make them clear! ;-)

I've also finally shown a bit of love to the rather neglected 1D option and the ASC-CDL controls (minus saturation) are now available to 1D LUTs. :)

Anyway, please give me feedback and set me on any bugs. I've tried hard to make sure it's working before posting this, but it's very easy to miss something!

Ben
 
Last edited:
Well,

wouldn't be right if I couldn't find at least one wee bug! ;-) The re-analyse button in LUTAnalyst wasn't working. Nothing earth-shattering and obscure enough that hopefully I can be forgiven, but I've now uploaded the fix to all versions,

Ben
 
I'm having a play in the Chrome version of the app - looks cool. Thanks for doing this - obviously a lot of effort.

I haven't figured out how I'd use this in my workflow.

I'm currently shooting FS7 SLog3 Cine EI without a monitor LUT and loading Sony's standard SLog 3 Cine post production LUT within Lumetri in Prem Pro, on top of which I use the 3 way corrector to further tweak exposure and colour balance.

Sorry for the numpty question, but how might I use LUTCalc as an alternative to this? Would I need a set of different LUTS for different exposure pushes/pulls, for example? Or is this primarily aimed as a monitor LUT generator?

Cheers,

Ben.
 
Hi Ben,

going by your current workflow, you may well have found that if changing CineEI the Premiere Pro / Lumetri plugin doesn't pick up on the metadata and make the exposure adjustment automatically. You can use LUTCalc to make expsoure adjusted versions of whichever gamma / colour space combo you favour. I have an F55 with a jog wheel that allows plenty of variation (+/- 2 stops in 1/3 stop increments), but the FS7 has the three-way gain switch so three versions will cover you. Specific to the Premiere / Lumetri workflow, DIT James Marsden pointed out that Lumetri expects data in to data out LUTs to look correct. Having a bit of a play myself I can see what he means. It's actually a subtle difference, but the camera (and Sony published looks) are data in to legal out and the normal workflow in Resolve expects data in to legal out. You can use LUTCalc to produce a data to legal version for the camera and data to data for Lumetri.

Second, may I ask why you don't use a monitor LUT whilst operating? Shifting the iris by a stop will shift the midtones by around 9% IRE with S-Log3. Even the low contrast LC709 and LC709A MLUTs will shift midtones by nearer 14% and with Rec709(800%) it's getting on for 20%. As such whether by eye, zebras or waveform it's much easier to set accurate, consistent exposure with a viewing LUT than with the underlying log. Log is really only meant for conveying data, not looking at, but I suspect that the popularity of the C300 with it's very limited MLUT/viewfinder assist option has got people used to working off the log as the 'right' way to do it. ;-)

In addition, if you want to change ISO in CineEI then you have to use some kind of monitor LUT in order to see the effect, even if you choose S-Log3 as the LUT.

LUTCalc also displays a representation of your chosen gamma curve, along with %IRE values which are 'correct' exposure, ie as designed and recording the underlying log correctly. You can use these to set zebras or to work off a waveform.

I'm told that the FS7 doesn't display waveform in CineEI unless all the MLUT options agree, including 'SDI1 & internal' which then bakes in the look. This seems silly, but that being the case LUTCalc also includes the option of false colour. This changes specific luminances in the frame to blocks of colour, giving a very accurate sense of exposure, like multiple colourful zebras. Personally I find it distracting, but it came from experiments when Simon Bailey on the Sony forum asked about a false colour LUT and seems to work well for some. Just DON'T BAKE IT IN! :) One other note about false colour, if you CineEI down (lower ISO), maybe for lower noise or to keep the iris wide open, being a LUT the near white clip and white clip levels won't show. Green, Pink and Orange (if turned on) - that is 18% gray, one stop over 18% gray and 90% white - will all be fine.

LUTCalc now has some other tweaks, such as colour temperature and green/magenta adjustment to make up for the limitations of CineEI shooting, plus you can scale the black and user selectable highlight scene reflectance output levels to whatever you want (I've written before about reasons for adjusting the highlight level of LC709A).

ASC-CDL gives a simple, standard set of RGB controls and PSST-CDL can produce pretty complex effects quickly (I knocked up a LUT which smoothly set everything except skin tones and bright reds to black and white in a matter of seconds).

Finally, if you already have a LUT you like (cube file) or have built in other software, you can use the LUTAnalyst option to read it in, convert it for use with S-Log3/S-Gamut3.cine or S-Log2/S-Gamut and build exposure adjusted versions as discussed earlier.

Sorry for the endless post, but hope that some of that helps!

Ben
 
Last edited:
One last thing,

in case it's not clear, you do not need to make exposure adjusted versions of an MLUT. The picture, zebras and (when available) waveform will appear to shift with CineEI changes. The underlying recording doesn't change (it's just metadata), but the viewfinder and SDI/HDMI (if selected) will to all intents and purposes appear as if changing gain in Custom mode.

This is in contrast to the C300 which shows the waveform and (as I recall) the zebra levels of the underlying log when 'viewfinder assist' is on. The C300 bakes in gain in log to get around being 8-bit. For the way that log works in the 10-bit FS7 (exposure adjustment applied in post, no loss of dynamic range), the Sony way is much better; simpler and more consistent with ISO or MLUT choices.

Just thought I'd better mention this in case new FS7 / former C300 users aren't aware of the difference,

Ben
 
Last edited:
Thanks Ben.

Second, may I ask why you don't use a monitor LUT whilst operating?

Mainly because I've got into a groove since buying the camera, learning lots about it and then having to work around it's not inconsiderable bugs.

I'm generally using the 18-105 lens which won't auto-correct with MLUTs on at 1080P - or at least it didn't, but may have been rectified in v2. Not documented, but worth me doing a check.

Also, I'm using zebras as my main exposure tool. IIRC, the zebras work on the monitor stream, so they won't give me the SLog 3 clipping point (set to 61% IIRC) - just the REC709 MLUT-manipulated clipping point.

I like the idea of a more WYSIWYG viewfinder - but, to be utterly candid, I get scared off by the FS7's ability to bite me on the arse (eg the shutter resetting each time I change the frame size) and have been sticking with the devil I know. It feels like I need to spend a day resetting my knowledge of what does and doesn't work - and right now, that's looking like October...

you may well have found that if changing CineEI the Premiere Pro / Lumetri plugin doesn't pick up on the metadata and make the exposure adjustment automatically. You can use LUTCalc to make expsoure adjusted versions of whichever gamma / colour space combo you favour.

Yes - I have to use the colour corrector to bring it into line. I've spoken with Adobe and it's on their roadmap for PP to pick up the metadata and display exposure accordingly.

Would you mind giving me an idiot's guide as to how you see LUTCalc being used in a Prem Pro workflow? I'm getting baffled by some of the terminology - but I also have the baggage of almost 30 years as a film editor, so there's a very good chance I'm stuck in "old think."

BTW, I'm currently enjoying non-destructive flexibility in post - ie chuck a LUT/CC on a clip or adjustment layer for a "first light" effort, cut the film and return later, once it's locked, to the untouched underlying clips do the final grade. I've moved away from the idea of baking in a first light to my editing clips, as I'm finding Prem Pro is powerful enough to cut in full res, unrendered, with LUT/CCs added to all clips.

Or have I completely missed the point? I suspect the answer is "yes"...

Cheers,

Ben.
 
This is what I've deduced so far for a Prem Pro/LUT Calc workflow - please correct any misunderstandings, Ben.

1) In Prem Pro, export a jpeg from the a sample of untouched SLog material.

2) Import that jpeg into LUT Calc and tweak accordingly, adding any exposure/colour correction

3) When you're happy with the look, export the LUT with a meaningful name.

4) In Prem Pro, add a Lumetri effect to the SLog clip. Reference the LUT you just gave a meaningful name.

5) Add sharpening to taste.

So far, I think it's fab. The ideal scenario further down the road would be to have a LUT Calc plug-in that can go straight onto a clip, like a Lumetri effect, and to then have the same tweaking power within the NLE.

In the meantime, thanks Ben for making this - it's brilliant.

Ben.
 
Back
Top