FS7: also F5/55 F3 - LUTCalc

L

Lenilenapi

Carbonite Member
Have you guys seen the app by "cameramanben" called LUTCalc:

https://cameramanben.github.io/LUTCalc/

There is a thread on the Sony community site about it called False Color.

http://community.sony.com/t5/FS7/Fal...497849#U497849

I'm not totally clear on it only because Ben's language is sometimes more complex than I grock and it seems to good to be true - but it seems as if LUTCalc will read any LUT you choose then make adjustments to the LUT and create a new one with those adjustments.
The thread began with Ben explaining how it could make a false color LUT for use in the FS7 but more exciting to me if I understand this correctly is that it could also alter a LUT to compensate for fluorescents or to lower blacks etc. Haven't tried it yet but it looks exciting.

"LUTCalc The Adjustment Tool
As well as displaying the basic appearance of different looks, LUTCalc can make a number of adjustments. Output levels can be scaled to preferred values for black and arbitrary reflectances, different colour spaces can be applied to shadows and highlights and the colour temperature can be adjusted, either as a CTO / CTB shift or by directly setting recorded and output temperatures.
Once the input, output and adjustment options have been set LUTCalc can then output .cube LUTs suitable for use in camera as monitor LUTs or in post to provide one-click base looks."
 
This tool looks absolutely remarkable. It takes a bit for the javascript app to load due to it's complexity. There is no hand-holding with this app so you will have to know what you are doing.

Thanks for pointing this out.

EDIT:

A few things of note. It allows you to do cross conversions of LUTs from cameras like the a7S and the F3 which are often left out in the cold as far as LUT utilities go. Second, it has compensation tools for Fluro lighting with minus green adjustment sliders. It also has a similar adjustment for color temp differentials and laterly LED compensation which would be near impossible to quantify in my view due to the way LEDs generally use slices of the color spectrum. Lots more. You got to dig around and study it CAREFULLY.
 
Last edited:
Hi All,

I'm the one writing LUTCalc. Thanks for the comments. Thought I'd better chip in as I've just discovered a really dumb bug that crept in recently whilst tidying up for release. One line of code which seemed to turn off almost all the adjustments, so sorry if it didn't seem to do anything! Please do give it another chance.

LUTCalc Main Page
LUTCalc For Mac Release Page

Anyway, it should work as expected now, plus if you look in the box with the charts, there is a button marked 'Instructions' which then brings up a graphic you can click for info about any of the settings. Hopefully clearer than my rambly posts which tend to get splurted out in the middle of the night,

Ben
 
Last edited:
If it about exposing to a LUT and to Log, then the figures in LUTCalc are all the correct baseline ones. Correctly exposed at f2.8 on S-Log3 is correctly exposed at f2.8 on S-Log2 and correct on LC709A - that's also why MLUTs are there. LC709A correctly exposed will put caucasian skin tones rather low (55%-58% IRE) - same with 90% white reflectance (71% IRE) - for most tastes as it holds within the legal range of 0-100% yet captures a little over six stops above mid gray. That's a heck of a lot to squeeze in considering that Rec709(800%) and HG7 and HG8 capture a little under 5 1/2 stops above mid gray and go all the way up to 109%. At 100% legal clip they are more like 3 to 3 1/2 stops above mid gray.

Out of the tin the LC709 looks are intended as guide looks which include the full dynamic range but expect further adjustment in post. I've posted elsewhere about how stretching LC709A can help, and I frequently bake-in LC709A variants I produce. In LUTCalc the basic method is that you do that by ticking 'Highlight Level', entering 1350% in the 'Reflected' box and then 108.9% in the 'Maps To' box. Then tick 'Camera MLUT' in the top right box to make sure that all the ranges are appropriate. The important thing is that ANY LUT exposed as designed will be correctly exposed for any other standard curve. It may not look nice, but then you adjust to taste. LUTCalc will give you the 'correct' IRE values in the bottom right box.

If it's about the ISO of the camera, the native ISO settings in the top left box are used to calculate stop shifts - the absolute value isn't relevent (with the Sony approach to log), so don't worry.

Log curves are extremely forgiving of exposure if handled correctly - that's the whole point. Sony captures the full range of the camera and stores metadata to tell post software how much to shift. Trouble is lots of software doesn't automatically pick up on that metadata. The result then is that footage at non-native ISO either looks too bright or too dark, and if you change CineEI it bounces up and down. That was a big reason for the origins of LUTCalc. You can get it to do pure exposure shifts by setting input and output gamma to the same thing and then just changing 'Stop Correction'. Setting 1D LUTs is best for that if you are using it in post software like Resolve (though Resolve still offers 1D cube LUTs under '3D LUTs' for some reason!). You won't want either clip options ticked in the top right box.

Log recording is much more about consistency than absolute values. Overexposing a couple of stops is EXACTLY the same as setting a CineEI of two stops below native but without the metadata - less headroom but also less shadow noise. That's a fairly extreme example but... pays your money, takes your choice.

The only really important thing with log is that ISO is NOT adjusted with gain. It's a bit more complex than that and gain will give terrible results if you try to use it to shift CineEI where exposure metadata hasn't been read. I do know people who do try to change log curve exposure with gain and it just makes things ugly and unforgiving.

I and a number of camera people I know regularly shoot F55s and F5s at 500-800 ISO when we don't need six stops headroom, both as an alternative to ND and to reduce shadow noise. I also know a very good cameraman who technically speaking overexposed an entire production because he set S-Log2 to what looked good in the viewfinder. Post were very happy as he was very consistent and still with plenty of headroom, so it was just as if he had CineEId down a stop for cleaner blacks. Which, cough cough, is what he intended all along!

If you meant something else, sorry for the waffle! Just fill in a bit more about what you do mean. Thanks,

Ben
 
Last edited:
Sorry SpareWheel,

just read through that thread you linked and presumably you just meant that LUTCalc got mentioned so I should tell them about the bug and fix. Thanks for the heads up and sorry about the enormous, probably stating-the-deeply-obvious ramble,

Ben (who should learn not to post late at night!)
 
Last edited:
Ben,
Already posted on the Sony site... but then I just found post #6 above which explains the "app" even better. So - another quick thanks, for posting late at night which helped me understand a few more tricks to play with on the LUTCalc.
 
Ben,
I can't thank you enough for what you've done with LUTCalc and with your detailed explanations on the board here. Very helpful for me to get a handle on this and finally explains why I have been confused by results from 709TypeA.
I have a few questions about the LUTCalc:

I find the Preview window a nice addition but have a couple of questions about it.
If you adjust your ISO down (say in SLog3 from 2000 to 1000)_ the preview window gets darker. Is that because you only have one shot there and it was exposed at native ISO ( 2000) ? Thus to actually get a feel for what it would look like with other adjustments , should you temporarily leave the ISO at 2000, then change it before generating your LUT?

Similar question though this is more obvious . Re the Fluorescent adjustments: In Advanced you can choose a specific bulb like Cool White, but the correction in Preview only seems to reflect the Kelvin part of that bulbs output. Does that mean I still need to add a magenta/green offset based on my own guess or trials?

Third is more an aesthetic question. Why not always place the black level output for "0". Wouldn't this be the natural place for it? Why bother to elevate the level at all especially like 709TypeA does at 5%IRE?

4th - Would I be correct is assuming that for making an altered version of 709TypeA that 709TypeA should be chosen both for Output Gamma and Output Gamut.

Lastly please stop criticizing yourself . It is charming to see how humble you are but - you know a great deal, have been very generous in sharing that knowledge, and have created a very valuable little app - no need to apologize about anything - certainly not for being long winded or boring.

BTW - Will you be at NAB ? If so I'd love to say hello. I'll only be there Tuesday.

Lenny
 
Sorry Lenny,

I'm afraid I'm english and that makes me genetically obliged to apologize for everything. Oops, there I go again, sorry. D'oh!

As for your questions:

the preview window shows what the corrections and adjustments will do to picture data; set the CineEI down a stop and you need a stop more light, the picture gets darker. CineEI one stop up in ISO and you need less light, the picture gets darker. The preview image is, as you suggest, at a reference exposure and again, as you suggest, I would say that the best approach is to do all your adjusting (or read in a LUT) with no exposure adjustment and only then create all the CineEI versions you need.

Two other points about exposure. First, careful producing exposure adjusted LUTs which might end up in the camera as MLUTs. Including any ISO shift in the filename is a good idea. I did a test once where I deliberately produced plus and minus three stop shift MLUTs (along with plus and minus one stop). I was testing out comparisons of baking in exposure-shifted log against doing the shift in post on 10-bit material. I was taking it to an extreme beyond the basic +/- 2-stop range. Basically I got a bit mixed up doubling up CineEI and LUT shift and managed to end up shooting ISO 40,000 on my F55! Won't be trying that again in a hurry. ;-)

The other thing is that the ASC-CDL adjustments (in fact all the adjustments) are done at a scene linear stage, to keep the mathematics clean and relatively simple. What this means is that the 'Slope' control, which is basically a multiplication, acts as a true exposure control. a value of 2 is one stop brighter, a value of 0.5 one stop darker. Handy to be aware of if you want to control the response whilst steering clear of the ISO options.

I've consciously tried to make the controls 'cameraman-y' and simple, but keeping the underlying parameters available if needed, eg colour temperature as CTO/CTB, but also the more strictly accurate temperature to temperature and choice of chromatic adaptation model (basically colour temperature is adjusted once converted to a colour space mimicing the response of human eyes, then converted back to the original colour space; the different models are because there are different interpretations of eye responses, and because the data is anyway limited to three values out of a continuous spectrum).

Going to the Fluori/LED, this is one I'm still mulling the best way to show the controls. Where colour temperature shifts along a line of 'black body' white values known as Planck's Loci, I've made the Fluori/LED a shift at right angles to that. The basic slider is around daylight temperature, which is distinctly green/magenta and in extent and scaling very roughly like plus and minus green gel (I checked out the specs of a couple of manufacturers). At low colour temperatures that tends more towards cyan/yellow as the Planck Loci are a curve. I've included lamp temperatures in the 'advanced' setting as I have some ugly very warm (low kelvin) energy efficient bulbs at home which definitely do come out rather more yellow than green (and under which I also have to tweak the colour temperature adjustment), but when you factor in that the phosphers in fuoris are varied and distinctly not full-spectrum it's more a guide before trial and error with imported framegrabs than something to be strict about. It's probably confusingly literal and I am trying to mull a better way of putting it, but for now it's hidden behind an 'advanced', aka 'there be dragons!' checkbox. ;-)

I can't honestly say why the black levels are where they are. I could give a few reasonable guesses, but a transmission engineer or colourist would be better placed to reel off the truth. Anyone?

Finally, yes, if you set output gamma and output gamut to LC709A it should be a facimile of the look released by Sony. In fact, I tuned the LUTs that they are stored in so that if you set input gamma to S-Log3, input gamut to S-Gamut3.cine and tick the MLUT box, the resulting LUTs will be numerically identical to the Sony LC709 or LC709A looks to all six decimal places.

That's not to say that it would be such a dead-on match with S-Log2/S-Gamut as the inputs, but the idea with all these is to give a base to work from, either just as MLUTs or as a starting point for a grade.

Sadly, I won't be going to NAB - I'm due to be in Birmingham on Tuesday (the soggy one in the midlands of England, not the sunny one in Alabama ;-) ) Enjoy it,

Ben
 
Amazing work! I'm particularly interested in the chromatic adaptation tool, achieving accurate white balance in post is a vital step to me (kelvin & tint), with RED cameras it's easy to just shoot grey chart and use eye dropper to quickly correct the white balance.

Since Sony does not offer "pick WB" functionality in any of its post processing softwares, I was wondering if it's possible to have a plugin (After Effects or OFX) to white balance shots? Something similar to this: http://aescripts.com/ft-auto-white-balance/, but with more accurate Bradford transform method.
 
Please forgive my ignorance but is LUTcalc opening the possibility of using 3D LUTs with the F3? Could I cross brand and use Arri LUTs (Slog2 or 3) in my F3 + Odyssey 7Q package (I know I can but is it technically safe and correct since the F3 is before all these LUTs started)? although I have the PMW F3 for almost 2 years I haven't really learned (Trying with courses now) the ways to manipulate gammas inside and outside this camera. I understand the only Log the F3 has is Slog1 EE R800 and it is in most cases not a very nice option for narrative films with this camera and recorder. I purchased your app a few minutes ago and I am looking to understand it! I use Avid MC but I am looking to start de-logging with DaVinci using LUTs I can create or at least test on location and like. Any suggestion? Thanks a lot.

Sebastian
 
You must log in or register to reply here.
Back
Top