dermot.shane

a fool and their money are soon parted....

https://www.lightillusion.com/luts.html

on the good side, Arri LUTs and Light Illsuion's seem pretty clean and useable without creating any needless dammage to your image, but test tes test always.....

one disadvantage of using the numpad ui is not being able to adjust say red down and blue up interactivly, you can adjust one -or- the other, and unless you have near endless time, you will never know how much better you could have made the image.. that interactivity thing

a worth tradeoff for precision? i'd say yes,  i use those tool early in the stack /layer / node tree, aim is to normiise the image cleanly there

For me precsion comes in later, the race is not won in the first corner, the grade is not finished in the first stack / layer /node

on all the surfaces i use when working in Resolve (Advanced/Mini/Elements/Color) they are mapped already, some better than others, not sure about Wave/Micro/Ripple tho

under Baselight, Color & Elements both offer the functional equliivent, as does Slate/Blackboard

under Nucoda Elements offers the functional equliivent, as does Precision

no idea about Luster and Scratch, id be surprised if exposure is not avb on a knob / ring / ball of some sort..

my thoughts ran to why would someone prefer useing the num pad in 2018 to adjust exposure rather than useing a knob to get to the same place much faster + more interactive feedback?

as to the tools themselves, i turn to them first on any system, no matter what they are called....

i get all that - but where's the advantage to the "printerlights" ui?

10 hours ago, Amada Daro said:

Old tech but the only primary controls with full tonal range control. 

care to expand on that thought?

arn't printerlights in Resolve nothing more than a diffrent ui for Offset?

Baselight's "Exposure", Nucoda's  "Brightness", Resolve's "Offset" are all the same thing?

here's a rip from Nucoda's manual on the tool;

Print stock filled the role of a ouput LUT/RRT, we called printer lights while viewing a answer print,  looking at something though an effective  RRT/LUT

try looking as something through a 2383 lut, then a 2393 lut, the print stock is your lut in a lab finish

that said, i prefer subtractive color and tend to use the a & b channels of L*a*b to ballance an image

i've used printer lights for real, and see no reason to turn back to 1941 tech

ugly, nasty, noisy, dark shite / lurvelry skintones / clippy, nasty, souless bright shit?

I hae 2383 lut's dateing back to the 90's when i used Kodak Display Manager for callibration, and Flame for comping, long long long time before Steve started Light Ilusion

Mocha remove is in the Resolve BCC ofx package

there is no "precisley matching" a film stock, too many variables in lab tollerance for any match to be more than lab X on day X at best

shoot a test, cut the camera roll, send it to three labs and the resualts are three densities, send it again a week later and there will be three more densities.... rinse and repete....

and they are all going to be within lab tollerance....

or use lum-vs-sat curve 1 node powergrade and turn down key output as needed?

about the same as Alex for a feature,  10-20 days depending on budget, although i've not seen a feature with only 800 cuts in it for a deacde or more (if ever), 1800-2200 is more common in 2018 with my clent base

but for TV i can go alot faster once we have looks in place, and that usualy happens when gradeing the pilot or next season's sizzle reel and trailers

typcialy 10-12 hr per 48 min ep, more at the begning of a season, and less at the end, more cut-n-paste of looks from prevoius shows in the same set by ep12, we are really only looking at new sets and QC issues. and that goes fast

some LUTs have applaingly bad maths,

i drop any LUT over a grey scale ramp and look for math errors before using, usualy i re-create in the software to avoid said errors, as they usualy contain math errors.

i often run it in groups, but being from the film era, i know the grain of Fuji500t has next to zero to do with the grain of Kodak 50D...  so i often group by 50 / 160 / 500 to match the scene

usualy i'll run one of many grain pass i made about two decades ago in Flame using it's match grain to match to the grain in the scans of a grey card

from a user viewpoint, BLE's tools are awesome complement to MC,

but not being able to navagate the timeline so one can play a sequence for review in context (BLE is locked to one clip at a time only) is problematic,

the workaround of closeing the software playing a seq in MC's ui, then launching the software is also problematic

Unlwess / Untill that is sorted, my hopes are for Daylight to get ported to Linux... i'm not going to buy a trashcan, and last time i looked that's the only machine it will run on

On 1/4/2018 at 9:56 AM, Mike Leisegang said:

Dermot, in the old days when you pushed film transparency  to a radical point, 3 to 4 stops you would - could start to see the DMax of the film showing.

prolly around the same time (early 2000's)Technicolor had a process called Dmin/Dmax that one lighted a roll with a cineon-ish encodeing, based on density between the sprokets.. all from memory and i never transfered a reel this way, but i have graded a few features (on Cyborg2k) from rushes transfered to DPX with this method, worked well for the time given the priceing of 2k scans was still by the frame mainly for VFX, and it was rare for anyone to be doing full DI back then, the Dmin/Dmax made it afforable

every print lut i've seen swings to warm as it assume a 709 viewing enviroment, and try to replacte the white point of a film projection on a d65 / 709 screen

i think Jussi is correct - it's the gain ball mapped to two axis, or as Resolve has those controls implemented they can be replacated with the gain hue offset ball

another option that i use alot in Resolve is setting a node to L*a*b and bypassing channel 1 so you have only *a*b, for me that's the simplest cleanest workflow, but be aware that the surface will react more like it's in 1998 mode

Baselight's tint/temp controls are somewhat diffrent btw, and really feel closer to L*a*b than to Resolve's gain offset method

true color temp is more a horseshoe shaped matrix than a straight line from blue to yellow

select ACEScct under pref's / color management

select project IDT if aplicable (ie; not a raw project)

select ODT

bingo bango bongo

i think it's a quiz...

dmax = digital maximum, goes with dmin, combined a method of scanning flim for digital work on a telecine rather than a scanner whiel retianing all the information from the ocn

i remembered going through this years ago, and it was nothing like known 2383 emulation luts. I compared it to two luts that are un-encumbered by active IP both of them are solid...  i compared to a number of other 2383 luts from Resolve, Nucoda and Baselight, all very similar, as were the Luthier and CineByte lut's

from memory, there was some wonky maths in the shadows,red channel doubling back over top of it's self, and some jaggy steps in the skintones area

i rolled my own when i had them presented by a producer, was not crazy about dealing with the outfall of the maths in Koji

i'm rendering overnight, (feature film DCP has to be in LA tomorow, mix arrived at 4.30 pm.. sigh)  will try to squeeze in a test before client arrives tomorrow

i have the entire load of koji luts somewhere on my server, i can drop some over a grey scale in the next few days

a cleint bought them for me to use on his project, and we ended up not using them

a bunch of board mixes from DJ's in Ibiza & CBC music streams

i have had log encoded DPX arrive at my doorstep, not  clue how they got there, it's what the DiT created with DiT magic

no issues with the files or the footage, nice looking highspeed food shots, flying peppers and dropping  / bounceing msuhrooms in a studio, no issues with expoure or highlight rentention, but there was grownup behind the camera.

awesome article Andy!

i've been using L*a*b in Resolve for a while now, and more recently  in BLE -  both myself and my cleints and find the resualts to bring us close to our happy place quickly

a questions for you if you have the time;

where are the diffrences between using R-Lab and Basegrade?

mainly the 4 zones and interactive surface mapping?

or are the maths substaintily diffrent under the hood?

next - a question about the article it's self, when you talked about Filmgrade + exposure you mentioned exposure not being truly linear due to input log curves, when in AP1 with raw sources is exposure also non-linear?