DAVINCI RESOLVE - MOVING ON FROM LUTS?

[morgan_moore]

Moving on from LUTs.

As per some other threads fold have been suggestingthat CST is the way it should be and LUTs are too fixed being tabular rather than formulaic.

How?
Why?

 

[CharlesPapert]

and also:
What?
(check your thread's name, Sam!)

 

[morgan_moore]

Yep I broke the thread name. Its supposed to be DAVINCI RESOLVE - MOVING ON FROM LUTS?

Colour space transform. Take the log from the camera and applies a lood 'red, arri, canon'

So here is a shot with canon log 3 on the R6

One has the canon log 3 Colur Space Transform and the other has a Canon LUT (top?), which has been my base to grade from. The CST bottom is clearly niceer ??

 

[AmbiSonde]

There's nearly always a slight difference between camera manufacturer approved LUTs and Davinci's CSTs but they are quite trivial.

I have got complete confidence in CSTs so I let Resolve manage mine behind the scene by using Davinci YRGB Color Managed colour science. Resolve will interpret from the footage which input CSTs to be applying, 99/100 it guesses correctly on the odd occasion just use the Input Color Space setting to select the correct colour space. Resolve applies the output CST based on your output colour space setting in the Project/Timeline colour management settings at the end of your node chain, that includes the Timeline Node space if you use that for a look across the whole timeline.

Just letting Resolve do thing donkey work means files arrive in a pleasing place to start grading, in Sam's example above the CST is a much more pleasing place to start and closer to where we'd want the image to be before adding looks over the top. It's also ready for air in a quick turnaround situation.

For those new to CSTs they are 32 bit floating point math functions rather than a tabular list of discrete values as in LUTs. If you do use Look LUTs and it's has been designed to work with a different camera space or different colour space than you are working in then you can use a CST sandwich to convert to the LUT space then switch back to your working colour space. For example if you've changed your camera but had some favourite LUTs you used with the old camera it's a trivial process to get those LUTs working again with CSTs.

 

[jmone]

I'm a CST convert (pun intended!) and using Wide Gamuts for grading (like DWG or ACES). LUTS for me are now in the "creative" space only. Such an LUT example would be Cullen Kelly's Kodak 2383 print stock for ACES or DWG - Kodak FPE Freebie (procolor.ist)

 

[AmbiSonde]

I'm a CST convert (pun intended!) and using Wide Gamuts for grading (like DWG or ACES). LUTS for me are now in the "creative" space only. Such an LUT example would be Cullen Kelly's Kodak 2383 print stock for ACES or DWG - Kodak FPE Freebie (procolor.ist)

DWG has been life changing and Cullen's 2383 is the best example I've come across. 2383 is getting a bit overused as a look but mixing it in at 20%-30% tames a lot of the digital look of modern cameras quickly and easily.

 

[morgan_moore]

So I made a bad video this morning.. and CST has become my my goto - so thanks Ambi who has pushed me that way.

I need to test some really bad exposures befre being a full convert.

Maybe I willll BLF (bracket like f..) a test card and then gain more unsderstaing.

And also try with my C200 the camera I use less but care about more

S

 

[morgan_moore]

AMbri made a comment about a red.

With the LUT - Im finding that when I push the sat a little the visual saturation goes wild.

small sat boost..
not much sat -> correct sat
a bit of sat -> way too much sat

I guess that is the tabular mathematics that CST dont rely on.

 

[jmone]

To me the best thing about CST and a Wide Gaumt in resolve is that various input footage is "normalized" automatically into a very neutral look so you can creatively grade to your output format. It's even easy to produce both an HDR and SDR version for example.

 

[morgan_moore]

Yes it is good to normailse - the R6 H265 does not do automatically.

But if you fill the WFM when shooting (in log) the normailazation process hits the contradiction..

-i must add contrast and sat
-i must not induce clipping

Which means introducing and rolled off 'knee' (putrid) or allowing highs to clip (fine if you can recover them with a secondary)

Coming from stills colour im used to recovering with a secondary (burning the sky)

Its that testing of the extremes that I must do.. unless fold can link to YT or other sources.

 

[morgan_moore]

This is such an obvious concern that any 'tuber that doesnt instantly bring is up is weak. Probably why I shyed away from aces for years.

 

[AmbiSonde]

AMbri made a comment about a red.

With the LUT - Im finding that when I push the sat a little the visual saturation goes wild.

small sat boost..
not much sat -> correct sat
a bit of sat -> way too much sat

I guess that is the tabular mathematics that CST dont rely on.

The CST is also probably taming things in the Tone Mapping Method area. The Davinci default setting tempers the worst instincts of the footage usually.

 

[AmbiSonde]

Coming from stills colour im used to recovering with a secondary (burning the sky)

I'm shooting a lot of landscape video for a project and the HDR wheels are just perfect for teasing out contrast in the monotonous seas of green of grass land and making a tree canopy appear more 3d. The images have lost contrast in the process of uploading here but you get the idea.

**I deleted the images because they were flattened out by either the forum or the utility I used to recompress them and it was peeing me off that they looked that crap. I suppose this is the wild west of browser colour management.

Looks like the convo has come to a screeching halt anyway so, I'll see you around.

 

[morgan_moore]

So (and Im in a hurry to deliver of course) .. im struggling here

This is my canon raw, with colour managed the footage is applied with a hard look (too hard) inside resolve and when exported im just seeing what looks like untouched LOG..

Due to time restrictions Im gonna run back to LUTs? !