Improving the Camera: The Image

[Hrvoje Simic]

Material from the last Friday shoot we did. I got the clearance.
DoP: A.S.



Camera: Ursa Mini Pro 4.6K
Lens: Zeiss ZE's

First and last image pairs are just test shots.

First shot I told the grandpa to stand there on the sun in completely uncontrolled light prior to the shoot so the DoP and me could see highlight performance of the camera and take that into consideration later for exposure decisions. Last shot with the girl at night we have set the camera on 4200K so we reduce the blue in the background from the HMI on the upper terrace and get warmer faces from 3200 light sources in the front + small fire on the left side.


First image in the pair: Primer 02 NTR
Second image in the pair: Resolve LUT for Ursa Mini 4.6


This is ungraded imagery, without any light and cast compensations done in post.
Just the shaping/transform tools applied. So first test shot goes to the upper border (deliberately) girl is under (DoP call) and reddish (camera temperature setup decision) and table should be a reference.



Ursa Mini & Mini Pro supports 17^3 cubes for on-board monitoring. On board LUT which comes with the camera behaves just like those in Resolve.

We used Primers on set and DoP exposed based on them.




Primer

Resolve Ursa Mini LUT

Primer

Resolve Ursa Mini LUT

Primer

Resolve Ursa Mini LUT

Primer

Resolve Ursa Mini LUT

 

[Wayne Morellini]

I like that resolve little, but the top end has to be compressed down a bit. With the level of high end you have achieved over the resolve, it is obvious. Don't they have controls in resolve to tone it down a bit. Is this Bayer corrected in resolve, or in camera resolve lut settings usually used with Prores? Definitely something useful as a starting point for .... Cameras like stills and low end prosumer. If only we could bypass internal color miscorrection more completely. Are there still efforts to use custom physical filters to correct camera misbalance. A combination of the two would produce more usable range on 8 bit codecs, which are two tight in sd, which means in HDR such a combination would better accommodate 10-12 bit footage. Maybe you should search for such a manufacturer to do a combination deal. Hmm, I'm just get a mechanism forming in my head to do on set color correction to establish look early. I wonder if color pullers in the old film days had something like this?

I basically completely turn off of all 8 bit 12 stop cameras due to vast degradation issues in footage, that despite rising stop levels, doesn't seem to go away (not funny). I presumed that the information was distorted or lost, even part of one chip demosiacing, and overzealous low pass filtering to make up for its deficits. However, you have shown me the information seems much more recoverable, and explains why the Arri can achieve such a good picture with 13-14 stops, which I expected. If only BM could get camera resolve to do this they could sell more 4k mini cameras. Even the 4.6k looks funky to me, and the Guardians of the galaxy 2 monochrome poster, which has the toning similar to the mini 4.6k. Thankfully we have helium now.

 

[Hrvoje Simic]

Added more rationale in previous post.

 

[Hrvoje Simic]

New Primers are online. Three new sets.

They are made for shooters, DIT's and colorists working on:

- Canon C300 Mk1
- Canon C100 Mk2
- Ursa Mini & Mini Pro

Canon sets are for "cinema" mode.

Ursa Minor & Minor Pro are for "film" mode.




Primers EQX designed for professional motion imaging cameras now cover:

Red Dragon
Red Helium
Sony FS7
Sony FS7 mk2
Sony FS5
Canon C300 mk1
Canon C100 mk2
Ursa Mini
Ursa Mini Pro



Have a great Sunday, creative folks...



http://omeneo.com/primers/

 

[Hrvoje Simic]

Sony FS7 / FS7 Mk2 / FS5 Primers are updated.
New set is online.


Links hit the emails over the weekend, I'm on a shoot until then.

 

[luigivaltulini]

based on Hrvoje Lut
Grading is purely artistic, but the basis for lut primers is just amazing.
I am very happy with these primers.


Dragon and dji phantom.
https://vimeo.com/214902231

 

[Hrvoje Simic]

Grabs from recent TVC we did.
Ursa Mini 4.6


DoP Mario T.


Mario used Primers in camera preview for exposure decisions.
I used them in post.





Resolve Ursa Mini 4.6 LUT

Ursa Mini Primer

Resolve Ursa Mini 4.6 LUT

Ursa Mini Primer

Resolve Ursa Mini 4.6 LUT

Ursa Mini Primer

Resolve Ursa Mini 4.6 LUT

Ursa Mini Primer

Resolve Ursa Mini 4.6 LUT

Ursa Mini Primer

...





Final grades using Primers as a basis:

...

 

[Wayne Morellini]

Looking better Hrvoje.

 

[Maira Fravel]

It does and it doesn't. All depends on what you are aiming towards.

Yes, I think its about playing with that line of subjective and objective reality.

 

[Joe Cage]

What is the difference between base and NTR, vivid and the others. Are they just preference?
Or should you use the base monitoring lut for example in low light conditions and then only use the base post lut? Or what is best practice with these?

Thanks,

 

[Waltraud Mowen]

Nice images. I love the shots!

 

[Hrvoje Simic]

What is the difference between base and NTR, vivid and the others. Are they just preference?
Or should you use the base monitoring lut for example in low light conditions and then only use the base post lut? Or what is best practice with these?

Thanks,

In a nutshell, specific signal distribution affecting contrast.
Not a simple contrast though.

I suggest NTR for monitoring in most scenarios, others per specific taste or need. They all keep the full DR but distribute it differently. For low light Base will show you more in shadows but Vivid for example will show you better how noisy is signal getting if you go under. You can choose all on preference but Push is...self explanatory.

In grading start from the one visually closest to your goal, you will get a "healthier" image.

 

[Wayne Morellini]

Thanks Hrvoje.

 

[Hrvoje Simic]

Summary of results.
On Dragon material.



FIRST image
Log3G10, RWG transformation to 709 with IPP2 LUTs


SECOND image
Log3G10, RWG transformation to 709 with Primers LUTs


Directly from Resolve. No color modifications.
In RAW selected ISO, temperature and tint.
On a few shots after the Primer upper end of curve was linearly brought up.



Many thanks to all who contributed and all who help the project.



Have a nice Sunday...

...

...

...

...

....

...

...

...

...

...

...

...

...

...

...

.

 

[Kwan Khan]

Thanks

 

[Mike Nagel]

red push in all skin tones

 

[luigivaltulini]

nice work MAN

 

[Hrvoje Simic]

nice work MAN

Great sample shots, Luigi.

 

[Hrvoje Simic]

red push in all skin tones

Nope.


It just may seem that way because of yellow skin examples affecting interpretation.

 

[Hrvoje Simic]

Important factors.


1) Make sure you check each example separately, without standard LUT examples in your eyesight.

They will affect your colour interpretation. Been there myself in the start of this research. And got back many times to re-check measurements and signal transformations. Let your eyes adapt for 10-20 sec and feel the image without the conditioning.


2) What you see in second examples is practically referent colorimetry .

I use "practically" because mathematically it can always get more precise but past a certain point it is not visually distinguishable. If you were standing in front of these scenes these images seen on a 709 calibrated OLED would coloristically match what you see in front of you.

3) Contrast distribution part varies, and the choices I made there were done with regards to grading, DIT and on set preview and practical first light, all taking into consideration signal "health", which is why there are more versions for each main route.

When you apply these and compare them to standard ones, they seem a bit darker. That is normal.
This is more referent, typical LUTs overboost the signal.

Mildest and more conservative versions allow more grading range, more defined ones practical dailies, on set preview and quick route for single shooters working on their material. All improve the colorimetry and overall signal distribution and keep the full DR. What you see is what is there. What you don't see, don't expect to find later and bug the colorist to squeeze more. If you want a healthy and a well defined image.


4) To stress again...these image optimizations are done within the last image transformation part before the post and done on Red's and Graeme's base IPP2 Log3G10/RWG mapping color science.

This is NOT "undoing" of it, this is a complementary optimization process made later in the signal chain.



5) These are designed per sensor response + color science signal mapping so they are "sensor referred" unlike "scene referred" type of image transformations.

Meaning they are scene independent.

Just make sure you set correct temperature and tint in raw and colours should sink in.