Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

EPIC MX Highlight Rolloff thread....

Status
Not open for further replies.
I have this example:

A005_C012_0809PZ0000516_zps6a271ab3.jpg
 
Hey Daniel,

Can you provide the snapshot as a R3D or a RAW TIFF?

Thanks,
 
I'd like to see a snapshot, too...

The rolloff was the 1st thing I loved with the Epic.
Now I can overexpose all over again!

1146311_10151790929240126_1823780661_o.jpg

AAAAh man.
I was just waiting for Mr G to join the party. :-)
 
The highlight magenta is mainly due to metadatasettings, Daniel.

I did a quick development which is not focused on "nice", but "clean".

Works well...

What lens was this shot on BTW?
 
Ah yeah, we have a thread here!

The scenes in Elysium of Max's childhood (the first 10 min and beyond) looked like fine examples for consideration here, and in the optical v. digital thread, I think.

Bring on the high key GG!
 
Last edited:
It's kinda soft for that kind of shot it seems, and it could be some CA in the lens causing your issues, too (says the guy who gladly shoots with a 40'ies angie that cannot be sharp...)
 
I'm using a small package for this documentary I'm shooting in Denmark right now. It's not unlikely that the shot is just ever so slightly out of focus, as I'm going run and gun.
I'm also using a Tiffen screw on Pola that I'm not super happy about, but the fact of the matter is that I think the roll off is actually quite good. My example was on the plus side :)
 
Here's one example of a shoot I did a while ago, where I purposely overexposed the window, but you still have some detail on the blinds:
High1_zpscaebbecc.jpg

Sorry, don't have the R3D's on me...
 
OK...


Hahahahahaha


Running in to defend!

LOL

And I agree. The magenta or yellowish ways are coolr.

I'd like to warn you a bit on looks in cam, though...

But that is just a pet of mine! ;)
 
I think Antonio knows how to protect his highlights.
If I understand Antonio's point, he wanted to clip his chanels as opposed as protecting the highlights.
Cheers
B.

Brice, I might have explained it wrong, what I don't want is exactly to not clip. Clipping for what I understand is beyond pure white. It's burned pixels. It's bad. Pure white is just pure white, no pixels burned, you can grade this white any way you want. But you can't grade a hole full of dead pixels (clipped).

I thought it would probably be a good idea to talk about the ZOne System, I'm sure many of you know what this is. And maybe I can explain more clearly what I have in mind.

a943743c1fd8467c852cb4e779cc463a.jpg



As you can see, all that we need and can see is between zones 0 to X. What I think the problem with digital (or Red) is, it lets the sensor read past Zone X, into the clipping area. So instead of us having a clean white glaring sun, we have a 'hole' in our image full of dead pixels. This might not be a problem as for the naked eye it looks just like pure white, but it does often create artifacts and those artifacts will in turn create a bad rolloff. This wouldn't happen if the sensor would stop at Zone X, not allowing any more light to burn the pixels.

Now, some people here have been posting beautiful examples of glaring highlights, some I think that in many shots this might not be a problem as much of the image is covered in the flare from the highlights, creating a glow. But, if the image doesn't have any flare, or if that flare doesn't cover the entire image around the clipped highlight, then I think we start to see artifacts.

But I might be wrong about this. But by looking at your pictures that represent a good example of specular highlights against others that I have seen that have artifacts, I try and analyze and understand why.

Now judging by the pictures that Flip kindly posted about his tests with a Low Contrast Filters seem to really help about this issue. As I also seen in Robert's grade of my frame grab that he fixed in photoshop.

So, what I gather so far is, using a low contrast or a Black Pro Mist filter will really help to avoid artifacts in the rolloff. And exposure has to be dead set on Zone X of the light meter. Because the MX will not stop it for me at Zone X and will let light in until it creates clipping.



A little off topic, a little trick I learned with the wonderful John Blackmore on how to find out how much DR your camera has. It really is very simple.

Get a white shower towel (so the white has some texture) try that it is not rugged and as smooth as possible.

Find a place in your house (or outside) in the shade.

Place towel on a wall and frame the towel so that it fills the entire frame. Probably use like a 50mm for this.

Shoot the first image on middle grey. (Use a light meter for this.) Or use the false color and set your aperture so that the towel reads Grey on the false color.

Now shoot one stop up every time until all your frame has nothing but white. (you should have 5 or 6 shots)

Now do the same but stepping down, until what you are shooting is nothing but black. (If you're lucky you should have another 5 or 6 stops)

Now, bring that onto RedCine or whatever other app or camera you are using and review the shots. Count on how many shots you can see the texture of the towel, both up and down. And that is going to be the dynamic range of your camera. You'll know how many stops it can read into shadows and into highlights.

If you really want to get funky with it, try different ISOs. And see for sure what difference that makes.

You might find good or bad things with this simple but effective test...
 
Last edited:
Lowcon and diffusion always works. That is what is built into the alexa. And personally I prefer to control optical parameters myself.

But I see how those built in makes life easier... At first.

BUT.

Your post makes me think I should really write that "linearlight for dummies" article, which I constantly try to avoid writing.

AND a
How does an AD process work

article

Because with current tech and math, you are asking for something that does not exist.

Clipped is basically = white

BUT

If you have ever heard a digital audio recording overoading, you would know that digital clipping = bad.

So what one has is filtering that recognizes clipping as white.

Problem is, that you have three channels, not one.

And THAT causes most of the issues. One channel overloading while the others are still fine...

Or white-ballancing.
 
Lowcon and diffusion always works. That is what is built into the alexa. And personally I prefer to control optical parameters myself.

But I see how those built in makes life easier... At first.

BUT.

Your post makes me think I should really write that "linearlight for dummies" article, which I constantly try to avoid writing.

AND a
How does an AD process work

article

Because with current tech and math, you are asking for something that does not exist.

Clipped is basically = white

BUT

If you have ever heard a digital audio recording overoading, you would know that digital clipping = bad.

So what one has is filtering that recognizes clipping as white.

Problem is, that you have three channels, not one.

And THAT causes most of the issues. One channel overloading while the others are still fine...

Or white-ballancing.


I might very well be wrong about this, but this is where I think you are wrong. Pure white is not clipping. Pure white is as simple as you shooting a white piece of paper, and if you expose it accordingly it will be pure white. You can bring it to the grade and take to all the different greyscale levels until pure black.

Clipping is something else entirely. Clipping is dead pixels! Bring a frame onto your post app that has clipping and see what it does. I tell you, on the Red you will have a pink/magenta area when you bring it all the way down to black. It will not go through any of the grayscale levels, it will go from white (burned pixels, not pure recorded white) to pink!

This is where I think people get confused.

Pure white IS NOT clipping.
 
Nope.

I have shot a 4k HMI and the sun directly into Epic MX.

It definitely clips.

And it CAN look good, and it can look bad.

the magenta has allways been due to bad metadatasettings, though
 
The image I posted is shooting directly into the sun in a dark room, exposing for the skin.
It definitely clips! :)

Should clip, by any means.

And not a trace of magenta in there...
 
Personally I think the roll off is determined more by how you grade it.

9482278056_b9b224cbbd_o.jpg
 
Nope.

I have shot a 4k HMI and the sun directly into Epic MX.

It definitely clips.

And it CAN look good, and it can look bad.

the magenta has allways been due to bad metadatasettings, though


EDIT: Brice took offense in me posting an example of one of his shots being manipulated to show what a clipped highlight does to the image. So I took it out.



Of course it clips! But that is never a good thing! Because the difference between a clipped highlight and pure white is none visually. The difference is that you have an intact image without artifacts if you have pure white, but if what you have is a clipped highlight, you have nothing.

Of course not in all shots is this a problem. In your example, the entire frame is filled with a flare. BUT, if you wanted the sun in the frame without the flare being so big and have a more defined image, then I think clipping becomes a problem.

I understand that this is not a problem for you. But it is for me (to a lesser extent now from what I have learned and seen in this thread) and it is to a lot of people. And it is a problem with digital in general!

So, with a not clipped highlight you are able to bring that white value to whatever value you want until pure black. That's my definition of RAW and good grading material.

But, with a clipped highlight, that's all you ever going to grade, a clipped highlight. It will go from white to pink. There's no middle values. And that's my definition of a bad grading material.

Note, if David or Brice oppose me posting their frame grab here as an example after I completely fucked it up on Red CineX, I will take it out immediately.


Let's start with a normal white. This guy's shirt is white and is was not captured clipped. It's pure white.

53b560be060fef51b7c3d8374ee0ae6d.jpg




Middle Gray

2ef267ffe5f14bf069446445170a9171.jpg



Pure Black (well, just one step away, so there isn't just a black frame...)

add90161a3d905a2ebd745db3b8efed1.jpg




Now going up. I can make my whites go to much brighter values as well.

A bit brighter than normal, maybe Zone VIII.

c9c093499c03fe729983f8e6c5c6c604.jpg




Just one step away from all white,
a7964a71fefbe81c261a35ba909cfeee.jpg



So, no. Pure white is not the same as a clipped white.

You might be fine with that. But I'm not. Mainly because I don't think anyone ever needs a clipped white! So mind as well find a way of not letting the sensor clip so that people can light and capture perfect grading material every time.

From the examples I posted above, I really fail to see how having the option to clip is a good option or even an option at all....
 
Last edited:
Status
Not open for further replies.
Back
Top