DR ..a good discussion!!

[David J. Buchanan]

Monstro Hard Light retry. Still, however has "hotspot" on the wall behind Luigi's head.

[/url]A001_C007_0113Q5.0000002F by rand thompson, on Flickr[/IMG]

[/url]A001_C007_0113Q5.0000002F by rand thompson, on Flickr[/IMG]

The biggest issue is the nose gradient, which is why it seems the Gemini has more DR. Could be because of the pixel pitch as well. Unsure.

 

[rand thompson]

David,


That gradient on the bridge of the nose mainly came from recovering the right side of his face. I'm sure I could have recovered the right side of his face a little less for a more even transition between the right(brighter) and Left (darker) sides of his face. It's a shame you have think about something like that though. Plus I think I remember adding a 65 from 50 sharpening in Redcine-X, which could have exacerbated the issue further.


You could however add a "blur" in a "power window" in Resolve and just track it.

 

[David J. Buchanan]

David,


That gradient on the bridge of the nose mainly came from recovering the right side of his face. I'm sure I could have recovered the right side of his face a little less for a more even transition between the right(brighter) and Left (darker) sides of his face. It's a shame you have think about something like that though. Plus I think I remember adding a 65 from 50 sharpening in Redcine-X, which could have exacerbated the issue further.


You could however add a "blur" in a "power window" in Resolve and just track it.

In the grades, have you found it easier to recover the Gemini more so than the Monstro? ... or even the Venice over Monstro?

 

[rand thompson]

David,


I've found Gemini, in low light mode, easier to recover in both highlights and shadows than Monstro. Gemini in "Standard Mode" not so much. The Sony Venice has better "Highlight" recovery tools in its Camera Raw tab which helps it get better highlight recovery than Monstro, but maybe slightly better highlight recovery than Gemini. The Sony Venice has a "shadows" function in its Raw tab that's very helpful. The Venice has the best looking shadows( least amount of purple/magenta noise), but has the most noise earlier in the ISO range than the Monstro or Gemini, the gemini having the least amount of apparent noise IMHO.

 

[David J. Buchanan]

David,


I've found Gemini, in low light mode, easier to recover in both highlights and shadows than Monstro. Gemini in "Standard Mode" not so much. The Sony Venice has better "Highlight" recovery tools in its Camera Raw tab which helps it get better highlight recovery than Monstro, but maybe slightly better highlight recovery than Gemini. The Sony Venice has a "shadows" function in its Raw tab that's very helpful. The Venice has the best looking shadows( least amount of purple/magenta noise), but has the most noise earlier in the ISO range than the Monstro or Gemini, the gemini having the least amount of apparent noise IMHO.

I see. So the Venice needs proprietary software to even manipulate the image. Correct me if I'm wrong, but it sounds like the Gemini has the most well-rounded image in your opinion?

 

[Adrian Jebef]

Rand is only using Resolve's tools in developing the Venice images. You do not need proprietary software though different software can sometimes make a difference in how the image is finally developed. It is best to test.

 

[rand thompson]

David,


Adrian is right. I think my use of the word "tools", might have thrown you off little, my mistake in nomenclature , "Shadow" and "Highlight" are just a sliders like Iso, Tint, Kelvin...etc. Also as Adrian said, results may vary from grader or user so it's best to do your own tests if that's at all possible.

 

[rand thompson]

Adrian,


Also, truth in advertising, along with the Sony raw tab, I am using some Slog-3/S-Gamut3.Cine, Tethrahedral, 65 cube sized Luts, Which use Custom Curves I made, for a buddy of mine a year or so back for I think his Sony A7s using "LUTCalc". I recently noticed that they were still in my Luts Folder.

http://cameramanben.github.io/LUTCalc/LUTCalc/index.html

 

[rand thompson]

I Re-did the Hard light grade in RedCine-X.


Here is the .RMD below. Most of the grade is under the "CDL" Tab


Luigi Sofa Monstro Hotspot Grade


https://www.dropbox.com/s/221j51y5h58ov3j/luigi sofa monstro hotspot.RMD?dl=0

[/url]A001_C007_0113Q5.0000002F by rand thompson, on Flickr[/IMG]


Here is the Epic-W Gemini in Standard Mode with the same grade applied

[/url]A001_C008_0604CM.0000081F by rand thompson, on Flickr[/IMG]



Edit:



You can get rid of the gradient on Luigi's nose to an almost undiscernible level on the Monstro image if you make theses changes to the grade;


change the


1) 3200K to 3000K
2) Uncheck the "Unsharp MasK" box under the "Post Image" Tab

 

[Kemalettin Sert]

I just cant believe Red still cant beat sony cameras DR wise thats so sad

 

[rand thompson]

Kemalettin,


Here is the original Sony Venice image before highlight recovery, as you can see it faired better in the highlight than Monstro and Gemini(In Standard Mode).

Here it is recovered to the Max "-100" "Highlight" setting in the Sony "Raw" Tab

I think Red chose the right approach, Focusing more on Lowlight and less so on Highlights. You can use a ND Filter or more diffused lighting to hold on to more highlight info., but being able to pull up less noisy Shadows that keep more info. is something more desirable IMHO.

 

[Kemalettin Sert]

Rand what about Phil rating monstro 17+ DR and Sony rated their venice 15+?

 

[rand thompson]

Kemalettin,


I'm shit with Xyla charts, so I will have to defer that subject matter to someone who is much more knowledgeable than myself. I don't want to pretend to know something I don't

 

[Sabovic Adis]

Rand what about Phil rating monstro 17+ DR and Sony rated their venice 15+?

Don't know about Phill's findings, but here's what I've found:


Venice LL; Shadowdetail (curtains, seets) but blown highlights (lighting, screens):

Venice STD; Highlights recoverable but man, shadows are so gone:

And Monstro; No blown highlights plus detail in the shadows at the same time, yet less noisy than that of Venice LL:

Thanks again, Rand, for supplying me with the TIFFs. :smile5:

 

[rand thompson]

Sabovic,


No problem man! glad I could help!


As you stated, the shadow of the Sony Venice, when pushed equally with Monstro and especially Gemini in Low Light Mode will fall short. Also as you stated, the highlights are recoverable especially when raw files are used.

 

[luigivaltulini]

Better up or better under ehhe
Monstro bring home the best compromise without any switch.
This for me is already a considerable advantage.

[/url]comparison by luigi valtulini, su Flickr[/IMG]

 

[Michael Tiemann]

Rand what about Phil rating monstro 17+ DR and Sony rated their venice 15+?

Let's again look at what this really means (which, again, means referring to what Ansel Adams teaches in his series, The Camera, The Negative, and The Print).

Whatever the dynamic range of the sensor, its white point and its black point need to be tied to the white point and the black point of the display device. If the sensor can only capture white or black we could call it one stop of dynamic range (counting from zero) or two stops (counting from one). If the display device is a typical REC 709 display with 80-120 nits of white output (in a darkened room) and a 500:1 contrast ratio, is the result then reported as 9 stops of DR (from display white to display black) or 1 stop (from source white to source black)? And if there's some amount of sensor noise that finely interleaves black and white pixels, does this give us an extra one or two stops of measurable sensor DR as intermediate brightness gradations can be represented by clusters of pixels that are either 100% black, 100% white, or some ratio in between? (Note that the quantum nature of sensors is such that it's not as if a source that's 80% white will get 80% of the pixels to say white and 20% of the pixels to say black. Above or below a certain threshold, almost all the pixels are going to see things exactly the same way. At the 50% level, the pixels are going to be most open to the pure chance of whether they read up to white or down to black. The lower the sensor noise, the sharper that transition.)

So the first thing to keep in mind is that whatever the sensor's DR, the top stop and the bottom stop have to be tied to the display's top stop and bottom stop.

Now, in between those two we can look at how well the sensor's DR maps to the display's DR, and why not all stops are created equal. When looking way to the right of the XYLA chart luma values (preferably in a waveform monitor), we see that there's very little (or no room) for meaningful intermediate values to exist. I previously talked about digital "eye" charts, which show (or don't show) the decisive space between reading something as a 1 or reading something as a zero. If the noise of the 1 signal overlaps the noise of the 0 signal, there is no "eye", which means that there's no real discrimination possible between signal and noise. Of course one can draw a digital line in the sand and say "everthing above this line is a 1 and everything below this line is a zero", but that means that up to half the time we'll be rendering a 1 as a 0, and vice-versa. And all those guesses will lead to a dithering of 1s and 0s, creating a kind of digital gray. That gray will be 50% gray if the 1s and 0s are full scale, and it will be a much darker gray if the 1s and 0s map to the bottom limit of the DR of the display device.

If we pick a chip where the eye is 6x the height of the 1 noise and the 0 noise, then there are 6 distinct gradations of values that can be teased from the sensor between the 1 value and the 0 value, 8 levels if we include the values of the 1 and the 0. That's 3-bit color! Now, not all part of the image need to be rendered in fully expressed 10-bit color gradients to look good. But if the blockiness of 3-bit color is going to be a problem for a given part of the tonal range, then even if we have the DR to distinguish one stop from another, the image rendered with that stop can fail subjective tests (blockiness in the shadows). So we must ask: how much do we have to crush the shadows before the limitations of gradations disappear? Alternatively, what chip of the XYLA chart gives us an eye that's large enough to hold enough gradations to render acceptably given the noise measured at the top and bottom of those scales?

The same calculations can be done in the highlights: when does blockiness in the highlights become a problem, and how large is the eye (vs the noise) large enough at the top level to deal with that?

Absolute DR is one interesting metric. Useable DR is another--and I think more important--metric.

Finally, in the same way that we might throw away pixels to reframe images (especially when stabilizing gimbal shots), we might use curves to push the levels (or the colors) in one direction or another. That comes out of the DR budget! If you never, ever color correct, then you have the full DR of the sensor/display system to play with. If you make adjustments that could be +/- one stop in one channel or another, you've just cut 1-2 stops of DR out of your sensor's effective DR.

So in reality, useable DR is calculated first by subtracting the grade DR from the sensor DR, and then looking at how many bits of luma/chroma smoothness you need at each level of display DR. You can test whether your can map the remaining sensor DR to the display DR range, subject to the gradation smoothness you need. If you are lucky, you have a few stops of latitude in case your original exposure was off by a stop in either direction. If you are professional, you budget in those stops up front.

I think that one of the biggest pitfalls I've seen here on REDUSER.net is people giving too many jobs to the same DR buckets. That is to say, if you loosely say "I'm going to allocate one stop for highlight protection and two stops for digging into the shadows. And I'm going to allocate two stops for taming highlights in the grade and three stops for lifting shadows in the grade." And then you assign those roles to the same DR buckets, then the minute somebody underexposes something, those shadows you thought you could lift are going to look UGLY. If you think "well, I cannot underexpose and overexpose a shot at the same time, so my -2 and +3 protections really just amount to 3 stops of total DR reduction" then you are not truly protecting both ends of your exposure. You are protecting one side or the other, or both sides half as much. Once you are clear about the actual role that each bucket plays in the overall production pipeline (which is the key message that Ansel Adams brings home again and again when he writes about the Zone System), then you can be assured that if you do your homework during acquisition, you will get excellent results when it comes time for exhibition. There might be a way to characterize a weighted sum of SNR at each DR level to produce a more meaningful measurement of sensor DR than just "how many chips can you see?", but I don't know of any.

 

[Patrick Tresch]

So the first thing to keep in mind is that whatever the sensor's DR, the top stop and the bottom stop have to be tied to the display's top stop and bottom stop.

Yes, if you have an 8bit display you could effectively only diplay values of 256 different stops. Now one measured light stop wouldn't match one light stop on the diplay.

 

[Björn Benckert]

Rand what about Phil rating monstro 17+ DR and Sony rated their venice 15+?

Writing that it sound like you mix up light sensitivty and range. if both sensors are exposed to their cliping point monstro will reach 2 stops deeper into the shadows. Also monstro is more light sensitive so, yes it will clip before sony. Now light sensitivity is normally a good thing. Sure a "highlight" mode on monstro would be great but thick ND´s will do the same trick.

 

[Fabricio Morato]

It is also curious that Panavision rates the DXL2 a 1600 native ISO. They probably feel confortable with the amount of noise at this ISO while also gaining 1stop in the highlights compared to Red's recommended ISO 800 (though they do not state anywhere I could find what they consider the native ISO).