Full Frame Cinema Shootout 2020...

[Noah Yuan-Vogel]

NO.

On both accounts.


It wouldn't.

Noise performance and highlight handling are two completely different sensor properties.

Ah you must mean absolute (scene referred) white clip point? Never heard "superior highlights" used that way. That is easily solved by closing the lens slightly or adding a little ND...

[Hrvoje Simic]

The Venice transform would ideally have been s709 LUT. It was introduced for the Venice. The 709 LUT is Sony legacy.

"Ideally"...as in the context of the test.

Ideally...nope. But that's for another topic.

Ah you must mean absolute (scene referred) white clip point? Never heard "superior highlights" used that way. That is easily solved by closing the lens slightly or adding a little ND...

You are talking about the image. Not the camera properties.
Camera properties are fixed and exposure determines what you put inside that usable range. And highlight handling. Defined by sensor properties.
Yes, scene mid-point is a reference to be able to see how much there is where, chips on a DR chart in vacuum are not. Digital ISO equivalent image push algorythm changes nothing with camera properties, it is image enhancement mimicing gain.

[John Chater]

[QUOTE=Hrvoje Simic;1889115]"Ideally"...as in the context of the test.

Ideally...nope. But that's for another topic.


Ideally instead of 709? Yes if you are looking for the manufacturers camera baseline..

Tests are complicated, hard work and always involve compromise as Im sure you know.
Please share your ideal way to do this.
Sincerely. Im always looking to learn.


John

[AndreasOberg]

Interesting test.
A few things I noticed:
- Monstro looks great
- Arri is very good, even though it has too low resolution for our needs, I was surprised how well it did when underexposed. From other tests I thought it would get a lot of noise, why is this not the case?
- Canon C500 was surprisingly poor.
- Sony FX9 surprised me. Both with how sharp it was, but also how good colors it just looked. Of course it has other downsides like only 4K for now etc.
- Gemini was very good in low light, no surprise there.
- Not so much to say about Helium which is still our go to camera, soon getting another one.

Andreas

[rand thompson]

Interesting test.
- Arri is very good, even though it has too low resolution for our needs, I was surprised how well it did when underexposed. From other tests I thought it would get a lot of noise, why is this not the case?Andreas

Andreas,


I asked Evin a similar question and this is what he said.

All I did was increase exposure in the RAW controls in Resolve, I think the higher key of the studio test masked the noise a little. Also this being the LF the magnifications is a bit lower compared to the Plus/mini....

[Hans von Sonntag]

Hans...you are not judging just the camera, but the camera + transformation.

: )

Hrvoje, I'm only judging what Evin shot and shares. I'm sure that this test has some value but, of course, there are many steps in a workflow that can be altered while conducting such an analysis that will, in the end, lead to different conclusions. But I'm convinced that Evin did all in his powers to make this test as objective and meaningful as possible.

Because this test is what it is - an experiment - I used the phrase "regarding Evin's tests" when I was mentioning the Venice. As you said, there are many aspects, workflows, transforms and whatnot that will lead to a result. This test is a real-world video test in video-land and, by that, I'm assuming Evin did all the work in Rec709 colour space. He probaby used regular/video tools that people have commonly at their finger tips when shooting with said cameras and eyeballing the pictures on their video devices.

But I know from some experience that quite a lot will change if you take the ACES route. So all of this has to be taken with some grain of salt beginning with ISO numbers and ending with LUTs and the normalisaton of the results based on the hope to give us an idea what which camera actually is doing.

In the end, there is no substitute to conduct your own test. We all know how subjective all that is. Even the Alexa's oustanding over- and underexposure performance will be rendered pretty much moot if the competition's cameras are well exposed. There are a myriad of reasosn why not to use/buy and Arri camera although we all know an Alexa delivers the best picture - what ever "best picture" means.

[Noah Yuan-Vogel]

You are talking about the image. Not the camera properties.
Camera properties are fixed and exposure determines what you put inside that usable range. And highlight handling. Defined by sensor properties.
Yes, scene mid-point is a reference to be able to see how much there is where, chips on a DR chart in vacuum are not. Digital ISO equivalent image push algorythm changes nothing with camera properties, it is image enhancement mimicing gain.

So that's a yes then? You are talking about absolute (scene-referred) white clip as I suggested? So is your issue my describing it in photographic terms instead of sensor terms? Even if they mean the same thing? Sounds like your issue is that I presume there must be a lens on the camera because it is a camera. Whether the light hitting the sensor passes through a lens or not is essentially irrelevant as the clip point as it relates to other cameras is the same so it seems like an odd thing to get hung up on.

[Hrvoje Simic]

So is your issue my describing it in photographic terms instead of sensor terms? Even if they mean the same thing?

There is no issue over here. :)
We are not talking about the same thing. You are interpreting the image as camera performance. Image is the outcome of camera performance and exposure to scene. Which is why the same exposure and scene have to be the reference to be able to see at which point does each clip and how.

Camera has fixed response to light and highlight handing with clipping point, falloff, lattitude and many other properties, and numbers for rating and exposure are not arbitrarty.
With lower exposure you don't affect camera's highlight handling, you make the incoming light intensity lower so it doesn't hit the cameras light intensity capture ceiling.

[Ben Roper]

Why does the Arri (1) look totally fine 6 stops both under and over exposed and everything else looks like trash?

[Hrvoje Simic]

- I used the phrase "regarding Evin's tests" when I was mentioning the Venice. As you said, there are many aspects, workflows, transforms and whatnot that will lead to a result.

Yes, and with base colour science properties, image transformation properties you also have RAW development algorythm properties which people tend to take as is, assuming they just stand there neutrally and work perfectly.

Digital domain did not come with a bag of gifts only.
It also brought a bunch of data-centric illusions in the package.

All those RAW settings including ISO slider affect the material and they are also neither arbitrary nor perfect. They have their own properties and limitations. In case of Venice, I saw some things in RAW developing and ISO push which are not very flattering to the material. To say the least. Luckily, we have "offset" feature in Resolve which is least destructive to brighten the material.

So...with base colour science properties and LUT properties when you judge material push in this case (ISO slider in raw) you are also judging the quality of the gain mimicing algorythm.

That is 3 sets of factors at least on top of camera performance.

But I know from some experience that quite a lot will change if you take the ACES route.

No need for ACES for a dramatic change.

ACES also brings a few data-centric illusions btw.

In the end, there is no substitute to conduct your own test. We all know how subjective all that is.

Subjectivity is default state of mind, with preferred perspective, prism, point of focus and range of awareness.
If one really really really likes his glasses and view, subjectivity gets to stay locked down internally. If one doesn't fall love with default and seeks beyond, able to change the perspective and prism, there is room for knowledge.
And as knowledge expands, "subjectivity"
- in the context of personal bias, decreases.