Skip to the end of navigation

Welcome to the Community!

F5 & F55

Showing results for 
Search instead for 
Do you mean 
Reply
Expert
Posts: 2,490
Registered: ‎11-21-2012
Message 11 of 32 (1,538 Views)

Re: F5 F55 & Alexa

Haha, clever-cloggs!

Ah, oh well!
Expert
Posts: 416
Registered: ‎04-03-2013
Message 12 of 32 (1,527 Views)

Re: F5 F55 & Alexa

My first impression from these still images, based solely on the amount of light captured in the darkest regions of the image, is 1) Alexa, 2) F5 and  3) F55.

 

If you apply the cheat, logically, from the other guesses, I think it comes out the same.  1) can't be the F55 because it was the same on both guesses. In the one guess that got one right, switched only the F5 and Alexa, but only one is right. So, it must be the F5. Therefore, F55 and Alexa must change position, and match my first impression.

 

I suggest that much of the "noise" in the Alexa image are artefacts resulting from uprezing to 4K.  But, the F5 image is amazing as a still, in some ways, the best image.  However, I like the colour much better on the F55, with the Alexa next; that is, judging from these images.

 

And, I agree, the images from any of these cameras should mix with the others without much effort in post, especially for a 2K release.

 

But, for 4K HDR and motion handling, the F55 is way ahead.  Of course, I must declare my bias, as an F55 owner.Smiley WinkSmiley Very Happy

 

Best regards,
Will

Expert
Posts: 344
Registered: ‎02-13-2013
Message 13 of 32 (1,512 Views)

Re: F5 F55 & Alexa

Congratulations Will, you get the brass ring!

 

FYI, the noise in the Alexa is about the same before scaling to 4K. With the F55 at 1250 as a benchmark, I would say the F5 is half a stop faster, and the Alexa half a stop slower than the published specs.

 

I wouldn't take too much stock in color differences either, as my grading leaves much to be desired. I get a little different result each time I try a new version.

 

I already knew that shot properly the F5 and F55 could produce stellar images before this exercise, but was a bit surprised by the Alexa. No doubt, continuous praise raises unrealistic expectations.

 

These three and the other RAW frames from the camera test can be downloaded from CML for anyone who wants to give them a go: UWE Camera Test

 

Cheers.

Noel Sterrett
Admit One Pictures
Expert
Posts: 2,490
Registered: ‎11-21-2012
Message 14 of 32 (1,510 Views)

Re: F5 F55 & Alexa

I must confess, when I looked first, I was looking on a small laptop screen... now I look at the images at their native resolution (on a 5K iMac) I am seeing things differently.

#1 must be the Alexa - I didn't notice any of that noise before. Amazed at what I'm now seeing. It doesn't look like digital noise though... almost looks like soft film grain added in post!

In the face I struggle to see differences between #2 and #3, but on the charts I can.

I'm really torn, but I'd guess #2 is the F55 and #3 is the F5. I know my reasons for choosing that... but I might be interpreting the things I'm seeing incorrectly.
Expert
Posts: 2,490
Registered: ‎11-21-2012
Message 15 of 32 (1,508 Views)

Re: F5 F55 & Alexa

Ah, well done Will Smiley Happy
Specialist
Posts: 18
Registered: ‎09-01-2014
Message 16 of 32 (1,298 Views)

Re: F5 F55 & Alexa

Without reading the thread, my guess is -  1: alexa, 2: F5, 3: F55

 

But far out, makes you wonder, 1 and 3 are barely different.

 

:-)

Expert
Posts: 344
Registered: ‎02-13-2013
Message 17 of 32 (1,241 Views)

Re: F5 F55 & Alexa

Good guess Kirk, but I'm afraid Will got it first.

 

Cheers.

Noel Sterrett
Admit One Pictures
Expert
Posts: 416
Registered: ‎04-03-2013
Message 18 of 32 (1,203 Views)

Re: F5 F55 & Alexa

[ Edited ]

Hi, Noel,

 

Really, the appearance of as much noise before scaling? Hum, maybe the Alexa could benefit from an OLPF, or a better one if it has one.

 

If you zoom in, say by expanding the image about 400%, and look at the squares on the colour chart, you can see the "noise" creates a strong mottled effect on what should be a smooth even colour.  It is just averages out to be a near match to the F55 colour when viewed at normal magnification. Still, it is perceived as a noisy image.

 

I wonder if the dark and light regions in the Alexa image move relative to one another from frame to frame.  If so, it would create a stimulus similar in effect to film grain in the perception of the moving image.

 

That would go a long way to explaining why people, coming from a film aesthetic, might prefer the footage from an Alexa.

 

It plays to the illusion of resolution that is all part of that wonderful process we call the human sense of vision.

 

Best regards,
Will

Highlighted
Expert
Posts: 344
Registered: ‎02-13-2013
Message 19 of 32 (1,171 Views)

Re: F5 F55 & Alexa

Will,

 

I chose the three shots with the cameras set at the manufacturers' ISO specifications, and with as little signal processing as possible, i.e., RAW. So in one respect, it was a test of the accuracy of the manufacturers' ISO spec, and not everyone agrees on the correct way to determine ISO.

 

Another approach would have been to pick the exposures where noise level for all three is similar. That would more accurately determine the relative sensitivities of the cameras. As an example, in other shots from the same test, I found that the F5 more closely matched the noise of the others when underexposed half a stop.

 

So while I understand your point, I think that the Alexa spec, as compared to the F's, is just a bit too generous.

 

As for Alexa's success, I think some of the main reasons include:

  • Arri was a well respected a high quality film camera (cinema rather than broadcast) company.
  • Arri color science was initially developed with their film scanners, and was incorporated into the Alexa.
  • Arri limited the operators gamma/gamut choices.
  • Arri established a workable workflow.

 

As a result, the Alexa's "look" was both consistent, and good out of the gate. All of which gave Alexa the head start. However, it is now clear to me that other cameras have caught up, and/or surpassed the Alexa.

 

Arri sites "large photosites and a Dual Gain Architecture" as the principal reasons for the sensors "breakthrough performance" : Alexa Sensor

 

While it is true that "The larger a photosite is, the more light it can capture and the lower the noise.",  as the test demonstrates, the Alexa sensor (~7.5um) is less sensitive than the F's somewhat smaller photosite sensors (~5.9um).

 

As for color, I think that has caught up as well, except that with the myriad of gamma/gamut choices available with the F's there can be no such thing as a consistent "look". It took me a very long time to work through the endless possibilities to get to the point that I can consistently produce the "look" I like. I've also discovered is that CMOS sensors are remarkably alike. Much of the difference between cameras is in the digital signal processing applied. Lots of great information about sensors on DxOMark.

 

If you start with their RAW output, a great many cameras can be graded to match.

 

Cheers.

 

 

 

 

 

 

 

Noel Sterrett
Admit One Pictures
Expert
Posts: 416
Registered: ‎04-03-2013
Message 20 of 32 (1,160 Views)

Re: F5 F55 & Alexa

[ Edited ]

Hi, Noel,

 

I think you have it right on as far as how Arri was able to develop a consistent look.

 

Indeed, it is derived from a "film" aesthetic via their heritage in film camera and film scanner technology.

 

I was only speculating as to why the Alexa look is successful from an aesthetic perspective.

 

You are right in your explanation of why it is successful in a pragmatic perspective.

 

If the blotches of colour in the Alexa image, which we are referring to as noise, are behaving in a manner to create the same, or similar, visual stimulation as film grain, our brains are processing the "moving" digital images from the Alexa, in the same way they process film images.

 

I won't go into a long explanation of what I mean about our brains' image processing, as a great deal of my understanding of it is intuition and speculation.  Suffice it to say, the timing of our brain's image processing is not clock-based, like our digital image processors in our cameras and computers, but light intensity-based.

 

The proof of this latter statement can be seen in this little experiment you may have experienced in a high school science class.  That's where I saw it for the first time.  You will need a pendulum and an ND filter.  Mount the pendulum and set it to swinging.  Now, view the pendulum with one eye without and the other eye with the ND filter.  Be sure the ND filter covers the full field of vision of that one eye.  You will observe the perception of two pendula swinging out of phase with each other.  The only explanation for this illusion is that the brain is processing the images from the separate eyes at different rates.  The lower intensity image is processed at a slower rate that the full intensity one.

 

As you can see, the discussion could get very long to explain how our brain perceives and distinguishes static and moving images.

 

But, I just want to make a distinction between two types of noise.  For one type, there is the noise we are observing and discussing in the Alexa image, variants in the data that are within the range of intended/captured image; it is not detrimental to the perception of the image.  The other type of noise is where variants in the data exceed the parameters of the image. The latter type results in degradation of the perceived image.

 

Also, you are right on, in your second approach to assessing the accuracy of the sensitivity designations.  Each company has decided for themselves what is an acceptable level of noise in low light areas of the image in determining their designation of their chips' base sensitivity.

 

Best regards,
Will

Get Social

Share your ideas Watch YouTube Support Videos follow us on Twitter Visit us on Facebook