Welcome to our community

Be a part of something great, join today!

Log in
  • 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 :)

DR ..a good discussion!!

Venice,


40821028220_3a90568841_k.jpg
[/url]Venice guy in chair red bkgrd_1.4.2 by rand thompson, on Flickr[/IMG]



40821025360_fcea40ecce_k.jpg
[/url]venice guy in chair red bkgrd_1.4.1 by rand thompson, on Flickr[/IMG]
 
Thank you all for contributing to this thread!
This is the most fun I've had since reading the SALT VV lens comparison late last year.

I will likely be upgrading my S-W to Gem sooner than later. As a special request, Can someone shoot a quick XYLA with a Scarlet-W or point me in the direction of a thread that has one attached?
Thanks in advance.
 
rand thompson said:
Sabovic,

It's about time you showed up !! Ha Ha
Click to expand...

Yep, sometimes I even surprise myself. :rofl:
Thanks, bro!:cheers2:

What video codec is in them .mxf files? Sony RAW? :scared: Can't import in my NLE coz codec is not being recognized, fuke...
 
Sabovic,


Ha.Ha, It's a .X-OCN ST/LT file, Davinci Resolve supports it...at least Resolve Studio does.



Screenshot_35.png
 
Last edited:
rand thompson said:
Sabovic,


here's a dropbox link to .tiff images directly out of Resolve Studio for each of the Venice files. There is one labeled "gemini cellphone..." ignore the label it's a venice .tiff image. Each is about 139mb each and have a resolution of 6048 x 4032.


https://www.dropbox.com/sh/8gpkwzhgv4xo39y/AAAFo-oZOrhQgtZYdclf2ulOa?dl=0
Click to expand...

Wow, thanks a bunch, Rand!

Here, same shots as before, but this time with different contrast/color results, as I'm not that familiar with slog3/sgamut3 footage. Looks kinda WBd, compared to RED's:

87TLMmr.jpg


9sOri9H.jpg


r4auDO2.jpg


Grades very nice, thaough. :smile5:
 
OK...some new thoughts.

1. The "EYE"s have it. For those of you who have looked at digital traces on oscilloscopes, you know what an "eye pattern" is:

Figure%201%20Eye%20diagram%20analysis.png


When the "eye" is open, it means there's room to distinguish signal from noise. When the eye closes, the electronics can't really make the distinction, even if we convince ourselves that we can still see the differences.

2. Here's a screenshot of all six waveforms (Monstro, Gemini 800, Gemini 1600 LL, SONY Venice 500, SONY Venice 2500, Alexa) lined up and stretched out to help us see the "eyes" down in the noise:

https://www.dropbox.com/s/263ebj1psa4eldx/6x-XYLA-Waveforms.png?dl=0 (also attached, but probably too compressed by RED's severe attachment limitations).

For each image I looked at a monochrome image (which substantially reduces blue noise and the confusion of gradually degrading white balances as things tend to the shadows) and I applied curve that substantially lifts the lowest values--to better stretch up and see the eyes (if they are there).

3. Monstro has the best bars, and a very small eye 4 stops below the blue marker (which I placed at about 500 IRE to make it easier to count stops down consistently). Regular Gemini barely has an eye 3 stops down, but Gemini LL has a large eye 3 stops down (but not 4). Both Venices have eyes 3 stops down (but not 4). Alexa has a nice eye 3 stops down (but not as nice as Monstro).
 

Attachments

  • 6x-XYLA-Waveforms.jpg
    6x-XYLA-Waveforms.jpg
    81.3 KB · Views: 0
Michael Tiemann said:
3. Monstro has the best bars, and a very small eye 4 stops below the blue marker (which I placed at about 500 IRE to make it easier to count stops down consistently). Regular Gemini barely has an eye 3 stops down, but Gemini LL has a large eye 3 stops down (but not 4). Both Venices have eyes 3 stops down (but not 4). Alexa has a nice eye 3 stops down (but not as nice as Monstro).
Click to expand...

Thanks Michael for your tests.

I love that Monstro... Give us a S35mm cut. Pleeeaaase ;-)
 
Michael Tiemann said:
OK...some new thoughts.

1. The "EYE"s have it. For those of you who have looked at digital traces on oscilloscopes, you know what an "eye pattern" is:

Figure%201%20Eye%20diagram%20analysis.png


When the "eye" is open, it means there's room to distinguish signal from noise. When the eye closes, the electronics can't really make the distinction, even if we convince ourselves that we can still see the differences.

2. Here's a screenshot of all six waveforms (Monstro, Gemini 800, Gemini 1600 LL, SONY Venice 500, SONY Venice 2500, Alexa) lined up and stretched out to help us see the "eyes" down in the noise:

https://www.dropbox.com/s/263ebj1psa4eldx/6x-XYLA-Waveforms.png?dl=0 (also attached, but probably too compressed by RED's severe attachment limitations).

For each image I looked at a monochrome image (which substantially reduces blue noise and the confusion of gradually degrading white balances as things tend to the shadows) and I applied curve that substantially lifts the lowest values--to better stretch up and see the eyes (if they are there).

3. Monstro has the best bars, and a very small eye 4 stops below the blue marker (which I placed at about 500 IRE to make it easier to count stops down consistently). Regular Gemini barely has an eye 3 stops down, but Gemini LL has a large eye 3 stops down (but not 4). Both Venices have eyes 3 stops down (but not 4). Alexa has a nice eye 3 stops down (but not as nice as Monstro).
Click to expand...

I always wonder how best to do this. It definitely should be done in linear, in the full bit depth and color space of the sensor. At least unify the gamma and color space. Maybe compare the cameras in ACES?

Without linear the signal is getting stretched unevenly to fit a gamma, so the signal to noise ratio can appear different for different cameras.

An 8k camera at 8k will have 2 fewer stops that have the "eye" than an 8k camera at 2k. That's fundamental to super sampling and I think that's important to consider as well.

LUTs and other parts of color science can only reduce the total number of colors, never increase it. You can't create something that's not there, but you can simplify complex things. I always felt that Alexa's color science is somehow smoothing out colors. Look at images of gradients, their waveforms are made of smooth lines while red's are spikey blurs, but I don't perceive this noisiness visually... it just seems like red is leaving me a bit more of the random variation in hue and saturation and alexa is processing it into a single color.

Just my thoughts.
 
You must log in or register to reply here.
Back
Top