Is the F65 known for being noisy in the shadow areas or am I just exposing incorrectly?
https://vimeo.com/302783032
https://vimeo.com/302783032
F65: Noisy or underexposed?
- Thread starter bitnaut
- Start date
morgan_moore
Major Contributor
why dont you bracket your shots and learn to use your camera?
as it happens I think it looks fine, I think you should add NR last (or not at all) and I think you should open up a stop.
as it happens I think it looks fine, I think you should add NR last (or not at all) and I think you should open up a stop.
Grug
Veteran
Hard to tell much from the compressed vimeo output. Can you post a sample using just the SLOG2 LC709A LUT?
I can't really see any noise in the SLOG clips, and the pattern in the Filmconvert clips appears to be the film grain overlay that the plugin adds.
The F65 is suuuuper clean. And your exposure looks good.
I can't really see any noise in the SLOG clips, and the pattern in the Filmconvert clips appears to be the film grain overlay that the plugin adds.
The F65 is suuuuper clean. And your exposure looks good.
abreu-canedo
Well-known member
Hi Ed,
I wouldn't worry too much about noise. As you test and use the camera you'll find your different sweet spots with it. The F65 can get noisy in the shadows, like any camera that doesn't apply NR. Turn off the NR in an Alexa and explore the shadows... same thing.
It's different for everyone and DR and noise floor is so subjective that I don't really bother with hard and fast ratings. But I do understand the grain structure/quantity/quality at different IREs... I use that knowledge to light/shape parts of my image accordingly. So ultimately, you tell us... did you underexpose? You'll be able to answer this once you understand the way in which your signal falls into the noise floor and whether parts of your image were lit too close to the noise floor for your liking.
It's hard to judge from your vimeo, since it's too compressed. But if you turn on false color on your original media, you can tell what the grain looks like on your code values or IRE. There is not quite as high DR as Alexa. Btw, what IRE or Code Value is your window on frame right... on the original media, that is. And what format are you shooting?
I wouldn't worry too much about noise. As you test and use the camera you'll find your different sweet spots with it. The F65 can get noisy in the shadows, like any camera that doesn't apply NR. Turn off the NR in an Alexa and explore the shadows... same thing.
It's different for everyone and DR and noise floor is so subjective that I don't really bother with hard and fast ratings. But I do understand the grain structure/quantity/quality at different IREs... I use that knowledge to light/shape parts of my image accordingly. So ultimately, you tell us... did you underexpose? You'll be able to answer this once you understand the way in which your signal falls into the noise floor and whether parts of your image were lit too close to the noise floor for your liking.
It's hard to judge from your vimeo, since it's too compressed. But if you turn on false color on your original media, you can tell what the grain looks like on your code values or IRE. There is not quite as high DR as Alexa. Btw, what IRE or Code Value is your window on frame right... on the original media, that is. And what format are you shooting?
Last edited:
abreu-canedo
Well-known member
Also, if you want, you can map the way the signal in any camera falls into the noise floor. This essentially maps the exceeding levels of noise as your signal descends into the noise floor of the sensor. For example, shooting a DR chart or a bracket test with a control card. The smaller steps you bracket, the more fidelity with which you'll be able to map the noise curve as your signal descends into the noise floor of the sensor. Then you can take this curve and use it to create custom NR settings that are in direct proportions the way the signal rolls into the noise floor, essentially cancelling it out.
Then, you ideally do the same test with a film stock (with a control card), and scan the grain structure at different stops (or code values once you scan it). You'll overlay this scanned film grain to the image, to counteract the smoothing effect of NR as it reduces noise deeper into the noise floor.
That Vision 3 LUT that you are using is quite extreme on my monitor. Perhaps it's as intended. But also look into color space transforms or ACES (if you haven't already). You will find that very few LUTs out there were buit for the gamut and gamma combinations of the F65.
Then, you ideally do the same test with a film stock (with a control card), and scan the grain structure at different stops (or code values once you scan it). You'll overlay this scanned film grain to the image, to counteract the smoothing effect of NR as it reduces noise deeper into the noise floor.
That Vision 3 LUT that you are using is quite extreme on my monitor. Perhaps it's as intended. But also look into color space transforms or ACES (if you haven't already). You will find that very few LUTs out there were buit for the gamut and gamma combinations of the F65.
Great ideas abreu-canedo and morgan_moore!
To answer your question, I put the bright window on frame right at 61 IRE and shot in SLOG2/S.Gamut Cine mode at 800EI. The Vision 3 look was achieved with Film Convert using their F55/SLog2.SGamut camera pack and turning off the grain pattern in the first sample, and bumping to 50% grain in the NR'd sample.
I was experimenting with the F55 looks since it's supposed to have the same colour filter array as the F65.
Noisiness aside, I do love the results I'm getting from the F65 and can't wait to shoot something more substantive with it soon.
To answer your question, I put the bright window on frame right at 61 IRE and shot in SLOG2/S.Gamut Cine mode at 800EI. The Vision 3 look was achieved with Film Convert using their F55/SLog2.SGamut camera pack and turning off the grain pattern in the first sample, and bumping to 50% grain in the NR'd sample.
I was experimenting with the F55 looks since it's supposed to have the same colour filter array as the F65.
Noisiness aside, I do love the results I'm getting from the F65 and can't wait to shoot something more substantive with it soon.
Last edited:
abreu-canedo
Well-known member
I'm almost certain the CFA's are different between F55 and 65... color space is a bit wider in F65. It just means some unreliable results when remapping values.
Did you record RAW, or to SR Codecs?
rbirnbaum
Well-known member
Ed, I think NR kills the image. Makes it look plasticky. In terms of LUTs and Grades... well, it is always a personal preference, so crushing, lifting, rolling... I mean they are all creative options. Crushing always helps with noise in the shadows. You just crush the noise into oblivion 
But if it looks crushed... it is crushed... hahaha
I follow this workflow in case it works for you with my F35.
In Resolve, I use the 'color space transform' plugin on a clip. Or I directly convert the whole timeline using Resolve Color Management in the Color Settings tab in general preferences.
Convert
sgamut to Alexa
Slog to LogC.
Then apply Alexa Luts to the footage. I get better results this way, because LogC is very close to Cineon, which was the base for most LUTS. try it that way. FC should also work great.
But if it looks crushed... it is crushed... hahahaI follow this workflow in case it works for you with my F35.
In Resolve, I use the 'color space transform' plugin on a clip. Or I directly convert the whole timeline using Resolve Color Management in the Color Settings tab in general preferences.
Convert
sgamut to Alexa
Slog to LogC.
Then apply Alexa Luts to the footage. I get better results this way, because LogC is very close to Cineon, which was the base for most LUTS. try it that way. FC should also work great.