Thread: Gemini Noise Issue

Reply to Thread
Page 3 of 6 FirstFirst 123456 LastLast
Results 21 to 30 of 60
  1. #21  
    Senior Member Bill Totolo's Avatar
    Join Date
    Dec 2007
    Location
    Los Angeles, CA
    Posts
    337
    My take without NR, (from the jpg).


    [IMG][/IMG]
    Bill Totolo
    billtotolo.com
    Reply With Quote  
     

  2. #22  
    Senior Member rand thompson's Avatar
    Join Date
    Aug 2011
    Posts
    6,383
    Kyle,

    That would be great if you have the time. You can in general get better noise reduction if you use both temporal and Spacial noise reduction together. A short video will have multiple frames for the Temporal and Spacial noise reduction tools at least in resolve, to better analyze the pattern and the movement of the noise over multiple frames.
    Reply With Quote  
     

  3. #23  
    Senior Member rand thompson's Avatar
    Join Date
    Aug 2011
    Posts
    6,383
    Bill,


    I thought when I was finally able to download the .R3D file that it would be much better in the noise on the wall to the left and right of the Bed/or sofa curtains. However, except for color blockiness in the jpeg, the .R3d looked to have that same fine grain noise pattern.
    Reply With Quote  
     

  4. #24  
    Senior Member rand thompson's Avatar
    Join Date
    Aug 2011
    Posts
    6,383
    Redcine-X Pro

    Low contrast/Soft highlight Roll-off

    ISO=1000

    Kelvin=5487

    Temp=-2.985

    Custom RGB curves and Luminance curve, Saturation and Contrast


    Reply With Quote  
     

  5. #25  
    High compression on higher frame detail (scene detail + noise) leads to "muddy"/"dirty" texture.
    Heavy noise reduction turns image into a smudge fest.

    Shooting an UHD camera with high compression and relying on noise reduction defeats the purpose of UHD camera and leads to a smudge fest with reduced resolution and inconsistent texture.

    Noise reduction is not designed to remove compression artefacts.
    Noise has a footprint, compressed noise is variable.

    Ergo/hence/therefore >

    Computative compensation A (compression)
    +
    Underexposure
    +
    Computative compensation B (NR)
    =
    Frame becomes a CG smudge fest and image sequence a dance-floor for micro-goblins, the offspring of NR and wavelets feeding of texture.
    Analog > Apollo wooden handgrip http://omeneo.com
    Digital > Primers - professional image transformation tools http://omeneo.com/primers

    imdb


    "Como delfines en el fondo del oceano
    volamos por el universo incentivados por la esperanza"

    "L'esperanza", Sven Väth
    "It's a poor sort of memory that only works backwards"
    Jung/ Carol
    Reply With Quote  
     

  6. #26  
    Causes of noise:


    Capture:

    - Inadequate black shading
    - Underexposure
    - Poor light quality with inconsistent spectrum coverage (off the shelf household lights, work lights, street lights etc.)

    Post:

    - WB RGB channel push away from sensor native WB
    - Inappropriate first light/primary CC or image transformation
    - Grading workflow with cumulative effect to image signal values
    Analog > Apollo wooden handgrip http://omeneo.com
    Digital > Primers - professional image transformation tools http://omeneo.com/primers

    imdb


    "Como delfines en el fondo del oceano
    volamos por el universo incentivados por la esperanza"

    "L'esperanza", Sven Väth
    "It's a poor sort of memory that only works backwards"
    Jung/ Carol
    Reply With Quote  
     

  7. #27  
    Senior Member rand thompson's Avatar
    Join Date
    Aug 2011
    Posts
    6,383
    Hrvoje,

    Thanks for the detailed explanation of problem! No matter what I did with noise reduction, I kept running back into the same thing.
    Reply With Quote  
     

  8. #28  
    Senior Member
    Join Date
    Jul 2009
    Location
    New York
    Posts
    958
    Quote Originally Posted by Hrvoje Simic View Post
    Causes of noise:


    Capture:

    - Inadequate black shading
    - Underexposure
    - Poor light quality with inconsistent spectrum coverage (off the shelf household lights, work lights, street lights etc.)

    Post:

    - WB RGB channel push away from sensor native WB
    - Inappropriate first light/primary CC or image transformation
    - Grading workflow with cumulative effect to image signal values
    Thanks you for this. What do you mean by WB RGB channel push away from sensor native WB?

    Since WB is metadata, why should this matter? Just trying to understand how WB contributes to noise whether during production (setting it/ignoring it altogether) and during post as well. Many thanks.
    ---
    Nikhil Kamkolkar
    Writer/Filmmaker: http://www.IndianCowboy.com
    Latest action short film: "LAILA" https://youtu.be/wq0z8F-LJww
    Reply With Quote  
     

  9. #29  
    Senior Member Nick Morrison's Avatar
    Join Date
    Nov 2011
    Location
    Brooklyn
    Posts
    8,828
    Quote Originally Posted by Nikhil Kamkolkar View Post
    Thanks you for this. What do you mean by WB RGB channel push away from sensor native WB?

    Since WB is metadata, why should this matter? Just trying to understand how WB contributes to noise whether during production (setting it/ignoring it altogether) and during post as well. Many thanks.
    What he means is this.

    Almost all RED sensors are natively 5000 kelvin. So the more you stray from this, the more you stress the sensor. This is true of almost every CMOS sensor, not just RED.

    So yes you are shooting RAW, and you can change the WB later, but that isn't changing the linear light that's actually HITTING your sensor.

    You have to be aware of how your lighting and exposure are affecting every channel of your image - R, G, and B.

    IE there are decisions you can't change later that happen at a sensor level (how much light, and what color it is).

    And then there are decisions you can change later at a metadata level (WB, ISO, etc). But these won't change the physics of what you actually shot, they'll just help you manipulate the recording of this more easily later in post.

    Does this help?
    Last edited by Nick Morrison; 02-26-2019 at 10:12 AM.
    Nick Morrison
    Founder, Director & Lead Creative
    // SMALL GIANT //
    smallgiant.tv
    Reply With Quote  
     

  10. #30  
    Quote Originally Posted by Nikhil Kamkolkar View Post
    Thanks you for this. What do you mean by WB RGB channel push away from sensor native WB?

    Since WB is metadata, why should this matter? Just trying to understand how WB contributes to noise whether during production (setting it/ignoring it altogether) and during post as well. Many thanks.
    White balance as a channel ratio is metadata in raw.

    White balance as a function - affects balance :) - of RGB channels, and as you go off native sensor's WB you are pushing the material to achieve RGB channel balance to compensate colour values which were not recorded. In case of daylight balanced sensor white balanced to tungsten you are pushing the red channel and accentuating noise. So for the cleanest shot optical correction with filters will give the best result, at the cost of light loss.

    When you shoot raw you have the ability to fine tune white balance without penalites to separation and RGB luminance consistency, which in practice means getting a more accurate and natural image compared to baked-in RGB shot off WB and color corrected in post.
    That will not remove the penalty of increasing noise by pushing the channels.
    Analog > Apollo wooden handgrip http://omeneo.com
    Digital > Primers - professional image transformation tools http://omeneo.com/primers

    imdb


    "Como delfines en el fondo del oceano
    volamos por el universo incentivados por la esperanza"

    "L'esperanza", Sven Väth
    "It's a poor sort of memory that only works backwards"
    Jung/ Carol
    Reply With Quote  
     

Tags for this Thread

View Tag Cloud

Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts