«

»

May
06

Pixel Density vs Noise: Do Higher Megapixels make Noisier Pictures?

For more, SUBSCRIBE and like http://fb.com/NorthrupPhotography Buy the #1 book with 9+ HOURS of video: http://amzn.to/1dCRJWA Worldwide use 10% off coupon ‘YouTube’: http://sdp.io/sdpbook …

[ccw-atrib-link]

45 comments

No ping yet

  1. Tony Northrup says:

    A nerdy video studying the impact of pixel density & high megapixel sensors
    on image noise.

  2. Aron Gatt says:

    i think is the dynamic range that makes the difference here between these
    two cameras , i agree that more pixels will not necessarily = more noise
    but will definitely absorb less light, hence the noise

    its almost always or a big part of the equation, total light gathered.
    just like the crop sensor debate

  3. Eddie O'Connor says:

    this is how I’m classing all the major photographer you-tubers compared to
    tv channels
    +Tony Northrup is like the discovery channel of youtube
    +DigitalRev TV is like the comedy central of youtube
    +Jared Polin is his own TV channel
    haha

  4. Hammerhaus Kennel says:

    GREAT stuff Tony, thanks!

    As at least one other person has mentioned though, aren’t the big pixels
    (lower density) important for video where scaling isn’t as viable? Of
    course now there is the option for shooting in 4K and scaling, but even
    that is still a little bit of a different deal…

  5. Tech Innuendo says:

    Off-topic: Is that the MX Master mouse? If so, do you also have problems
    with the scrolling resetting every once in a while? It drives me nuts.

  6. Adrian Reyes says:

    If Tony goes missing, we’ll know sony got to him. Take care out there in
    Connecticut you beautiful scientific bastard.

  7. Ken Strong says:

    Tony, you are making my hair hurt….stop…ask this question..If Ansel
    Adams was alive, would he care? If not, take pictures and enjoy…..

  8. Sal Tutorialz says:

    great great video thank you!

  9. gurudeclan says:

    What about dynamic range? The d750 for example will give you substantially
    better latitude for shadow recovery compared to say a Canon 6d. This does
    in some ways, add to the noise performance.

  10. WADE CH says:

    Hi Tony, did you test them at ISO12800 and higher? Is it possible dxomark
    counts the score using weighted mean from ISO100 to 12800 or even 25600,
    while A7s might show better result at that extreme high ISO?

  11. Prince Gopalakrishnan says:

    Hi Tony, Do you own a D810 or not??? I thought you liked it and would have
    got one.:)

  12. Ren Lok says:

    Thanks Tony, just blew my mind as I was considering the A7s for low light.

  13. Norbert Rakosi says:

    This video is really interesting. A great explanation on how does the pixel
    density of the camera sensor related to the amount of noise you get in your
    images.

  14. Christos5120 says:

    Because nobody commended on it, I will put another parameter: Maybe DXO
    Mark rates Sony A7S “better” because their view is “obstructed” by the
    color GREEN 😉
    And I would have no problem believing it.
    That’s why I would rather rely more on Tony’s reviews than in any other
    “professional” site.

  15. 吴校杰 says:

    As far as I know that superb low light capacity of A7s shows when push the
    ISO up 25600 to 102400, and the image quality still usable. That’s not
    gonna happen on D810. (I DO NOT own any of these two cameras, but as those
    reviews I’ve seen, that’s my opinion)

  16. Kaoukabi Jaouad says:

    there is big difference between DXOMark and DPreview, DXOMark are testing
    the sensor regardless of the lens attached to it …. there more scientific
    … DPreview are more real world test, they try to find the best lens for
    any given camera to portray at best the capabilities of the sensor ….. i
    remember reading a review on the D5500, they mention that the sigma art
    lens 50mm dosen’t give the same optical quality than an older 50mm nikon
    lens …. the sigma introduces more noise than supposedly higher quality
    sigma art lens you can read that at the bottom in green (8.Lab Test :
    Studio Scene int he review of the d5500) and that seems logical if a lens
    has an effective resolution, they can logically affect noise too ….
    anyway a lot of details …. you are giving me headaches on this video …i
    didn’t understand all of it …. but it’s good from time to time for one to
    use his brain and get to the nitty gritty of things …..

  17. Baba Wethu says:

    I totally understand what you’re getting at, and yes, it’s true that more
    pixel density won’t make the image look more grainy at the same relative
    zoom. However, if each pixel gets too little light, every individual pixel
    is going to lose too much tonal quality. The D810’s three pixels gather as
    much light as one of the A7s. Let’s say each of the D810’s pixels acquires
    1 light unit. 1+1+1=3 which is A7s’ light/pixel. However, if each pixel
    gets a third of the light, each pixel is going to have, let’s say for this
    example, a third of the quality of each pixel the A7s, and the thing with
    downscaling is it takes an average of the three pixels. This makes it look
    just as smooth as A7s’ pixels, but the thing is that the tonal qualities
    and other qualities beyond just grain will look worse. (1+1+1)/3 compared
    to 3/1. Since every individual pixel looks a lot more shit on the D810 than
    every individual pixel on the A7s, when viewing the entire image as a
    whole, the D810 will look more degraded, even if it isn’t more grainy at
    the same relative zoom. The average of three shits is still a shit which is
    less than one less-shit.

  18. supercurioTube says:

    Another video developing well the need to compare cameras on total image
    noise and not per pixel noise.
    Both the data and analysis are sound, however while I’m still watching all
    I’m thinking about is “Yes but the A7s”

    So of course there’s a general rule that works in most cases, however
    there’s always gonna be a exceptions 😉
    In this case, the A7s, with its low big pixels count will produce a lot
    cleaner images than the D810 in the same conditions even once both output
    downscaled to the same resolution.
    Sometimes then, sensors can be optimized for high sensitivity, and it still
    seem to work better when designed with a lower pixel density.

  19. Joshua Adams says:

    Very enlightening, Tony. Thank you for another very well done and
    informative video!

  20. Mike Spivey says:

    Tony, your testing confirmed my suspicion that we weren’t comparing like to
    like. However, your testing was mostly at 6400. I shoot sprint car racing
    at night. My 5D Mk III is almost, but not quite good enough to shoot
    without flash. The A7s would not work because of the slow AF. But still, I
    would be interested in comparisons at 12,800 ISO and higher. Thanks.

  21. george aronis says:

    Make art and not megapixel war…

  22. cp dukes says:

    Really interesting episode. I’m trying to understand the noise I get with
    my D7100. Maybe the tstops?

  23. noonsound says:

    Excellent, and so glad you used the same lens in your testing. 

  24. sunny kiri says:

    Great comparison

  25. Fabio Gomes de Carvalho Monteiro says:

    Do you scale down the images in Camera Raw or in Photoshop?

  26. Razor2048 says:

    I feel it all depends on your needs. The D810 is not a low light camera, as
    after ISO 800, it loses far too much resolution.

    Since most images are shared on the web, 1:1 is also very important since
    there is a desire to see images in their full resolution. Remember, digital
    images, means that we can scroll through the image. This allows users to
    more closely examine an image, and essentially explore the images in order
    to notice more in the scene.

    If I buy a D810, I do it because when I upload a photo online, I want users
    to be able to see the picture of the tree, but not only that, but zoom into
    the image 1:1 and see if the squirrel in the tree is really making a gang
    symbol with its front paws or not.

    PS, one of the most annoying things possible in a camera review, is when
    some websites review a camera, and only upload sample images that are
    scaled down to like 2000x resolution or smaller. What that is done, it
    completely defeats the purpose of sample images as ignoring lens
    characteristics, if an image is scaled down like crazy, the sample photo
    could be taken with a point and shoot camera for all you know.

  27. Clark Kent says:

    +Tony Northrup i might still get the d700 tho, is still amazing

  28. Vincent112 says:

    Suppose I walk into the Bibliothèque nationale, the Bodleian Oxon, BL, NLI
    or the TCD library to photograph the illuminated manuscripts that are on
    display. Which camera will render the best image. Given that all of these
    places are aware that the items under their protection are degraded by
    light, and so the visibility pushes the eyesight nevermind a camera.
    Assuming that you have permission to record the images and don’t need the
    Sony’s relative size and pocketability. But are also using handheld.
    It’s one thing shooting little Johnny or Lainy galloping with little legs
    to cast a wing and a prayer at a basket, another again to record a page
    that’ll be seen once every twenty years with you living 5000 mls away.

  29. far0ashgar says:

    So GH4 handle noise and pixles better than the new canon 7D ? From the
    statistics slide. 

  30. mj rooney says:

    My Apple watch takes better low light pics than both the Nikon and Sony
    plus I can also know what my blood pressure was while taking the pics 

  31. randomgeocacher says:

    Great video as always. The obvious moot point which you seem well aware of,
    but still: video 🙂

    A7S users bought the camera for video – not stills, and most other cameras
    do line and pixel skipping, I.e only high megapixel cameras often use a
    small percentage of the sensor…

    The 5dmkII video buffer isn’t even HD as shown by Magic Lantern raw hack,
    and full of jaggy lines, clearly skipping pixels so less than 10% of sensor
    size is used in video.

    That’s where the A7S shines by using all of the pixels in video.

    Very good that it’s clarifies that these benefits aren’t stills benefits. I
    can see a bunch of people mistaking amazing video reviews results and
    believing that video benefits have any impact on stills benefits. And will
    be interesting to see what Dxomark skew is caused by (…I saw the theory
    on extreme ISO analog gain in another comment thread)

  32. TANGOMANification says:

    Simply, brilliant! I see a nerdy mathematician applying hard facts and
    logic to the photographic process. I like your tutorials. However coming
    from an older generation where we only had the use of film and study
    photographic history at A level (UK) digital technology is breaking new
    ground every day compared the days of Fox Talbot, Henry Western, the likes
    of the Daguerreotype etc. etc. Again the process of film tech was evolving
    but at a much slower pace! non the less fascinating though!Unless you’re a
    professional the tech is moving so fast as to bankrupt one finacially in
    what seems like a technocrat consumer war frenzy. It seems one has to be on
    a constant steep learning curve with both computer programs and new camera
    data on this on that; I, I, I, Bla, Bla, Bla!! Whatever??????

    Mind blowing it maybe BUT what has become a pretty dawned continuously
    expensive hobby!!!

  33. RocMat1 says:

    Have you contacted dxoMark about the results? Have they responded?

  34. HaouasLeDocteur says:

    Thorough and detailed. Perfect analysis. Thank you!

  35. Cory Hultquist says:

    So dry I feel like I need a drink of water…. lol. But I love this stuff!

  36. Bowie Sensei says:

    It would be good to see still image comparisons at a range of isos, as my
    understanding was the d810 gave cleaner images when you need a low iso, but
    high pixel density means you cant have high iso performance in the say 12k
    to 50k+ range. My d810 images at 12k iso are utterly horrible, but i don’t
    have an a7s.

  37. David Schmetzer says:

    Sorry +Tony Northrup, you say Nikon glass are sharper… I agree with you
    on zooms, but not on the prime side. According to dxomark.

  38. The Post Color Blog says:

    Thank you for pointing out the d4s (the low light king before Sony came
    along, especially at its megapixel count) and also pointing out that video
    is processed differently. It’s a compressed codec, unlike raw stills. 

  39. Saint Si says:

    Is it true that Sony make all Nikon sensors.

  40. cadmus777 says:

    Another great video, thanks Tony. I’m wondering if you tried different
    scaling algorithms at all? I’m sure noise scales somewhat differently
    depending on the algorithm you use to scale it up or down. Just curious…

  41. Ruben Kristianto says:

    but…d810 is heavier than a7s . LOL

  42. Camera von Theuns says:

    Oh, and comparing the D810 with A7s on dynamic range is rather painful – at
    51k ISO it’s 6.39 EV vs. 8.19 EV. That’d certainly impact how images
    respond to processing.

  43. Glen Johnson says:

    An excellent video, thank you.

    A better comparison may have been the A7r vs. the A7s, that way you could
    have used the same lens on both cameras. I suspect you had to use an
    adapter that could have effected your results, although I would expect that
    to effect them negatively for the nikon.

    In terms of the first generation of Sony cameras, one reason to use the A7s
    over the A7r was focusing speed. Its possible that having less MP makes it
    faster to focus.

  44. Saumyadip .Kumar says:

    Great video. Though you previously discussed that pixel density does not
    impact total noise this just made everything more clear. Thanks

  45. BlackWipeout says:

    I say only… Not all people need 36Mpxl and you need a good PC/Mac for
    handle this data in Lightroom 6 ;)

Comments have been disabled.