Latest / The Nerdy Photographer Podcast / 176 - Bit Depth Demystified: The Math of Color

Transcript

176 - Bit Depth Demystified: The Math of Color · The Nerdy Photographer Podcast

• 0:00
  Casey Fatchett: Hey, welcome back to the Nerdy Photographer
• 0:08
  Podcast. I'm your host, Casey Fatchett. Today we're going
• 0:11
  somewhere equally nerdy as we did before, but in a completely
• 0:14
  different direction. We're going inside the image file itself,
• 0:19
  specifically we're talking about bit depth, what it is, why it
• 0:24
  matters, and why it matters enormously for editing, and why
• 0:28
  the high efficiency or compressed raw settings on your
• 0:31
  camera might be quietly costing you some image quality in
• 0:35
  situations where you need it the most. If you've ever pushed a
• 0:39
  raw file hard in Lightroom, and watched it hold up beautifully.
• 0:43
  Then tried the same thing to JPEG, and watched it fall apart
• 0:48
  with banding noise, ugly transitions. Bit depth is
• 0:52
  exactly why that happens. And once you understand the math,
• 0:57
  you'll never think about file formats the same way again. So,
• 1:02
  what is a bit? Let's start from the ground up. A bit is the
• 1:06
  smallest unit of digital information. It's a switch, on
• 1:09
  or off, one or zero. When you string bits together, you get
• 1:13
  the ability to represent more and more values. One bit gives
• 1:19
  you two possible values, zero or one two bits gives you 400011011
• 1:27
  Each additional bit doubles the number of possibilities. This is
• 1:33
  why bit depth is exponential, not additive. Going from eight
• 1:38
  bit to 14 bit isn't six more, it's two to the eight versus two
• 1:46
  to the 14th, that's 256 versus 16,384 a difference of more than
• 1:54
  16,000 steps. Now here's the part that makes the relevant to
• 2:02
  photography bit depth in an image file refers to the number
• 2:06
  of brightness levels available per color channel, and your
• 2:10
  camera sensor records three channels: red, green, and blue.
• 2:15
  The total number of possible colors in a file is calculated
• 2:20
  by multiplying those per channel values together three times, I
• 2:25
  know your understanding was that there would be no math, but so
• 2:29
  for an eight bit jpeg, 256 times 256 times 256 is roughly 16
• 2:36
  point 7 million possible colors. Now that sounds like a lot, but
• 2:43
  for a 14 bit raw file, 16,003 84 times 16,003 84 times 16,003 84
• 2:55
  over 4.4 trillion possible colors, that's not a small
• 3:02
  difference, that's a difference of roughly 260,000 times more
• 3:08
  color information sitting in your raw file compared to a
• 3:11
  jpeg, and the reason that matters isn't about what the
• 3:15
  image looks like straight out of the camera, it's about what
• 3:18
  happens when you start pushing and pulling in post, think about
• 3:25
  a smooth gradient, a clear blue sky shifting from deep blue at
• 3:29
  the top to a lighter haze near the horizon. In reality, that
• 3:34
  transition is continuous. There are no steps at all, but a
• 3:38
  digital file has to represent that continuous transition as a
• 3:42
  series of discrete values, and the number of values available
• 3:46
  determines how smooth those steps are. In an eight bit file,
• 3:50
  you have 256 steps of brightness per channel to describe that
• 3:54
  sky. If those steps are spread across a wide tonal range, the
• 3:59
  gap between one step and the next can be visible, especially
• 4:03
  after editing, when you increase exposure or pull up the shadows,
• 4:07
  you're essentially stretching that staircase, and the gaps
• 4:10
  between steps get wider, wide enough, and you start to see
• 4:15
  them, and that is called banding, folks, those ugly
• 4:19
  parallel stripes that make a smooth sky look like a cheap
• 4:23
  gradient from 2003 So, in a 14 bit raw file, you have 16,384
• 4:31
  steps for the same transition. The staircase is so fine that no
• 4:36
  amount of reasonable editing will make the steps visible. You
• 4:39
  can push the exposure slider two stops, pull the shadows up, and
• 4:43
  that sky will still look smooth, because there's enough data
• 4:47
  underneath to absorb the stretch. This is the single most
• 4:51
  practical reason to shoot raw instead of jpeg, not because the
• 4:56
  images look different straight out of the camera, often they
• 4:58
  don't, especially with a good. Jpeg profile, but because raw
• 5:02
  files are editing headroom, their insurance, the information
• 5:08
  is there when you need to recover a shot that wasn't
• 5:11
  perfectly exposed in the field. Now, most modern cameras give
• 5:17
  you a choice within raw 12 bit or 14 bit, some high-end medium
• 5:22
  format systems go all the way up to 16 bit, and you might wonder
• 5:26
  if 14 bit is so much better than eight bit. Is 14 bit
• 5:29
  meaningfully better than 12 bit? The answer is, it depends on
• 5:33
  what you're shooting. 12 bit raw gives you 4096 steps per
• 5:38
  channel. 14 bit gives you 16,384 That's a a four times increase
• 5:44
  in tonal resolution. In practice, for well-exposed
• 5:48
  images of typical subjects, you'll probably never see the
• 5:52
  difference, but for scenes with subtle gradients, misty
• 5:55
  landscapes, smooth studio backdrops, heavily underexposed
• 6:00
  shadows, you plan to lift significantly in post 14 bit
• 6:04
  gives you noticeably more latitude before banding appears.
• 6:08
  The trade off is file size, and on some camera bodies, write
• 6:12
  speed. 14 bit files are larger, and on cameras with slower
• 6:15
  processors or smaller buffers, shooting 14 bit can reduce how
• 6:20
  many frames you can fire in a burst before the camera starts
• 6:23
  to slow down. For sports and wildlife photographers,
• 6:26
  hammering the shutter, 12 bit might be the smarter practical
• 6:29
  choice for landscape portrait and studio work, where you're
• 6:33
  not sprinting through a burst. 14 bit all day. So, how about
• 6:39
  lossless versus lossy compression. Let's talk about
• 6:44
  compressed raw, because this is where a lot of cameras have
• 6:48
  quietly introduced a setting that can catch people off guard.
• 6:51
  When your camera writes a raw file, it can do it in several
• 6:55
  ways. Uncompressed raw is the raw sensor data with nothing
• 7:01
  done to it, maximum file size, maximum information. Lossless
• 7:06
  compressed raw applies an algorithm that reduces the file
• 7:09
  size without discarding any of the actual data. Think of it
• 7:14
  kind of like a zip file. The file is smaller to store, but
• 7:17
  when you unzip it, everything's exactly as it was before, no
• 7:21
  information lost for most photographers. Lossless
• 7:24
  compression is a smart default. You get meaningful file size
• 7:28
  reductions with zero quality penalties. Lossy compressed raw
• 7:33
  is different here. The camera actually discards some of the
• 7:37
  sensor data to achieve smaller file sizes, and this is where
• 7:41
  you really need to be careful, unlike a JPEG, which
• 7:43
  recompresses and loses quality every time you resave it. A
• 7:47
  lossy raw file only degrades once at the moment of capture,
• 7:52
  so it's not quite the photocopy of a photocopy problem with you
• 7:55
  have with JPEGs, but you are making a permanent one-time
• 7:59
  decision to throw away data that you can never recover. You can't
• 8:03
  get it back for a casual portrait or a scene with simple
• 8:06
  tonal gradients. You might never notice for a landscape with a
• 8:10
  subtle sky or a scene where you're planning to do heavy
• 8:13
  shadow recovery. You might. The question is whether you want to
• 8:17
  make the trade off at the moment of shooting when you don't yet
• 8:21
  know how hard you're going to need to push the file in post.
• 8:25
  My general advice: use lossless compression as your default.
• 8:28
  File size savings over uncompressed are meaningful, and
• 8:31
  you lose nothing. Only drop to lossy if you have a specific
• 8:36
  practical reason, like filling a buffer during a fast action
• 8:39
  burst, or managing storage on a very, very long shoot, so to
• 8:46
  recap, bit depth is exponential, not linear. An eight bit JPEG
• 8:51
  gives you 16 point 7 million possible colors. A 14 bit raw
• 8:55
  gives you 4.4 trillion. That difference is invisible until
• 9:00
  you start editing, and then it's everything. Banding is what
• 9:05
  happens when you try to stretch a staircase that doesn't have
• 9:08
  quite enough steps, and lossless compression is your friend.
• 9:11
  Lossy compression is a trade off you could, you should only make
• 9:15
  consciously and not by accident. Today's homework is what I call
• 9:19
  the stress test. Find a scene with a smooth gradient, a clear
• 9:23
  sky, a softly lit wall, anything with a gentle tonal transition.
• 9:29
  Shoot it simultaneously as a jpeg and as a 14 bit raw. Bring
• 9:34
  both into your editing software and push the exposure slider to
• 9:38
  plus three or plus four stops. Then pull the shadow sliders all
• 9:42
  the way up. Look at the smooth areas of both files at 100%
• 9:48
  Watch the jpeg start to fall apart, banding in the sky or
• 9:52
  noise in the shadows. Watch the raw hold together. Once you've
• 9:56
  seen that, you won't need anyone to convince you to shoot raw.
• 10:00
  Awe again until next time. Stay safe and stay nerdy. Hey folks,
• 10:10
  it's Casey. I'm back. There's a couple things I forgot to
• 10:12
  mention. One, please be sure to subscribe to the show. Two,
• 10:16
  please leave a review for the show. Reviews help us with
• 10:18
  discoverability, new people finding the show, so leave a
• 10:22
  review on Apple Podcasts, Spotify, wherever you can leave
• 10:26
  a review for the podcast. It is greatly, greatly appreciated.
• 10:30
  And finally, I would love to hear what you think of these
• 10:33
  short little nerd lit episodes that we're doing. Do you like
• 10:37
  them? Find me on social media at the Nerdy Photo on pretty much
• 10:43
  everything: Instagram threads, Reddit, TikTok, Blue Sky. Find
• 10:49
  me there, mention it. Let me know what you think of these
• 10:53
  episodes, whether I should continue doing them, whether you
• 10:57
  are enjoying them, and if there's a different topic you'd
• 11:00
  like for me to cover in a short episode, I would love to hear
• 11:04
  about it. And now, without further ado, stay safe. Stay
• 11:09
  here.