Chapter 3 — Exercises
Chapter 2 gave you the four settings that build an image; this chapter gave you the choices that decide what you can do with that image later. The trouble is that almost none of these choices are visible on set — the whole point of the chapter is that a good file and a doomed file look identical on the camera's screen and reveal themselves only in the grade. So most of these exercises share one method: capture something, then push it in post until it breaks, and watch where and why it breaks. That is the only way to make an invisible decision visible, and once you've seen 8-bit footage band with your own eyes, you will never guess about bit depth again.
A phone is enough for most of these. Where an exercise needs a control your phone hides (10-bit, manual bitrate, a Log profile), do the version you can do and read the rest as the reason to reach for a manual app or a camera that offers it — the understanding transfers either way. And keep one habit through all of them: when a clip breaks in the grade, don't just note that it broke — note where (the sky? the shadows? a colored edge?) and how (banding? blocking? smearing?). That "where and how" is the whole diagnostic vocabulary of this chapter, and naming it is what turns a vague "that looks bad" into a decision you'll make correctly next time.
Exercises are graded by effort and depth:
- ⭐ Warm-up — a few minutes; locks in a concept.
- ⭐⭐ Core — the real practice; do as many as you can.
- ⭐⭐⭐ Stretch — deeper or more ambitious; where the biggest gains are.
Model solutions and critiques for the starred and odd-numbered exercises are in Answers to Selected Exercises at the back of the book. Try each one before reading its answer. One rule throughout: watch and grade your clips on the biggest, best screen you have — banding, macroblocking, and the "flat Log" look are subtle on a phone preview and undeniable on a laptop or TV.
A. Seeing the format (Eye & Ear Training)
No camera needed — just attention, extending your Frame Log toward a technical eye for recording quality.
1. ⭐ Spot the banding. Watch for banding in the wild: a sunset or dawn sky in a streaming show, a smooth studio backdrop behind a talking head, a slow fade to black. Pause on a smooth gradient and look for stair-stepped bands instead of a seamless blend. Note one clear example. (You'll see it most on heavily-compressed streams and social re-uploads.) (Answer provided.)
2. ⭐ Hunt for over-compression. Find a video that looks "blocky" or "mushy" — an old low-bitrate upload, a heavily-forwarded social clip, a dark scene on a weak connection. Name the artifact you see (macroblocking? smearing? banding?) and guess the cause in one line. (Answer provided.)
3. ⭐⭐ The re-upload degradation test. Take one short clip and post/download it through a compressing platform (a messaging app or social upload is perfect), then compare the returned file to your original on a large screen. Self-review: what did the platform's compression do to gradients, shadows, and fine motion? Write two sentences. You've just watched lossy compression happen to your own footage.
4. ⭐⭐ Standard vs graded, in things you love. Pick a film or show known for a strong "look" (a heavy teal-orange grade, a warm nostalgic wash, a bleached thriller). Watch two minutes just for the color, then ask: could this look have come straight out of the camera, or was it obviously built in a grade? What tells you? (Answer discusses how to read a grade.)
5. ⭐⭐⭐ The dynamic-range audit. For one day, whenever you watch anything with a bright window or sky behind a person, freeze and judge: did the camera hold detail in both the person and the window, or did it sacrifice one? Note three examples and which end each one protected. You're training the eye that Chapters 5, 11, and 13 will arm. (Answer provided.)
B. Reading Described Shots (Read the Sequence)
The Described Shot is how this book replaces "watch this clip." Reading the recording decisions hidden inside one is the same muscle you use to plan a capture.
6. ⭐ Name the format decision. Re-read FIGURE 3.6 (standard vs Log). In your own words, list what the standard profile gave up to look good straight away, and what Log gave up to keep its flexibility. If you can't name both sacrifices, re-read §3.4. (Answer provided.)
7. ⭐⭐ Read the ladder. Look again at FIGURE 3.7 (the flexibility ladder). For each rung — standard 8-bit 4:2:0, Log 10-bit 4:2:2, RAW — name one real shoot from your own life or plans that belongs on that rung, and defend the placement in a sentence. (Answer provided.)
8. ⭐⭐ Diagnose the pipeline figure. Study FIGURE 3.1 (codec vs container). A colleague says, "Just change the file from .MP4 to .MOV and it'll import fine." Using the figure, explain why renaming the extension won't fix a codec problem, and what actually would. (Answer provided.)
9. ⭐⭐⭐ Write a format Described Shot. Choose a real shot you plan to make and write a full seven-field Described Shot for it — but make THE FRAME and THE EFFECT explicitly name your recording format choices (profile, bit depth if known, bitrate) and tie each to what you'll do in post. Self-review: could a stranger rebuild both your framing and your capture settings — and your reasons — from the box? (Model answer provided.)
10. ⭐⭐ The café grade, previewed. Re-read the 🔗 Connection about the Café Scene's future warm grade (§3.4). In three bullets, explain what you would need to capture now for that heavy warm grade to be possible later, and what would happen if you shot it 8-bit 4:2:0 with a contrasty look baked in. (Answer provided.)
C. Your format shoots (Shoot This)
The heart of the chapter, and the part that actually changes how you shoot. Every one of these follows the same three-step method: capture something in a specific format, take it into an editor, and push it until it breaks — because a format's true quality is invisible until you grade it. Do not judge any of these by how the clip looks on the camera; the whole lesson is that the camera's screen lies to you about format. A phone works for most; a manual video app unlocks the bitrate and profile ones, and if yours is locked to auto, do the versions you can and treat the rest as your shopping list.
11. ⭐⭐ The banding test. If you haven't done the 🎬 On Set from §3.3, do it now. Subject: a genuinely smooth gradient (dusk sky, a softly-lit plain wall, a white surface fading to shadow). Constraint: shoot it in your most compressed / lowest-bitrate 8-bit mode, then push it hard in an editor — heavy contrast, deep shadows, a strong tint. Self-review: describe where and how fast it bands. If you have a 10-bit mode, shoot the same thing in it and push it identically; keep both clips. (Model self-critique provided.)
12. ⭐⭐ Bitrate A/B. If your camera lets you set bitrate (or offers "standard" and "high" quality modes), shoot the same busy, hard-to-compress scene — wind in foliage, moving water, a textured crowd — at a low and a high bitrate. Watch both large. Self-review: where does the low-bitrate version smear or block that the high one holds? One sentence on what this teaches about matching bitrate to scene difficulty. (Answer provided.)
13. ⭐⭐⭐ Standard vs Log, graded. If your camera has a Log or "flat" profile, shoot the same subject twice — once in a standard profile, once in Log — under identical light. Bring both into an editor. Leave the standard one alone; grade the Log one back to a normal look (apply a color-space transform or the maker's LUT, then match it to the standard version). Self-review: which gave you more control to build the look you wanted, and what did the Log one look like before you graded it? (Guidance provided.)
14. ⭐⭐ The white-balance rescue test. Deliberately shoot a subject at the wrong white balance (set the camera to "tungsten/indoor" under daylight, so everything goes blue). Now try to fix it in post, once with a standard-profile clip and — if you can shoot RAW — once with a RAW clip. Self-review: which corrected more cleanly, and why does RAW make white balance nearly "free" to change? (Answer provided.)
15. ⭐⭐ The green-screen edge test. Shoot a subject (a hand, a toy, yourself) against the most saturated, evenly-lit colored background you can improvise. Pull a rough key in any editor. If you can shoot both an 8-bit 4:2:0 and a 10-bit (or higher-chroma) version, compare the edges of the key. Self-review: which had cleaner, less blocky edges, and how does this connect to §3.3's chroma subsampling? (Green screen is fully covered in Chapter 35.) (Answer provided.)
16. ⭐⭐⭐ Shoot the dynamic-range problem. Frame a person indoors against a bright window (the classic Café Scene situation). Expose for the face — note the window blowing out to featureless white. Re-expose for the window — note the face going black and losing detail. You are watching a scene that contains more stops than your camera can hold in one exposure (FIGURE 3.5). Now try to hold both ends: move the subject away from the window, add a bounce to lift the face, or shoot a flatter/Log profile that keeps more range. Self-review: one paragraph on which end you chose to protect and why, and what tool would let you keep both — previewing the exposure work of Chapter 5 and the lighting of Chapters 11 and 13, where you'll wrestle this exact problem with the Café Scene. (Answer provided.)
17. ⭐⭐ The four-setup format sweep. Pick your capture format for each of the four setups as if you were shooting all four this week: talking-head, walk-and-talk, product on a table, event in available light. Write profile + bitrate (+ bit depth/chroma if you can) for each, one line of reasoning apiece. Self-review: did the format actually change between setups? It should — a controlled product and a run-and-gun event want different things. (Answer provided.)
D. One subject, five ways (Five Ways)
The fastest way to feel how much a recording choice matters is to hold the subject dead still and change only the format, five times, back to back. A difference that's invisible when you look at a single clip becomes undeniable when you watch five versions in a row and one of them bands while the others hold. Shoot these with a phone or a camera, lock it on something steady, and view the five as a single sequence on the biggest screen you have — small previews hide exactly the differences these exercises exist to reveal.
18. ⭐⭐ Five compressions of one gradient. Shoot one smooth gradient five ways, changing only recording quality: (a) lowest bitrate 8-bit, (b) highest bitrate 8-bit, (c) 10-bit if available, (d) a phone "high efficiency"/HEVC mode, (e) if possible, a light-vs-heavy grade of the best one. Push each the same amount. View all five in a row. Self-review: rank them by how much grade they survived, and defend your ranking. (Answer discusses trade-offs.)
19. ⭐⭐⭐ Five profiles of one face. If your camera offers several profiles (standard, a flat/neutral one, HLG, Log), shoot the same face under the same light in each, then grade each toward the same target look. Self-review: which profile reached the target most easily, which resisted, and which looked best ungraded? Note that "best ungraded" and "best after grading" are often different clips — that difference is the whole standard-vs-Log decision. (Model answer provided.)
E. Settings and scenarios (Settings Drill)
Given a situation, write the capture format and the trade-offs — the reasoning matters more than the "right" answer.
20. ⭐ Read the pair. For each, say in one line what it means and how it grades: (a) 8-bit 4:2:0, (b) 10-bit 4:2:2, (c) 12-bit RAW. (Answer provided.)
21. ⭐⭐ Four briefs, four formats. Write profile, bit depth/chroma (if available), and bitrate for each, with a one-line reason: (a) a same-day-turnaround talking-head testimonial; (b) a moody branded piece you'll grade hard next week; (c) a green-screen product composite; (d) an all-day conference in mixed light on limited storage. Self-review: could you defend each to a client who asks "why not just shoot RAW to be safe"? (Model answers provided.)
22. ⭐⭐ The card-speed check. Your camera's top mode is 400 Mb/s. (a) How many MB/s is that actually writing? (b) Is a V30 card enough? A V60? (c) What happens on set if the card can't keep up? Show the arithmetic. (Answer provided.)
23. ⭐⭐⭐ The "should I shoot Log?" decision. For each shooter, answer yes/no to Log and justify in two sentences using the three strings from §3.4: (a) a creator with an 8-bit phone who wants a "cinematic" look but has never graded; (b) a shooter with a 10-bit camera, a moody brief, and a week to grade; (c) a wedding videographer delivering same-day highlights. (Model answers provided.)
24. ⭐⭐ Write the offload plan. For a one-day, two-card shoot far from home, write your card-and-offload workflow start to finish: card choice, when you offload, how many copies exist before you reuse a card, and how you confirm the footage is good. Name the chapter that formalizes this. (Answer provided.)
F. Fix the shot (Fix the Shot)
Diagnosis is its own skill, separate from shooting — most real format problems arrive as "something's wrong in the edit and I don't know why," and naming the cause is most of the cure. Each of these describes a failed capture or a post disaster; your job is to name the format decision at fault and prescribe the specific fix, sorted by the stage it belongs to. A useful discipline: for each one, decide whether the problem is recoverable (a workflow issue you can fix now) or permanent (baked in at capture, fixable only on the next shoot) — that distinction is the hard truth this whole chapter teaches.
25. ⭐ Fix the won't-import file. A clip refuses to import; another imports as sound with a black picture. The editor insists the files are "corrupt." Name the likely real cause and two fixes. (Answer provided.)
26. ⭐⭐ Fix the banded sky. A drone shot of a sunrise looks fine on the camera, but the moment the colorist deepens the sky it shatters into stepped bands that get worse as they push. Name the two possible culprits and what should have been shot instead. (Answer provided.)
27. ⭐⭐ Fix the grey, broken video. A beginner proudly delivers footage that looks flat, grey, and lifeless — "but it's the cinematic setting." Diagnose what happened and give the two-part fix (immediate and for next time). (Answer provided.)
28. ⭐⭐⭐ Fix the un-editable 4K. An editor's laptop stutters so badly on a batch of 4K clips they can't cut, though the files play fine in a media player. Nothing is wrong with the exposure or the framing. Name the codec issue and the standard fix, and the chapter that covers it. (Answer provided.)
29. ⭐⭐⭐ Fix the lost footage. A shooter formats a card to reuse it, then discovers the only copy of a morning's interview was on it and the offload had silently failed. Diagnose the two process failures and write the rule that would have prevented it. Constraint: the footage may be gone — your answer is about the habit, because this one can't be fixed in post. (Answer provided.)
G. Recreate it (Recreate It)
You own a technique by rebuilding it with your own subject — never by copying footage. Each of these asks you to reproduce a phenomenon (a flat-to-graded reveal, an over-compressed texture, a white-balance rescue) so you understand the mechanism behind it and can summon it, or far more often deliberately avoid it, on purpose. The goal each time is a defensible sentence: this format choice produced this result, and here's when I'd want it.
30. ⭐⭐⭐ Recreate the flat-Log-to-graded reveal. After Case Study 1, recreate the standard workflow that gives graded films their look: shoot a flat or Log clip that looks deliberately "wrong" (grey, washed out), then build it into a real look in the grade. Self-review: two sentences on why starting from "wrong-looking" flat footage gave you more control than starting from a pretty baked-in image. (Guidance provided.)
31. ⭐⭐ Recreate the over-compressed look. Deliberately make footage look cheap and "internet-degraded": shoot or export a busy scene at the lowest bitrate you can force, then re-compress it once or twice more. Self-review: name the artifacts you produced and one (rare) time you'd want this look on purpose (a "found footage" or retro-web aesthetic). (Answer provided.)
32. ⭐⭐⭐ Recreate the white-balance save. If you can shoot RAW (or your phone's most flexible format), recreate the classic RAW rescue: shoot a scene under mixed or wrong-colored light, then set the white balance entirely in post. Self-review: one paragraph on how much freedom this gave versus a baked-in standard clip, and why you still shouldn't rely on it as an excuse to be sloppy on set. (Answer provided.)
H. A taste of the edit (Cut This)
You don't need advanced skills to feel how these formats live in post. Any free editor works (DaVinci Resolve is the book's reference from Chapter 26 on).
33. ⭐⭐ Grade until it breaks. Take your most-compressed clip and your cleanest clip of the same subject. Apply the identical aggressive grade to both — deepen the shadows, push a color hard, crank the contrast. Self-review: document the exact point where the compressed one falls apart (the setting, the amount of push) and where, if ever, the clean one does. This is the ceiling of your capture format, measured with a number instead of a vibe — and it's the single most useful thing to know before any shoot where a real grade is coming. Keep both clips and the note; you'll want to point to them the next time someone insists bitrate "doesn't matter." (Guidance provided.)
34. ⭐⭐ Transcode a stubborn file. Take a heavily-compressed clip and transcode it to an editing-friendly codec (ProRes/DNxHR, or your editor's optimized/proxy media). Self-review: does it scrub more smoothly now? Note that transcoding makes editing easier but cannot add back quality the capture threw away — one sentence on why. (Answer provided.)
35. ⭐⭐⭐ Build the grade the format allows. Take your best-quality clip and build a real look on it (warm it, add contrast, protect the skin) — then take an 8-bit 4:2:0 clip of the same subject and try to build the same look. Self-review: how far could you push each before it broke, and what does that gap tell you to shoot next time you know a heavy grade is coming? (Answer provided.)
I. Interleaving with Chapters 1–2
New skills stick when you weave them into old ones. These pull Chapter 1's ideas (the video is not the footage; you shoot for the edit; fix it in pre) and Chapter 2's pipeline (sensor → numbers → reframe headroom) through this chapter's format choices, so the technical never floats free of the storytelling and the workflow it serves. If a format choice can't be traced back to a decision about the finished video and what post will do to it, that's exactly the flaw these exercises exist to surface.
36. ⭐⭐ Format serves story. Take your Project 1 concept (Chapter 1) and its technical baseline (Chapter 2). Now justify your Chapter 3 format choice from the story and the post plan: one sentence connecting your profile/bitrate choice to what the piece needs and what you'll do in the edit. If you can't, your format is arbitrary — fix it. (Model answer provided.)
37. ⭐⭐ The irreversibility check. Chapter 1 taught "you can only arrange what you shot"; Chapter 2 taught the camera test as "fix it in pre." Explain why the recording format is the most irreversible decision in Chapters 2–3, and name one format mistake a pre-shoot test would catch for free. (Answer provided.)
38. ⭐⭐⭐ Trace the pipeline end to end. In about 200 words, trace one shot from light hitting the sensor (Chapter 2) to a graded frame in the edit, naming every stop where information is kept or thrown away: sensor → demosaic → profile → bit depth → chroma → bitrate/codec → container → import → grade. Mark the two or three stops where a bad choice would be permanent. Teaching the whole chain is the strongest proof you own it. (Model answer provided.)
39. ⭐⭐ The three-projects format page. Write the intended capture format (profile, bit depth/chroma, bitrate) for all three portfolio projects — talking-head, documentary short, branded piece — each with a one-line reason tied to the piece's post plan. Self-review: they should not all match; the heavily-graded branded piece wants more than the fast-cut talking-head. If they're identical, you're defaulting instead of deciding. (Model answer provided.)
40. ⭐⭐⭐ Teach "the video is not the footage," format edition. In about 200 words, explain to a beginner why "the video is not the footage" (Chapter 1) is still true even though this chapter is entirely about the footage's technical form. Use bit depth or Log to show that the numbers only matter as room for the decisions post will make. If you catch yourself describing a format as "better" with no "in order to do X in post" attached, you've found the exact habit this chapter exists to break. (Model answer provided.)
When you've done these — especially the banding test (11), the standard-vs-Log grade (13), and the grade-until-it-breaks edit (33) — you can no longer be surprised by what your footage does in post, because you've seen its ceiling with your own eyes. You choose a format on purpose now, for the edit you intend to have. Next, in Chapter 4, we step in front of the sensor to the lens — where focal length and aperture shape the image more than the camera body ever could.