Chapter 36 — Exercises
Delivery is a doing skill wearing a checklist's clothes. You do not get good at it by understanding codecs; you get good at it by exporting real files, watching them fail in small ways, and fixing the settings that caused it. Do these with whatever editor you have — DaVinci Resolve is the book's free reference, but every one of these translates to any NLE, and several need only a phone. Take the starred and Settings Drill items seriously: they are the exact decisions a paying client will one day judge you on.
Exercises are graded by effort and depth:
- ⭐ Warm-up — a few minutes; locks in a concept.
- ⭐⭐ Core — the real practice; do as many of these 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 collected in Answers to Selected Exercises at the back of the book. Try each one before you read its answer — a wrong export you diagnosed yourself teaches more than a correct setting you only read about.
A note on numbers: several exercises reference bitrates, loudness targets, and platform specs. Every such number is typical, at the time of writing — treat it as a shape to reason with, and for anything real, confirm the current figure against Appendix H (Delivery & Codec Reference) or the platform itself. Knowing that a spec exists and changes is the actual skill.
A. Reading the file (Eye & Ear Training)
These need no camera and barely any software — just attention to files you already have.
1. ⭐ Name the two halves. Find any three video files on your devices. For each, write down its container (the extension) and, if you can tell from the filename, camera, or a media-info tool, its codec. Then answer: for which files could you not tell the codec from the extension alone, and why does that prove Chapter 3's box-vs-contents point? (Answer provided.)
2. ⭐ Watch the re-encode. Take one video you made, upload it privately to a platform (or note a past upload), and compare the platform's played-back version to your original file on the same screen. Write one sentence on what the platform's re-encode changed — sharpness, banding, color, loudness. (Answer discusses what to look for.)
3. ⭐⭐ Spot the caption type. Over one day, watching video normally, find three examples of burned-in captions and three of sidecar/closed captions (the toggleable kind). For each, say why that maker likely chose that type given where the video lives. (Answer provided.)
4. ⭐⭐ The loudness jolt. Watch a playlist or feed and notice each time a video is jarringly louder or quieter than the one before it. Note two examples. Given what Chapter 33 and this chapter say about platform loudness normalization, what did the too-loud maker probably get wrong at delivery? (Answer provided.)
B. Reading settings and specs (Read the Spec)
5. ⭐ Read FIGURE 36.1. Without looking back, list the seven decision areas the generic export dialog contains (filename/location count as one). For each, write one phrase on what goes wrong if you set it carelessly. (Answer provided.)
6. ⭐⭐ Read the preset. Look again at FIGURE 36.2 (the "YouTube 1080p" preset). It says "upload a clean, generous file because the platform re-encodes." Explain, in your own words and to a skeptic, why uploading a bigger file than the platform will keep is not a waste. (Model answer provided.)
7. ⭐⭐ Decode a real spec. Find the current published upload/delivery recommendations for one platform you actually use (its own help pages). Rewrite them as a filled-in copy of the FIGURE 36.2 preset for your typical resolution. Note the date you looked it up — because it will change. (Guidance provided.)
8. ⭐⭐⭐ Two specs, one table. Pick two very different destinations (e.g., a vertical social platform and a film-festival or broadcast spec you can find). Build a small comparison table with rows for aspect, resolution, fps, codec, container, bitrate, loudness, and captions. Where the two disagree most, write one sentence on why that destination is different in kind. (Model answer provided.)
C. Delivering real files (Deliver This)
This is the heart of the chapter. Export things and watch them.
9. ⭐ Your first deliberate export. Take any finished or near-finished clip and export it as an H.264 .mp4 that matches its timeline resolution and frame rate, at a generous quality. Then play the exported file in a plain media player — not your editor. Confirm it plays, looks like your timeline, and runs full length. (Model self-check provided.)
10. ⭐⭐ One source, two specs. Do the §36.3 🎬 On Set: export one short piece twice — a 16:9 web deliverable (H.264 .mp4, sidecar .srt) and a 9:16 vertical social deliverable (1080×1920, burned-in captions, subject in the safe zone). Build each to its own spec; do not reuse settings. Play both on a phone. Which field explains any quality difference from your timeline? (Answer discusses.)
11. ⭐⭐ Make and keep a master. Export a finished piece as a master — ProRes 422 HQ or DNxHR HQX in a .mov if you can, or a high-bitrate H.264 if storage is tight. Note its file size, and compare it to a normal delivery export of the same piece. Write the ratio. Why is that size correct for a master and wrong for a deliverable? (Answer provided.)
12. ⭐⭐ Derive, don't photocopy. From the master you made in Exercise 11, produce a small web deliverable by transcoding from the master. Separately, produce another by exporting from the timeline. Compare the two on screen. Then (for the lesson) make a third the wrong way — export a new deliverable from the first small deliverable — and compare all three. Which looks worst, and what is that damage called? (Answer provided.)
13. ⭐⭐⭐ The full delivery package. Take one project and produce a complete, professional delivery: (a) an archived master, (b) a 16:9 web deliverable with a sidecar .srt, (c) a 9:16 social deliverable with burned-in captions, and (d) a one-line "delivery note" naming what each file is and its spec. Run FIGURE 36.6's checklist on every file. This is a real client delivery in miniature. (Model package described.)
14. ⭐⭐⭐ 4K down to 1080. If you have any 4K footage, finish a short piece on a 4K timeline, then deliver both a 4K master and a 1080p web deliverable from it. Compare the 1080p delivery to a version you make by first exporting 4K and then re-exporting that to 1080p. Which path is cleaner, and why does deriving from the high-res source beat chaining exports? (Answer provided.)
D. Diagnosing broken deliveries (Fix the Export)
Each of these describes a delivery that went wrong. Name the cause and the specific fix.
15. ⭐ The judder. A creator cut a music video on a 24 fps timeline, then exported it at 30 fps because that setting was left over from a previous job. On playback, all the motion stutters faintly. Name the cause and the one-field fix. (Answer provided.)
16. ⭐⭐ The eight-second film. An editor exports a five-minute documentary and gets a perfect-looking 8-second file. Nothing is corrupt. What did they do, and where in the dialog? (Answer provided.)
17. ⭐⭐ The mush. A social clip looks razor-sharp on the editor's timeline but soft and blocky once uploaded. The editor exported at a very low bitrate "to keep the file small for upload." Explain the two-part reason it looks bad (their export and the platform's re-encode) and the fix. (Answer provided.)
18. ⭐⭐ The silent scroll. A vertical video gets terrible retention; viewers drop in the first three seconds. It has no captions, and analytics suggest most views are sound-off. Diagnose, and prescribe the exact delivery fix — including which kind of captions and why. (Answer provided.)
19. ⭐⭐⭐ The rejected broadcast file. A freelancer delivers a documentary to a broadcaster as a 1080p H.264 .mp4 mastered to -14 LUFS, and it is rejected. Give at least three distinct spec areas that are likely wrong for a broadcast delivery, and state the correct process that would have prevented all of them. (Model answer provided.)
20. ⭐⭐⭐ The photocopy of a photocopy. A client comes back three times over a month — "make it vertical," "now a square one," "now a small one for email" — and each new version looks worse than the last. Explain exactly what the editor did wrong each time, name the phenomenon, and describe the workflow that would have kept every version first-generation. (Answer provided.)
E. Settings drills (Settings Drill)
Given each scenario, write the export settings you would use and the trade-offs. This is the chapter's core skill; do several.
21. ⭐ The safe default. A friend needs "a video file that will just play on anything I send it to." Write the container, codec, and rough bitrate approach you would give them, and one sentence on why this pair is the safe universal answer. (Answer provided.)
22. ⭐⭐ Four destinations. For each, write container + codec + aspect/resolution + caption approach: (a) a 1080p talking-head for YouTube; (b) a 9:16 teaser for TikTok/Reels; (c) an archival master of a wedding film you'll keep for a decade; (d) a client who wants "the file for our website and maybe the lobby TV." (Model answers provided.)
23. ⭐⭐ Bitrate judgment. You are exporting two 1080p pieces to the same platform: one is a locked-off interview against a plain wall; the other is a fast-moving, confetti-filled event highlight. Should they get the same delivery bitrate? Reason it out using Chapter 3's "what's hard to compress" idea, and say which encoder mode (constant-quality vs restrict-to-bitrate) you'd pick and why. (Answer provided.)
24. ⭐⭐ The storage-squeezed master. A phone-first creator has almost no free space but wants to "keep a master" of an important piece. They cannot store a giant ProRes file. Write the compromise you'd advise and defend it — what makes a high-bitrate H.264 an acceptable master here, and what do they give up? (Answer provided.)
25. ⭐⭐⭐ H.264 vs H.265 decision. You have a 4K travel film to put online. Write the case for delivering H.265 and the case for H.264, name the specific factors that decide it (file size, decode/compatibility, the platform's re-encode), and make a recommendation with your reasoning. (Model answer provided.)
26. ⭐⭐⭐ The whole spec, from a brief. A client brief (Chapter 22) says: "A 90-second brand film for our homepage hero banner (autoplaying, muted) and a cutdown for Instagram Reels." Write the complete delivery plan — every file you'd produce, its spec, its caption treatment, and why the muted-autoplay detail changes your captions and even your edit. (Model answer provided.)
F. Captions and accessibility (Deliver Inclusively)
27. ⭐ Write an SRT by hand. For any 15-second clip, hand-write a small .srt sidecar file in a plain text editor: number, 00:00:01,000 --> 00:00:03,500 timecode line, the caption text, a blank line, repeat for 3–4 lines. Load it against the video in a player that supports subtitles. Why is a format this simple so universal? (Answer provided.)
28. ⭐⭐ Auto then fix. Auto-generate captions for a clip that includes a proper name, a number, and some background noise. Count the errors before you correct them. Write down the three worst mistakes. What does this exercise prove about shipping uncorrected auto-captions? (Answer discusses.)
29. ⭐⭐ Burned-in vs sidecar, decided. For each, choose burned-in or sidecar and justify in one sentence: (a) a TikTok cooking clip; (b) a YouTube tutorial you want multi-language; (c) a broadcast documentary; (d) your archived master. (Answer provided.)
30. ⭐⭐⭐ The full accessibility pass. Take one finished deliverable and run a delivery-time accessibility review: captions present and correct? on-screen text legible on a phone and held long enough? any essential meaning carried by color alone? any risky flashing? Write the punch-list of what you'd change, and fix at least the captions. (Model punch-list provided.)
G. A first taste of the edit meeting delivery (Cut This / Interleaved)
31. ⭐⭐ Recreate the master→deliverables flow. Draw FIGURE 36.4 from memory for one of your projects: the master at the top, the specific deliverables you actually need beneath it, with the correct spec noted on each. Pin it where you edit. (Guidance provided.)
32. ⭐⭐ Interleave — capture meets delivery. Take the Café Scene (or any piece you graded). Trace one quality decision all the way through: what bit depth you captured it in (Chapter 3), how hard you graded it (Chapter 32), and how that combination will behave when the delivery codec compresses it again here. Where in that chain is the quality ceiling set? (Answer provided.)
33. ⭐⭐⭐ Interleave — the whole pipeline, one sentence each. In a single sentence per stage, describe how one choice in each of these chapters lands in your final delivered file: Chapter 2 (frame rate), Chapter 3 (bitrate/bit depth), Chapter 23 (aspect/safe zones), Chapter 31 (correction), Chapter 33 (loudness). This is the book's spine, seen from the exit. (Model answer provided.)
H. Synthesis
34. ⭐⭐ Build your permanent checklist. Adapt FIGURE 36.6 into your own delivery checklist — add any lines specific to your gear, editor, or the platforms you actually use; cut any that don't apply. Save it where you'll see it at every export. Use it on your next real delivery and note which line first saves you. (Guidance provided.)
35. ⭐⭐⭐ Teach it back. In about 200 words, explain to a beginner who "just exports with the default settings" why the export is "the last place to ruin a great edit." Use the codec/container distinction, the frame-rate-match rule, the re-encode principle, and the master discipline — with one concrete example of each going wrong. Teaching it is the strongest test that you own it. (Model answer provided.)
36. ⭐⭐⭐ Deliver the portfolio. If you have done the Production Checkpoint, audit it: do all three projects have an archived master and captioned deliverables, each QC'd on a phone and a big screen? For any gap, close it. When every project is a set of correct, captioned, spec-matched files, you have finished the delivery craft — and Project 3 is truly delivered. (Self-audit rubric provided.)
I. Recreate it and five ways (Recreate It / Five Ways)
37. ⭐⭐ Five ways to deliver one clip. Take one 20–30 second clip and deliver it five ways, changing exactly one dimension each time: (a) 16:9 web H.264; (b) 9:16 vertical, burned-in captions; (c) a ProRes/high-bitrate master; (d) a deliberately tiny file for email (efficient codec, low target size); (e) a 1:1 square. Watch all five. Which trade-offs (quality, size, aspect, captions) did each force, and which was hardest to keep looking good? (Answer discusses the trade-offs.)
38. ⭐⭐ Recreate a spec from a file you admire. Find a video file you think looks great (a downloaded trailer, a creator's uploaded master, a sample clip). Open it in MediaInfo (or your editor's clip info) and read out its container, codec, resolution, frame rate, bitrate, and audio format. Then write the export preset that would recreate that spec. What, if anything, surprised you about how it was delivered? (Guidance provided.)
39. ⭐⭐⭐ Recreate a platform's look, then beat it. Upload a clip to a platform, then download the platform's re-encoded version. Study where it degraded (banding, softness, loudness). Now go back to your source and export a better upload candidate — higher bitrate, cleaner — and compare its re-encoded result to your first. Prove to yourself that a better source survives the same re-encode better. Write the one-sentence conclusion. (Answer discusses.)
40. ⭐⭐⭐ The delivery capstone. Take your most finished project and produce its complete professional delivery set from a single master: an archived master, a captioned 16:9 web file, a captioned 9:16 social file, and a one-page "delivery sheet" listing every file, its exact spec, its caption treatment, and where it's meant to live. Run the full checklist (FIGURE 36.6) on each. This is exactly what you'd hand a paying client — and exactly what the Production Checkpoint asks of all three projects. (Model delivery sheet described.)
When you have done these — especially the Deliver This and Settings Drill items — you can be handed any finished timeline and any destination and produce the right file, captioned, at full quality, the first time. That reliability is worth more to a client than any single beautiful shot. Next, in Chapter 37, we make sure the masters you just rendered are never lost.