Generate media when it helps Fictional AI presenter; AI-generated narration. Generate media when it helps Welcome to module 7. I'm Alex, a fictional AI presenter for Vidmoat Learning. In this lesson, we will use optional generated assets, reference-led consistency and sound generation with clear limits and alternatives. Let's work through a practical example. Decide whether a new asset is needed A missing shot does not always need an AI generation. If you need to show the real lamp switching on, filming the real switch gives more trustworthy evidence. If you need an abstract illustration of a late-night idea, a generated image may be appropriate. Decide whether a new asset is needed Write the job of the missing asset before opening Generate. Describe the subject, action, framing, lighting and intended placement. Ask for one useful asset first. Large batches can consume credits before you know whether the direction works. Check the available provider, plan and price shown in your account; this course does not require paid generation to complete the main projects. Decide whether a new asset is needed Generated media can invent lettering, change a product feature or create continuity problems. Inspect it before importing it into the final sequence. Do not present a generated demonstration as proof of a real product’s performance. When generation is unavailable, use the included illustrations, a simple title card or another shot you can record yourself. Try it in Vidmoat Here is a way to try this. Identify the exact gap in your story and write the job of the missing shot. Decide whether real footage is necessary to support the claim. If generation fits, open AI Generate and choose image or video. Review the displayed cost and settings; request a single short or still asset first. Inspect the result, then add it to the timeline only if it serves the brief. Make the decision yourself Now pause the lesson and make your own version. List three missing shots for your project. Mark each “film”, “reuse” or “generate”, and explain the choice. Complete one with the lowest-cost suitable method. When you are ready, compare your result with this standard. Every new asset has a defined purpose, and evidence about real products still comes from real material. Keep a character or product recognisable Consistency is a choice. A mood montage may need varied subjects; a three-shot story about one robot needs a recognisable robot. Decide before generating. A text-only request remains a valid way to create a fresh subject. Keep a character or product recognisable For a continuing character, prepare a clean reference image and a short description of the features that must stay. Our original reference is an orange desk robot with a round head, dark face panel, two pale eyes and a small antenna. Keep that description stable while changing only the action or camera view between shots. Keep a character or product recognisable Use Character / product reference in Generate when the selected provider supports it. In this edition, reference-led image and video generation uses Grok; current controls impose limits that differ by media type. Video references are limited to the supported resolution shown in the interface, currently up to 720p. Reference similarity is guidance, not an identity guarantee. Check colour, face, proportions, accessories and background continuity in each result. There is no saved character library to assume; keep your own reference file and brief. Try it in Vidmoat Here is a way to try this. Choose whether this story actually needs the same subject across shots. Attach resources/robot-reference.png using Character / product reference in Generate. Write the character’s fixed features and the action for shot one. Reuse the same reference and fixed features for the next shot; change only the action or framing. Compare results side by side. Reject visible drift before building the sequence. Make the decision yourself Now pause the lesson and make your own version. Plan three shots: the robot notices the lamp, reaches towards it, and sits in the warm light. Generate one if your account allows it, or use the original reference as a still and practise framing. Fill in the continuity checklist for each shot. When you are ready, compare your result with this standard. You can name what stayed consistent and what drifted. You do not claim that all generated shots are identical. Use generated voice and music thoughtfully Generated narration can help when you cannot record a voice, but the script still needs to sound like a person. Read it aloud before generating. Short sentences and concrete words usually survive better than a paragraph packed with slogans. Use the same available voice selection across a series when continuity matters. Use generated voice and music thoughtfully For music, describe the role: a light instrumental bed under speech, a sparse reflective opening, or a short energetic ending. If a generation option supports vocals, say whether you want original lyrics or an instrumental. Do not assume a successful generation request produced the requested words; listen to the output. Use generated voice and music thoughtfully Avoid imitating a real person without appropriate permission. Check the terms that apply to any generated or licensed asset before commercial distribution. Keep a copy of your prompt, selected settings and asset origin. If generation is unavailable or outside your budget, record your own narration and use the supplied practice bed. The editing skills are the same. Try it in Vidmoat Here is a way to try this. Write a narration of roughly 40 to 60 words for a short demonstration and read it aloud. Use the available voiceover generation control, or record your own voice. Listen for incorrect names, odd pauses and an unnatural pace. For an optional music generation, request a clear mood, duration and instrumental or vocal intent. Import the accepted assets, balance them against each other and verify any words by listening. Make the decision yourself Now pause the lesson and make your own version. Read the supplied lamp narration naturally and time it. If it runs long, remove one idea instead of rushing every word. Use recorded or generated narration and compare it with the silent version. When you are ready, compare your result with this standard. The narration fits the available time without sounding hurried, and the soundtrack supports rather than competes with it. Keep one useful decision Before moving on, write down one edit you made, why you chose it, and what changed for the viewer. The course workbook and practice files are available with this lesson. Continue when you can explain your decision in your own words.