CASE-BASED PROMPT GUIDE
Hypit prompt examples: turn an idea into a usable video brief
Use a real phone-stand comparison example to map the source goal, retained structure, replacements, asset files and acceptance criteria into an editable Hypit brief.
Edited by HypitAI · Reviewed
Watch the phone-stand comparison workflow
The video moves from a reference structure to a Chinese product video, then explores presenter, scorecard and language changes. The brief below turns that demonstration into editable fields.
Open video details →State the source-video goal
This case borrows the information structure—two price points, three tests and a scenario-based recommendation. The remix goal is to use your own verified product facts, presenter, narration and footage rather than copying the creator’s brand, media or conclusions.
A reusable brief
Replace each bracket with actual information. Mark unknowns explicitly.
Audience: [who]
One takeaway: [message]
Assets: [files and purpose]
Keep: [brand, lines, structure]
Change: [language, scene, presenter]
Delivery: [aspect ratio, duration, language]
Avoid: [unsupported claims or changes]
First return: [shots, script, missing inputs, budget]
Budget: [cap and approval boundary]Map what stays and what changes
Ask for a section-by-section mapping before production.
| Section | Structural role to keep | Replace with |
|---|---|---|
| Opening | Set up a low-versus-high-price question | Your models, dated sources and real prices |
| Tests 1–3 | Observe differences through three actions | Conditions, footage, results and limitations |
| Scorecard | Summarise observations | Evidence-backed wording, scores or unknowns |
| Ending | Choose by use case | Audience, limitations and accurate CTA |
A completed remix instruction
Replace bracketed filenames and facts with your own verified material. The instruction requests a plan first so missing inputs do not trigger paid generation.
Reference: [reference.mp4]. Borrow only the two-price-points, three-tests and scenario-choice structure.
Audience: [people comparing two phone stands].
One takeaway: [one verified difference].
Keep: four information beats, test-by-test comparison and scenario-based ending.
Replace: presenter, brands, models, prices, test media, narration, scorecard, voice and CTA.
Assets: products.csv, test-plan.md, product-a/*.mp4, product-b/*.mp4, presenter.png and approved-script.md.
Delivery: 9:16 video; attach a date to every price; keep captions clear of the products.
Avoid: unverified superlatives or sales claims; generated imagery must not prove real performance.
First return: mapping, storyboard, script, missing assets, required services and budget. Do not start paid generation without approval.Asset file checklist
Names can change, but the role and evidence source of every input must be explicit.
- reference.mp4: an authorised structural reference.
- products.csv: full model names, price sources and check dates.
- test-plan.md: conditions, actions, recording method and non-comparable factors.
- product-a/ and product-b/: matched product footage and test actions.
- approved-script.md: confirmed narration, pronunciations and CTA.
- presenter.png or presenter.mp4: permitted presenter material, or an explicit missing-input marker.
Request a controlled second version
Keep most dimensions fixed so the change is interpretable.
Change only the first three seconds to a direct audience question. Keep the demonstration, main narration, product facts, ending and output specification unchanged. List what can be reused, what needs new material and any additional cost first.Acceptance table
Do not rely on a completed task status. Record issues at exact moments and correct the corresponding asset or fact.
| Pass | Method | Confirm |
|---|---|---|
| 1 | Play with sound | Complete structure and aligned actions/dialogue |
| 2 | Watch muted | Readable captions, correct models/prices and unobstructed products |
| 3 | Listen without picture | Correct numbers, names, pronunciation and CTA |
| 4 | Compare with evidence sheet | A source for every claim and retained limitations |
Review before production
If the response contains polished prose without a mapping, missing-input list and acceptance method, ask again before production.
- Does every shot have an existing asset or explicit capture plan?
- Do the three tests state conditions rather than only conclusions?
- Do names, numbers and the CTA come from approved material?
- Are missing assets, uncertain charges and required decisions listed separately?