A product comparison video maker is useful when it helps a shopper choose between two models you already sell. The job is not to make one model look cinematic.
The job is to map a real use case to one model, using the same units, the same conditions, and source-backed specifications. TapVid turns existing product materials into a video that can explain that choice clearly, as long as the source pack already contains the facts the video is allowed to show.
If you opened this page looking for a split-screen generator, an avatar, or a cinematic product shot, pause before you upload assets. Those tools can show finish, size, or form. They do not decide which of your own two models a shopper should buy. Write one choice rule first. If you cannot state who should buy each model, you need a specification table, not a side-by-side look. A video that cannot state that rule is a feature parade, not a comparison.
Lock one buyer job before a product comparison video maker generates
Write one sentence before you open any generator:
After this video, a buyer who needs [this job] should pick [Model A or Model B] because [this sourced difference], under [this condition].
That sentence decides the models, the units, the assets, the duration, and the final line. It also creates the review standard. If a viewer still cannot tell which model to buy, the video failed even if the motion looks finished.
Use five fields in the brief:
- Buyer job: one task, such as taking notes on a commute or driving an external display.
- Model pair: two current, publicly sold models from the same brand family.
- Comparable unit: one quantity both models publish in the same unit and condition.
- Source date: the date the specification page was read.
- Choice rule: who should buy each model if the sourced difference holds.
If those fields are empty, do not generate. A comparison video that starts from “show the differences” will invent emphasis. A comparison video that starts from one buyer job can omit every fact that does not change the choice.
A documented TapVid test on the public product-demo page is a Studio outline-and-script review from supplied material. That is the review surface you will use later, in the twenty-minute check. It is not a same-brand comparison output. The outline on this still talks about a magic engine and skipping a timeline. Use the decision table in the next section as the brief. Use this screen only to confirm that an outline and script can stay visible.
Figure: A documented TapVid test keeps an outline and script visible for review. This still is a product-demo brief, not a two-model choice table, and not proof that a shopper will convert.
Build a same-brand decision table with comparable units
Put the two models in one table before you write a script. Every row needs a buyer use case, a comparable unit, a sourced value for each model, a condition, a source date, and a rule. If either model is missing a value, leave the cell blank and block the scene.
The comparison below is a worked example, not a TapVid customer case. It uses Apple’s official technical specification pages for the 11-inch iPad (A16) and the 11-inch iPad Air, read on 6 September 2026. The pair is useful because several headline numbers look similar until the units and conditions are held still.
Buyer use case: Everyday carrying
- Comparable unit: Mass
- 11-inch iPad (A16): 477 grams (1.05 pounds)
- 11-inch iPad Air: 464 grams (1.02 pounds)
- Condition: Wi-Fi models
- Source date: 6 September 2026
- Choice rule: If bag weight is the constraint, choose the lighter Air.
Buyer use case: Bag thickness
- Comparable unit: Depth
- 11-inch iPad (A16): 7 mm (0.28 inch)
- 11-inch iPad Air: 6.1 mm (0.24 inch)
- Condition: Official size spec
- Source date: 6 September 2026
- Choice rule: If slim fit is the constraint, choose the thinner Air.
Buyer use case: Reading and notes
- Comparable unit: Diagonal display, measured as a rectangle
- 11-inch iPad (A16): 10.86 inches
- 11-inch iPad Air: 10.86 inches
- Condition: Official display note
- Source date: 6 September 2026
- Choice rule: Display size does not decide this pair. Use another row.
Buyer use case: Web or video session
- Comparable unit: Battery energy and claimed runtime
- 11-inch iPad (A16): 28.93 watt-hours; up to 10 hours
- 11-inch iPad Air: 28.93 watt-hours; up to 10 hours
- Condition: Wi-Fi web or video
- Source date: 6 September 2026
- Choice rule: Battery claims do not decide this pair. Do not invent a winner.
Buyer use case: Moving large files
- Comparable unit: USB data rate
- 11-inch iPad (A16): USB 2.0, up to 480 Mb/s
- 11-inch iPad Air: USB 3, up to 10 Gb/s
- Condition: USB-C port spec
- Source date: 6 September 2026
- Choice rule: If transfer speed matters, choose the Air.
Buyer use case: Desk monitor work
- Comparable unit: External display support
- 11-inch iPad (A16): Up to 4K at 60 Hz
- 11-inch iPad Air: Up to 6K at 60 Hz
- Condition: One external display
- Source date: 6 September 2026
- Choice rule: If a higher-resolution monitor is required, choose the Air.
Buyer use case: Future storage headroom
- Comparable unit: Maximum published capacity
- 11-inch iPad (A16): 512 GB
- 11-inch iPad Air: 1 TB
- Condition: Official capacity list
- Source date: 6 September 2026
- Choice rule: If the buyer needs the 1 TB option, choose the Air.
Buyer use case: Pencil workflow
- Comparable unit: Supported Pencil models
- 11-inch iPad (A16): Apple Pencil (USB-C) and Apple Pencil (1st generation)
- 11-inch iPad Air: Apple Pencil Pro and Apple Pencil (USB-C), with hover
- Condition: Official accessory spec
- Source date: 6 September 2026
- Choice rule: If Pencil Pro or hover is required, choose the Air.
Apple publishes both inch and millimeter values, and both pound and gram values. Use one column unit for both models. Do not put grams on one side and pounds on the other. Do not compare a Wi-Fi weight with a cellular weight. Do not treat “up to 10 hours” as a measured test result. It is Apple’s claimed runtime on the specification page, under the stated activity.
The United States Federal Trade Commission states that advertisers need a reasonable basis for objective claims before those claims are disseminated. The policy covers express claims and implied claims. A split screen that makes one model look faster can imply a performance claim even if the voiceover never says “faster.” Company counsel still decides what the company may publish. The production rule is narrower: if the table cannot support the implication, cut the scene.
Figure: A same-brand decision table keeps the use case, unit, sourced difference, and model rule in one row. Empty cells block the scene.
This table also shows what not to dramatize. The two 11-inch models share a 10.86-inch rectangular diagonal and the same 28.93 watt-hour battery rating. A video that spends twenty seconds on “bigger screen” or “longer battery” would be inventing a difference the source pages do not give. The useful differences are mass, thickness, USB rate, external display support, maximum capacity, and Pencil support.
Turn each material difference into a use-case rule
A comparison video should end with assignment, not a trophy. The last scene should say who should buy each model. If you cannot write that sentence from the table, you do not have a comparison yet.
Use this pattern for every material row:
- Name the buyer job in one clause.
- Show both sourced values with the same unit.
- State the condition and the read date.
- Assign a model.
- State the leftover case: the other model is still the fit when that job is not the buyer’s job.
For the iPad pair, a clean close is:
- Choose the 11-inch iPad (A16) when the buyer wants the current 11-inch family at the lower specification set, and the table rows that differ do not match their job.
- Choose the 11-inch iPad Air when the buyer needs the lighter and thinner body, USB 3, 6K external display support, the 1 TB option, or Pencil Pro with hover.
Do not add “better for most people.” That phrase hides the job. Do not add a score out of ten unless you publish the scoring method and the source for every point. A score that is not in the source pack is a new claim.
Keep competitor names out of this video. A same-brand selector and a competitive battle card are different jobs. TapVid’s public competitive-comparison feature page describes battle cards, win indicators, and competitor variants. That is a different brief. Mixing the two jobs usually produces a video that attacks a rival and still fails to tell a shopper which of your own models to buy.
Map scenes to source assets, not cinematic contrast
Once the table is stable, build a scene list from the rows that actually change the choice. One material difference gets one scene. The scene must carry four attached objects:
- The model image or official product visual for each side.
- The exact specification string, including unit.
- The condition label, such as “Wi-Fi model” or “one external display.”
- The source date.
If any of those four objects is missing, the scene is blocked. A generated lookalike, a last-season colorway, or a stock desk shot cannot stand in for the model on sale.
Cinematic comparison tools often optimize for matched lighting and a visible shape difference. That can be useful for finish, size, or form. It is a poor default for specification-led choice. If the important difference is 480 Mb/s versus 10 Gb/s, a slow studio orbit does not carry the fact. Put the numbers on screen, in the same unit, next to the correct model.
Owned TapVid product pages describe a source-pack workflow: add a product URL, PDF, script, screenshots, or approved claims; review the outline and scene mapping; keep names, numbers, prices, and required wording literal; then rerun only the affected scene. That workflow fits a same-brand comparison when the source pack already contains both model records. It does not approve the claims for you. It does not record a live screen. It does not prove that a generated product shot is the SKU on the page.
Figure: A comparison scene is ready only after units, context, assets, implications, and the choice rule all pass.
A practical scene order for a two-model video is:
- Job: the buyer task, spoken in the first five seconds.
- Pair: both current model names, on screen, with no nickname.
- Shared facts: the rows that do not decide the pair, stated once so you do not fake a difference later.
- Deciding rows: two to four sourced differences, one scene each.
- Rule: who should buy each model.
- Next step: the product page, configurator, or sales conversation that matches the shown pair.
For most product-family pages, keep the deciding rows to four or fewer. More rows usually mean you have not chosen a buyer job. A full specification dump belongs on the comparison chart, not in the voiceover.
Review the finished comparison against the source pack
Run a twenty-minute source-pack check before you pay
You do not need a finished comparison video to decide whether a generator is safe for this job. Use the two specification documents you already trust.
- Save the official spec page or PDF for Model A and Model B. Write the source date on the brief.
- Complete the five-field sentence and the decision table first. Leave empty cells empty.
- Put both records in one source pack: the two specs, the approved product images, and the brief sentence.
- Open the tool only far enough to read the outline, script, or on-screen labels. Stop before you treat the render as publishable.
- Score four observations. Do not score a conversion rate.
- Locked numbers match the source string.
- Empty cells were not filled with a guessed winner.
- Shared values were not staged as a win.
- Captions or motion did not add “faster,” “stronger,” or “worth it.”
- If any check fails, do not buy more credits to “fix it in the edit.” Fix the source pack, or pick a workflow that can hold the table.
This check does not prove purchase lift. It only tells you whether the tool will hold the facts you already have. For step 4, use the same kind of outline-and-script review surface shown earlier. Do not treat that product-demo still as a finished model-choice video.
Review is not a taste pass. After that check, read the finished video against the table, then against the live specification pages.
Check these items in order:
- Identity: each model name, image, and colorway matches the current sellable record.
- Units: both sides use the same unit and the same condition.
- Literal facts: narration, captions, and labels match the source string character for character where the number is locked.
- Implications: motion, music, and framing do not add “faster,” “stronger,” or “worth it” unless the table supports that wording.
- Date: any time-sensitive value, including price if you choose to show it, carries a visible as-of date.
- Rule: the final sentence assigns both models, not only the one you prefer to sell.
TapVid’s public product-demo page says product names, model numbers, parameters, prices, approved claims, and source images stay outside generative rewriting and are rendered from the supplied material. Use that as a review promise, not as a substitute for a human check. A scene-level revision is useful when one caption or one number changes. It is not a claim that the first render is ready to publish.
Current public pricing, read on 6 September 2026, lists Pro at $49 a month for 2,000 credits, Max at $99 a month for 6,000 credits, and Ultra at $199 a month for 14,000 credits. Rendering is described as 3 credits per second. Those figures describe generation cost. They do not describe the cost of building the decision table, collecting assets, or reviewing claims. Do not treat a credit price as the cost of an accepted comparison video.
If a specification changes after the video is published, the affected scene is stale. Update the table, replace the scene, and keep the unchanged scenes only if they still match the current source. A comparison video has a maintenance cycle. A cinematic product shot does not remove that cycle.
Frequently Asked Questions
What should a product comparison video maker do first?
It should lock one buyer job, two current models, and a table of comparable units before it generates motion. If the tool starts by inventing a side-by-side look, you still have to supply the facts that make the look honest.
I came here for a split-screen or avatar tool. Is this the wrong page?
Only if your job is a visual face-off with no specification decision. If shoppers must choose between two models you already sell, write the five-field brief first. Then decide whether a generator, a live shoot, or a static chart is enough.
Can I compare my own two models instead of naming a competitor?
Yes, and that is usually the safer commercial job on a product family page. A same-brand selector tells a shopper which of your models fits their use case. A competitor video is a different brief, with a different legal and evidence load.
How many specifications should the video include?
Use the rows that change the choice. Two to four deciding differences are enough for most family-page videos. Shared values can be stated once. A full specification list belongs in a chart the viewer can inspect after the video.
Do I need the same units on both sides?
Yes. Mass in grams must face mass in grams. Battery energy in watt-hours must face watt-hours. A Wi-Fi weight must not face a cellular weight. If the source pages use different units, convert both sides or omit the row.
Can I use generated product shots in the comparison?
Only if the generated frame is still the approved sellable model and the specification overlay comes from the source pack. A prettier lookalike is a new product claim. If recognition matters, use the supplied product image.
What should I check in twenty minutes before I pay?
Upload or attach the two official spec records you already trust. Read the outline, script, or labels. Pass only if locked numbers match, empty cells stay empty, shared values are labeled shared, and no extra superiority wording appears. A pass is not a conversion result.
Does a public Studio screenshot prove a comparison video will convert?
No. The public product-demo still shows a reviewable outline and script from supplied material. It does not measure purchase rate, click-through, or return rate. Conversion claims need their own evidence.
What if the two models share a headline number?
Say so. In the iPad example, both 11-inch models list a 10.86-inch rectangular diagonal and a 28.93 watt-hour battery. Those shared values should be labeled as shared. They should not be staged as a win.
When should I avoid making the video at all?
Avoid it when you cannot source both models, when the important difference is not published, when legal review has not approved the implication, or when the buyer job still points to more than two models. A blocked brief is better than a confident video with an empty cell.
