Feature
When an avatar beats filming: five cases and a sorting test
Both routes make the same finished 9:16 ad and both live in one account. A render is about 393 credits for forty seconds. Filming the same ad is about 114. Here are the five cases where the render is worth that difference, a ten minute test that sorts your script list, and where the rest of your scripts should go.
- A finished forty second rendered ad
- 393 CR
- The same ad filmed on a phone
- 114 CR
- What the render costs against filming
- 3.3x
Case one: the script is a fact nobody wants to read on camera
Pricing changes, shipping cutoffs, policy updates, restock notices, spec comparisons, compliance language.
Nobody volunteers to deliver these and nobody watching cares who delivers them.
The information is the value. The face is the courier.
A rendered presenter carries that perfectly, because there is nothing personal in the sentence to perform.
It also carries it identically in January and in June. That matters for anything that repeats.
This is the strongest case in the list. It is worth 350 credits a minute without argument.
The alternative is not a cheaper ad. The alternative is the ad never getting made, because nobody wanted to say it out loud.
Teams usually discover this backwards. They render the hero ad first and dislike it. Then they never try the announcement they were dreading.
Start with the script nobody wants to read. That is where the render is unambiguously the right buy.
Cases two and three: nobody is available, or nobody should be identifiable
The person who films is on holiday, ill, or in another country, and the ad has to go out on Thursday.
A render costs more per attempt and takes nobody's diary. For that week it is the cheaper option, because the other option is nothing.
The third case is when the face should not belong to a specific person at all.
A stock or image-model presenter removes the permission question entirely.
It also removes the risk that your whole ad library depends on somebody who might leave.
That is a structural advantage and it is rarely mentioned. It only shows up eighteen months later.
Both cases are about availability. Availability decides how many attempts you get to make.
That is the number the whole account is built around, on either route.
Cost per attempt, both routes
The render buys availability. Both routes finish as the same 9:16 file.
- Filmed on a phone, batched, exported114 CR
- Rendered with a generated read393 CR
This is the whole editor
Highlight a phrase and a clip lands on those exact words. No timeline, no keyframes, no layers.
Cases four and five: identical delivery, and a series that must not drift
When you are testing which of six openings works, the delivery should be the only thing that does not change.
A person films six takes. Back come six energies, six lightings and six haircuts.
The winner you find there may be a mood rather than a message. That is an expensive thing to learn late.
A rendered presenter delivers all six identically. The difference in results is the words.
The fifth case is a series recorded across months. Same room, same shirt, same face, no continuity problem.
A filmed series drifts. Rooms change and people change.
Viewers notice at exactly the point you wanted it to feel familiar.
Neither case is about the render looking good. Both are about the render being constant. Filming cannot offer that at any budget.
Attempts needed to expect one winner
Published benchmarks put the winner rate at roughly 5 to 8 percent, from Motion's analysis of 550,000+ Meta ads. Cost per attempt is what decides whether you get there.
The ten minute sorting test
Print your script. Underline every sentence that only means something because a real person said it.
Underline every sentence that needs something held, opened, poured or pointed at.
If nothing is underlined, render it. The rendered version costs more per attempt and loses nothing.
If a couple of lines are underlined, film those and render the rest.
Join them in the editor by highlighting the phrase and dropping the footage in. You direct on the transcript, so it is a highlight rather than a timeline job.
If most of it is underlined, film the whole thing and batch it for 100 credits.
Ten minutes with a pen sorts this better than any comparison table, including the ones on this site.
Run the test on the last five ads you shipped rather than on the one you are about to write.
Most teams find two of the five were pure information and three needed a person. That tells them how to split next month.
Where to send the other scripts
Volume testing. Twenty rendered attempts is over 7,800 credits against roughly 2,280 filmed. Put the bulk on the phone.
Long-form. At 350 credits a minute, ninety seconds is 525 credits of render. It holds nobody longer than forty would have.
Anything tactile. Texture, weight and scale need a hand next to them. There are no hands in a rendered frame.
Founder-led brands. If people buy partly because of who made it, a render removes the reason.
None of those four is a reason to leave. Every one uploads as a source take and batches for 100 credits in the same editor.
You direct it on the transcript, cover phrases with footage, pick captions from six packs and export at 20 credits a minute.
The plain facts on both routes. Output is 9:16. A filmed source take caps at three minutes. Rendered audio caps at five.
There is no timeline, by design. Every lever is a decision about the ad.
One account, two ways in, the same finished vertical file at the end. No presenter tool offers the second way in.
Questions people ask
- Is a render ever better than filming, rather than only more convenient?
- Yes, in two places. Identical delivery across a test, so the words are the only variable. And consistency across months. A filmed series drifts in room, energy and appearance.
- Can I mix a filmed take and a rendered one in the same ad?
- Yes, and it is often the right answer. Film the lines that need a person, render the rest, and place each over the words it belongs to by highlighting a phrase.
- How do I know if my delivery is too flat for a render?
- Read the script aloud and listen for where your own voice does something. If it rises, hesitates or softens, a rendered read will flatten that. Those are the lines to film.
- Who should ignore this comparison?
- Anybody selling something tactile with nobody willing to film. Buy one creator shoot for reusable product footage first, then decide about presenters with real material in hand.
Five cases, one pen, ten minutes. Whichever pile a script lands in, it finishes in the same editor as a 9:16 ad.