All Arcads features

Cast AI actors for your ads from a single description

Describe the person you want on camera — age, wardrobe, energy, the room they are standing in — and Arcads renders a photoreal presenter holding your product, ready for a paid test.

Cast someone now

Step one renders the AI actor still from your description, and step two plays a motion preview of that same still.

Describe your product in the prompt — colour, packaging, label — and your AI actor will hold it.

Quick human check before generating.
Free · no sign-up · rate-limited to keep it free for everyone

Casting used to be the slow part

A traditional shoot spends its first week on people. You brief an agency, sift through reels, argue about whether the 34-year-old reads too corporate, book a day, and hope the light holds. By the time a face is locked, the campaign angle you were testing has already been replaced. Arcads collapses that into a sentence. You write who should be on screen and what they are doing with the product, and the ai actors engine renders them. If the read is wrong, you change three words and render again — the cost of a recast is one more generation, not one more booking.

That changes what you are willing to test. Instead of one presenter carrying every ad, you can run a 24-year-old and a 45-year-old against the same script, or the same person in a bathroom versus a gym, and let the numbers pick. Most teams find the winning combination is not the one they would have booked. Cheap recasting is what makes that discoverable.

Why generated presenters work

Any face, any brief

Age, build, wardrobe, hair, expression and setting are all things you write rather than things you search for. If your product sells to nurses on a night shift, you can put a nurse on a night shift in the frame instead of settling for the nearest stock approximation.

No booking, no releases

There is no talent to schedule, no day rate, no usage window that expires halfway through a campaign, and no re-clearance when you want the same creative in a new market. The presenter exists because you described them, so the creative stays usable for as long as the offer runs.

Recasting costs one prompt

Swapping the person is a text edit. That makes it practical to hold everything else constant — same product, same room, same framing — and isolate what the face alone is worth. Very few teams have ever been able to run that test cleanly before.

From description to cast presenter

Three steps, and the whole thing happens in the browser.

  1. Write the casting note

    Treat the prompt box like a call sheet. Name the age range, the wardrobe, the room, what they are holding and where they are looking. Specific casting notes produce specific faces; vague ones produce generic ones.

  2. Choose the frame and render

    Pick the ratio your placement needs — 9:16 for vertical feeds, 4:5 for the Meta feed, 16:9 for pre-roll — clear the Turnstile check, and Arcads renders the AI actor still in that shape.

  3. Judge the read, then recast

    Play the motion preview to see how the presenter carries themselves, then rewrite the casting note if the energy is off. Two or three passes usually lands a face you would have paid to book.

Questions about generated presenters

No. Every presenter is generated from your description, so there is no performer being represented and no likeness being reused. That is why there is no booking step and no usage window — the face was assembled for your prompt rather than filmed on a set and licensed to you afterwards.

Write the call sheet, skip the casting call

The presenter you need is one carefully written sentence away. Describe them, render, and see whether the face you assumed would work actually does.