A two-person casting exercise

Crazy Frog AI Video, directed by your cast

One person escapes on an impossible ride. The other follows on a red motorcycle. Turn two portraits into the cast of a Crazy Frog AI video, using a supplied chase clip to guide the action.

Kize gives each person a separate photo slot. Decide who leads, watch the motion reference, then shape the pairing with a short direction of your own.

Choose your two riders
128/10000
Seedance 2.0
Required Credits—

Add the required reference images to continue.

Loading video models…

15-second chase reference

A familiar chase. A cast you choose.

Number 1 matches a woman in lavender to the invisible-bike leader; number 2 matches a man in cobalt to the red-motorcycle follower.
Casting diagram: photo 1 leads the chase; photo 2 follows on the motorcycle. AI illustration, not a generated video result.

Photo 1 sets the escape

The first portrait belongs to the person at the front. Their exaggerated crouch and invisible handlebars create the visual joke. Choose the person whose expression or personality makes that escape amusing.

Photo 2 supplies the pursuit

The second portrait identifies the rider on the visible motorcycle. Together, the two roles make the scene understandable even before a caption explains why one person is chasing the other.

Identity and motion have separate jobs

Your portraits describe who appears; the supplied video describes how the chase moves. You do not need action photography or a matching riding pose. The model still has to interpret both sources, so inspect recognition throughout the completed clip.

A focused direction is easier to judge

Use the prompt for a simple performance note: playful determination, a surprised glance or a lighthearted rivalry. Keep the two assigned identities clear. Several competing scene changes can make it harder to judge whether the casting worked.

Take the cast from portraits to a moving scene

Build the sequence in four deliberate passes. The numbered storyboard below explains the workflow; the written steps remain readable on smaller screens.

Four numbered panels show separate portraits, role assignment, the two-person highway chase and a filmstrip review.
Production map: 1 select portraits, 2 assign positions, 3 generate the chase, 4 review identities in motion. AI workflow illustration.

01 / Select one face for each slot

Upload a clear individual portrait to each photo input. Check the labels before proceeding: the first slot is the leading rider and the second is the pursuing rider. Leave unrelated faces out of the reference images.

02 / Check the scene before directing it

Play the 15-second reference beside the form. Notice the spacing between riders, the visible motorcycle and the changes in camera distance. Add a concise prompt that works with this action rather than asking for a different plot.

03 / Read the current generation quote

Open the available video settings and check the credit amount shown for your selection. Start generation only when the portraits, roles and settings look right. Selecting files, changing the prompt and swapping roles are preparation steps.

04 / Review the whole performance

Follow progress in History, then inspect the completed video from start to finish. Look for each person in closer and wider shots, check role continuity and download the result when it meets your needs. A single good opening frame is not enough.

Recast the same scene and change the punchline

Side-by-side compositions swap a lavender-clad woman and cobalt-clad man between the invisible leading ride and the real red motorcycle.
The setting stays comparable while the two people trade positions. This is a casting illustration; each generated version is a separate task.

Try the less obvious lead

If the reliable planner usually follows, put them in the escape role. A small casting reversal can tell a different story without adding another character. Use Swap roles after both portraits are ready, then read the labels again before generating.

Change one decision per comparison

Keep your wording and format steady when testing a role reversal. That makes the cast change easier to evaluate. Swapping the inputs alone does not run a task; generating the alternative uses its own displayed credit quote.

Give the pursuit a reason your audience recognizes

A Crazy Frog scene works well with a small, familiar disagreement. Pick a premise that fits the people you cast; the video supplies the chase while your caption provides the context.

Four illustrated adult pairs act out an invisible-bike chase and red-motorcycle pursuit: game rivals, dancers, travel companions and podcast hosts.
Four story prompts, not preset locations: game night, rehearsal, travel plans and a podcast debate. The generator follows the supplied chase reference.

Game night: the last victory lap

Cast the friend who celebrates every win as the escaping rider and their rematch challenger as the follower. A short caption about the next round gives viewers a reason for the pursuit without requiring extra events in the clip.

Rehearsal: one more take

Pair two dance collaborators and let the leader play the person who thinks practice is finished. The pursuer becomes the teammate asking for another run. Their recognizable relationship carries the joke; a new stage or dance routine is not required.

Travel plans: who has the tickets?

Use two travel companions and build the caption around an ordinary planning mix-up. Decide which person should be holding the imaginary advantage. The highway action stays the reference scene; the travel premise belongs in the story you tell around it.

Podcast debate: the unfinished argument

Put two cohosts on opposite sides of a harmless recurring debate. The one leaving mid-discussion takes the leading slot. Review facial consistency before adding the finished clip to an edit, and check whatever audio you choose to include.

Give the model a face it can read

A sharp, evenly lit portrait is marked with a blue check; motion blur, a hand over the face and a distant full-body photo receive coral crosses.
Input check: clear eyes and visible facial features give a stronger reference than blur, obstruction or a tiny face. AI photo-quality illustration.

Start with a readable close portrait

Choose a single person with visible eyes, natural lighting and a well-defined face. Leave enough detail to recognize them at the size shown in the upload preview. The source photo does not have to match the wardrobe or camera angle of the chase.

Fix the weakest reference first

If a result loses one identity, replace that slot with a sharper photo and keep the other inputs comparable. Avoid changing the entire cast and prompt at once. New attempts can still vary, and a better input does not guarantee every frame will look correct.

Use portraits you are entitled to upload and share. Before publishing the video, review the pairing and context as well as the visual quality.

Before you direct a Crazy Frog AI video

Is this a frog character maker or a two-person chase?

This page is built for a two-person chase inspired by the Crazy Frog riding scene. It uses one portrait for the invisible-bike leader and another for the motorcycle follower. The supplied video is the motion guide, rather than a prompt to design a new frog character.

Can I choose who is in front?

Yes. Place that person in Photo 1. Photo 2 is assigned to the rider pursuing on the motorcycle. Once both photos are present, Swap roles reverses the inputs so you can prepare a different casting.

Where does the chase footage come from?

A fixed technical reference is supplied with the page and previewed beside the form. You do not need to find or upload the chase video. It guides the motion; it is not an example of a result made with your portraits.

Will the illustrations determine my output style?

The illustrations explain the roles, preparation and possible story angles. They are separate from the generation inputs and are not used as additional scene references. The video task uses your portraits, the supplied motion clip and the supported settings.

Which length and format should I expect?

This effect uses a 15-second output. The settings control shows the formats and resolution currently available from the selected model. Check those options and the current quote in the form instead of assuming every format is supported.

When does a role swap spend credits?

The swap button only exchanges the two portrait slots. Credits are associated with starting a new generation at its displayed quote. Two versions with different casts are separate generation tasks, even if the same photos are reused.

How do I check whether both identities survived?

Watch past the first close shot. Compare both faces during turns, wider framing and faster movement, and check that their assigned roles remain readable. If one person is inconsistent, try a clearer reference for that slot before changing several other choices.

Can I keep working after I leave this page?

Use the existing task and History workflow to check your generation again. Reopen the page with the same account or session. Confirm that a task has actually started before leaving, and review the completed output before downloading or sharing it.

Set the cast. Keep the action readable.

Choose two distinct portraits, give each a clear role and let the reference guide the pursuit. For a different kind of project, browse the other Kize video tools or start an image reference first.

Return to the casting form