dimeclips.

AI character creator guide · November 19, 2026

A character.
A habit.
A good scene.

The raccoon expects to be the strongest one in the gym. The tiny duck has other plans.

Give your original AI character something to want, a recognizable appearance and a situation that tests them. This guide turns that idea into a scene brief you can reuse and a result you can judge.

Whether you are planning an AI influencer or a fictional comedy series, start with one scene before committing to a season.

A raccoon struggles with a purple barbell. A tiny duckling lifts it overhead while the raccoon reacts. Effort sounds and a squeak accompany the action; there is no dialogue.

AI-generated Dimeclips campaign · Omni 1.2 directly on Google Cloud · 8-second source, 9-second edit · 720 × 1280

01 / Decide what stays

Write a character brief
you can recognize.

A name helps you organize files. Visible details help you compare the actual results. Record a few distinctive traits, then keep a separate line for the action that changes in each episode.

Appearance
Stocky grey raccoon, dark eye mask, orange headband. Use concrete features you can check on screen.
Behavior
Very confident before a challenge, quietly bewildered afterward. The reaction is part of the idea.
Visual style
Clay animation, simple indoor set, full body visible. Keep the important action inside the frame.
This episode
A purple barbell and a much smaller rival. The duckling completes the lift the raccoon struggles with.

Reusable planning template

Character: [build, colors, face and one distinctive accessory]. Behavior: [what they usually expect or do]. Style: [medium, setting and framing]. This scene: [one challenge]. Ending: [the visible reaction or reversal]. Keep unchanged: [the features you will compare in every take].

Save this brief alongside your prompts and chosen results. It is a planning document you maintain, rather than a character-lock feature in Dimeclips.

02 / Make the behavior visible

The smallest spotter.

The joke needs three things to remain readable: the same barbell, the size difference and the raccoon’s reaction. A single fixed view gives you a clear way to inspect all three.

  1. Setup · 0–3 secondsThe raccoon strains with the weight. The duckling is already present, so its later action has a setup.
  2. Reversal · 3–5 secondsThe raccoon steps back. The duckling moves under the same bar and lifts it overhead.
  3. Reaction · 5–8 secondsThe duck holds the weight while the raccoon looks toward the camera and its own paws.

Exact selected-take prompt · Omni 1.2 Flash · Text to video · 8 seconds · 9:16 · 720p

Original eight-second vertical clay-animation gym comedy with two charming fictional animals. Locked medium-wide camera, full bodies and whole barbell always visible. A stocky grey raccoon in an orange headband stands center-left behind a purple barbell with two oversized round weights, on a simple indoor gym floor. A tiny yellow duckling waits at screen right. From the first frame through three seconds the raccoon strains comically to lift the bar with both paws, but the bar stays planted on the floor. At three seconds the raccoon releases it and takes one step back to the left, looking exhausted. At four seconds the tiny duckling waddles under the middle of the bar and lifts the SAME rigid purple barbell cleanly above its head with both wings, like an absurd little champion. Hold the duckling with bar overhead from five through eight seconds. The raccoon slowly turns to stare at the camera, then looks at its own paws in disbelief. Strong clear size contrast. Raccoon has two arms and two legs; duckling has two wings and two feet. Same characters, single unmoving camera, consistent simple geometry. Deliberately impossible cartoon strength, no injury. Original audible effort grunts, a rubber squeak on the duck lift, two little foot taps, no dialogue or music or writing or logos.

These timings describe the request. The result differs: the raccoon makes a partial lift, and the bar bends slightly. We kept the take because the smaller character’s overhead lift and the reversal were clear enough for this stylized scene.

This was the third take for the gym concept. The first two used a different exercise gag that did not read clearly enough. The opening caption and one-second dimeclips.com ending were added afterward. This campaign was generated directly on Google Cloud; it is not a recording of the Dimeclips product workflow or proof of matching a character across episodes.

03 / Change the situation

Keep the character.
Test a new challenge.

For a text prompt, keep your appearance and style description together and change the action separately. Google’s video guidance recommends reusable character descriptions and focused scenes. That approach helps organize the request; it does not guarantee the same face, outfit or voice in every generation.

For this raccoon, a next-episode idea could be an oversized jar that the duckling opens effortlessly. That is an untested brief, not a generated continuation. Keep the orange headband, clay style and confidence-to-disbelief reaction, then compare the new output with the gym scene.

If the selected model supports an image input, choose a clear image of the character and check the input requirements before uploading. For image-to-video, Google recommends describing the desired movement instead of repeating everything already visible in the image. A start image, multiple reference images and a continuation video are different input modes; use the controls the chosen model actually provides.

Dimeclips supports individual image and video generations with model-specific inputs. A recurring character still needs review between shots. A finished AI influencer series also needs your editorial choices, disclosure and publishing work; a prompt alone does not run an influencer account.

04 / Choose the take

Review the whole scene.

Identity
Compare face, proportions, colors and accessories at the start, during movement and at the end. For another episode, compare it beside the accepted scene.
Action
Does the intended character perform the action? Does the object remain recognizable? Watch hands, paws and contact with props.
Payoff
Can a viewer understand the ending without your explanation? Check the scene at phone size and allow time for the reaction.
Sound and edit
Listen through the entire cut. Check speech when present, unwanted sounds, abrupt endings and whether captions cover the action.

Record the requested model, input, settings, attempts and reason for keeping or rejecting each result. Separate a prompt that worked once from a repeatable character workflow you have actually tested.

Budget by the model you intend to use. Dimeclips plans have a supported short-clip allowance and separate premium credits. Veo, Omni and generated reference images use premium credits. Allow room for discarded attempts; the number of included generations is not a promise of that many finished episodes. Check the current plans and model allowances before choosing.

For more scene structure, see the prompt guide. If the character is part of an ad, use the ad workflow to keep the offer and each creative variation clear.