How to Make Malfoid-Style AI Videos with Idoli

If Malfoid-style clips have been appearing in your X or Instagram feed, you may already have an idea for your own scene: a platinum-haired heroine, a tense exchange, and a confession neither character quite manages to say aloud. But getting from a reference image to a convincing video can be harder than choosing a storyline.

You might try different tools and prompts without getting the performance you want. A realistic reference image can also run into a real-person-related restriction when used directly, interrupting the project before the scene is generated.

Idoli’s approach is to create and save a character first, then reference that character in image or video mode. For supported projects, this is the character-based route Idoli provides for the restriction associated with directly using a real-person photo. The saved character also becomes a reusable subject for your next scene.

Step 1: Choose how to build your Malfoid-inspired character

There are two starting points. Use image-generation mode if you want to design your own interpretation. Use character-generation mode if you already have a suitable reference and want to build a saved character from it.

In either case, make the interpretation your own. Reusing a reference does not make the source artwork yours; use images you created or have permission to use. The examples below describe adult characters and a fan-created romance scene.

Step 2A: Design your own version in image mode

Open image-generation mode. Start from a text description, or include an authorized reference if you want visual guidance. Describe the appearance you want rather than relying on the character’s name alone.

For example:

Create a realistic waist-up portrait of an adult woman in her early twenties with long platinum-blonde hair, pale skin, gray eyes, and a composed, slightly proud expression. She wears a black academic robe with emerald lining over a white shirt and a green-and-silver striped tie. She stands in a stone castle corridor, softly lit by warm wall lamps and cool light from an archway. Natural skin texture and realistic fabric. A live-action cinematic portrait, not an anime illustration.

Use the result to decide which features define your version. Image mode remembers the conversation, so you can continue with a targeted change:

Make her expression more reserved. Keep her facial features, hairstyle, and clothing the same.

Once you have an image you like, use it as the reference in character-generation mode, generate the character, and save it. An image result is the starting material for this route; do not assume it has automatically become a saved character.

Step 2B: Build a character from a reference image

If your reference already has the look you want, begin in character-generation mode. Upload the authorized reference and enter:

Generate a character based on the reference image.

Review the generated face, hairstyle, and clothing. If you need a detail adjustment, use Fine-tune (微调) on the character. For example:

Change the robe lining to emerald green. Keep the face and hairstyle unchanged.

Character-generation mode currently does not have conversation memory. Use Fine-tune for changes to the generated character, and save the version you want to use.

Both routes now lead to the same place: a saved character that you can select in later creations.

Step 3: Save both characters before building a two-person scene

The key sequence is:

Create the character → Fine-tune if needed → Save → Reference in image or video mode.

Idoli uses the saved character as a virtual subject in this workflow, rather than requiring the original real-person photo as the direct input for every scene. Saving and referencing the character is the important distinction; simply uploading the same photo in video mode is not the workflow described here.

For a two-person romance, prepare the second character as well. The linked example pairs a platinum-haired woman with a dark-haired man wearing round glasses. Create and save each separately so you can select the intended subjects for the scene.

You can use clear project labels such as “Malfoid” and “Potter” to distinguish them in your own planning. In the scene prompt, also assign unambiguous visual roles, such as “the platinum-haired woman on the left” and “the dark-haired man on the right.”

Step 4: Reference the characters and write the video scene

Open video mode and reference the saved characters. Then describe what happens: where they stand, who speaks, how the other character reacts, and where the clip ends.

For a starting scene, open Malfoy Wants Me to Confess I. It provides a fifteen-second romantic confrontation in a rainy castle cloister, ending on the woman’s demand: “No. Swear on my life.” This is a fan-created dramatic scenario, not a quotation from the original books.

Use the template as the scene structure after selecting your saved characters. To make their roles clear, you can introduce the scene with:

Use the referenced platinum-haired adult woman as the woman on screen left and the referenced dark-haired adult man with round glasses as the man on screen right. Keep each character’s appearance consistent across the scene. Follow the timed actions and assigned dialogue below.

Then include the template’s complete timed scene instructions. Do not send only this role-assignment paragraph: the action, dialogue, and timing still need to be part of the generation request.

Video mode currently does not have conversation memory. Provide the complete instructions and intended character references for each generation. Begin with one short scene and review the faces, eyelines, speaker assignment, and movement before making the next part.

Step 5: Choose a model for the scene you want to make

Idoli offers multiple models for image and video generation. Studio membership provides access to all models. Select among the options available in your account for the project you are creating.

For this particular scene, check whether the selected video model supports the spoken dialogue you want. A visual-only result and a clip with synchronized English dialogue are different production requirements. The character workflow establishes your subjects; the selected model still determines the available generation capabilities.

Saving a character addresses the direct-photo issue described in this guide. It does not guarantee every generation will succeed or remove all content restrictions. Your prompt must still meet the applicable requirements, and generating material uses credits.

Start with the characters, then build the story

A saved character gives your Malfoid-inspired project a practical foundation. You can refine the appearance before asking for a performance, reference it in an image to explore a setting, and then use it in video mode for the dramatic scene.

Start with the character route that suits your materials. Save the result, prepare the second character, and try Malfoy Wants Me to Confess I as your first dialogue-led scene. Once you have reviewed the result, continue with your own setting, dialogue, and emotional turn.