Create a Mukbang YouTube Channel with Your AI Influencer

Watch the finished AI mukbang example

Start by watching the result with sound. This is an AI-generated food-performance example: the influencer, food interaction and audio are generated. The brief aims for a quiet, continuous eating scene with expressive reactions and close eating sounds.

Treat your first episode as a reusable production method. Keep the influencer recognizable, choose a clear food theme, and make each new video an intentional variation: different textures, a different dish, or a different reaction.

Plan a recognizable channel and first episode

Give the channel a specific promise that a viewer can understand in one sentence. For example: “AI food scenes with satisfying textures and no talking.” A clear concept helps you decide which images to collect and which sounds should dominate the video.

• Choose the AI influencer who will appear across the series. Keep a clean identity reference you can reuse.
• Pick one food theme for the first episode, such as a spicy dumpling tray or a crunchy snack plate.
• Decide whether the episode is silent ASMR, spoken commentary or a reaction format. This example is silent ASMR.
• Choose a consistent setting and camera style. Here, the camera is fixed across the table and the food fills the foreground.
• Plan the opening bite and the final moment. Even a short food scene benefits from a beginning and an ending.

Find inspiration with a clear food composition

Look for a scene whose structure already supports the episode: a readable face, visible hands or chopsticks, an unobstructed food tray and enough room for the action. In this walkthrough, the Pinterest search is used to explore food arrangements and framing. Use reference material you have permission to adapt.

The prompt asks to remove the visible caption from the reference and reconstruct the image underneath it. Inspect that area in the generated result: asking for clean detail does not guarantee that every food edge, finger or fabric pattern will be correct.

Give each reference a different job

The influencer portrait is visible in the right-hand source thumbnail in the supplied setup. Its clothing, room and pose are not meant to carry into the result. The prompt keeps the floral top and food composition from the first image while using the second image’s identity.

Create the starting image with Max Edit

1. Open Images from the side navigation and select Edit Image.
2. Choose Max Edit in the model selector.
3. Upload the food composition and your influencer identity image. Confirm their order matches @img1 and @img2 in your prompt.
4. Paste the image prompt below and insert the matching reference tags in FanPro.
5. Set the output format in the controls. The supplied setup changes from a 4:3 view to 1:1; its downloaded result is square. Choose your intended format deliberately.
6. Select Generate, then inspect the result before moving into video.

The useful part of this prompt is the division of responsibility. It describes the composition in detail, restricts which traits come from the identity reference, and asks for a clean image without captions. Adapt those details to your own scene instead of retaining a food list that no longer matches your reference.

Review the generated image that becomes the video input

• Identity: the face should match your selected influencer, with a consistent hair colour and recognizable features.
• Interaction: inspect fingers, chopsticks and the point where the food approaches the mouth.
• Food: check the bowl rim, egg, glaze and separate items for merged shapes or distorted details.
• Composition: leave space for the planned actions without clipping the tray or the influencer’s head.
• Text: review the former caption area and remove unwanted text through a corrected generation if necessary.

The video prompt adds a lavalier microphone, which is not visible in this supplied starting still. That asks the video model to introduce an object as well as animate the scene. If a perfectly matched first frame is essential, prepare a starting image that already includes every required prop before running the video.

Animate the scene in Studio Marketing

1. Open Marketing from the side navigation.
2. Use the UGC tab and choose Optimus.
3. Under Media, choose Image and upload the generated starting still.
4. Confirm that IMG1 is the generated food scene, then insert its reference tag in the video prompt.
5. Set the duration and resolution available for your run. This example uses 30 seconds and 1080p.
6. Choose the actual output aspect ratio. The recorded run uses 4:3 despite the prompt’s 9:16 wording.
7. Review the current credit amount displayed in your account, then select Create.

Use the preview to confirm the input before generating. Uploading the original Pinterest reference here would direct the video from the wrong identity. Adding several unrelated images would also make the simple one-scene brief harder to interpret.

Build a sequence of bites and reactions

The video brief breaks the episode into five six-second stages. Each stage names an action and an end state so the next one has something concrete to continue. This is the intended sequence; use it as a checklist against the generated result.

Keep the action load realistic for the available time. The source brief’s fourth stage mentions one bun being eaten but an end state with two buns gone. When adapting it, make the action and end-state counts agree. The original text is retained below so you can see exactly what was supplied.

For a simpler first test, direct fewer bites and one clear expression change. Once hand movement, food continuity and audio are working, add more variety. Changing one issue at a time makes the next result easier to evaluate.

The original ASMR mukbang video prompt

This is the complete video prompt used for the example, separated into its original named sections. Replace the scene-specific details for your own episode and check the output controls separately. Text such as “exactly” and “never” expresses the requested direction; it does not guarantee a flawless result.

Review the finished video before publishing

• Watch once for identity and scene consistency: face, hair, top, sofa and tray.
• Watch again for hand-to-food contact, chopsticks, bites and mouth movement. Look for items that duplicate or reappear.
• Listen with headphones at a comfortable volume. Check whether sound lines up with visible bites and whether unwanted speech, music or background sound appears.
• Check the lavalier microphone and whether it stays attached in a consistent place.
• Review the exported dimensions and duration, then watch the opening and ending at normal speed.
• If a detail breaks the scene, revise that part of the prompt or starting image and compare the next run.

Keep a record of the selected input, prompt, model and output settings alongside the final file. That gives you a reproducible starting point for the next episode and a way to track which change actually improved the result.

Set up the channel and publish your first episode

Choose a channel name and handle that fit the influencer and the food concept. Use a consistent portrait and write a short description of the experience viewers can expect—for example, AI-created food scenes, close eating sounds and no talking.

1. Sign in to YouTube. In your account settings, open the channel-management option and create a channel with your chosen name, handle and profile picture.
2. In YouTube Studio, choose Create → Upload videos and select your reviewed export.
3. Add a specific title and description that match the actual food and format. Choose a thumbnail that accurately represents the episode.
4. Complete the audience, AI-use disclosure and other applicable upload settings, then review YouTube’s checks.
5. Preview the upload before choosing its visibility or scheduling publication.

Channel setup and upload instructions: YouTube Help, “Create a YouTube channel” (https://support.google.com/youtube/answer/1646861) and “Upload YouTube videos” (https://support.google.com/youtube/answer/57407), checked 11 September 2026.

For Shorts, YouTube’s current guidance covers square or vertical uploads up to three minutes. This example’s 4:3 export is wider than square, so prepare a square or vertical version if you want the Shorts format. Check that any reframing still includes the hands, food and face. Source: YouTube Help, https://support.google.com/youtube/answer/15424877, checked 11 September 2026.

Turn one episode into a consistent series

Keep the influencer and the recognizable channel style while giving each episode its own purpose. Try a soft-versus-crunchy texture contrast, a single dish with a careful close-up, or a quieter scene with fewer movements. Write a fresh food and action description each time so the prompt still matches the image.

Review audience response alongside your own quality notes. Keep the parts viewers respond to and improve the parts that interrupt the scene. Build the channel around deliberate episodes, with a clear subject and a reviewed result.

Connect accounts and plan your publishing calendar

See how FanPro’s publishing workflow organizes connected accounts and scheduled posts.

Direct a five-stage AI UGC performance

Study the Porsche walkthrough for another way to combine image references and timed scene direction.

Build your AI influencer from scratch

Prepare an influencer identity to use across your next content series.