VidMuse FAQ

VidMuse is an AI video creation platform built around music, audio, and storytelling. You can start with a song, an audio file, a Suno link, an idea, or a product concept, and VidMuse will help turn it into a complete visual video.

VidMuse can create:

• Music Videos / MVs Upload an audio file or a Suno link, and VidMuse can analyze the rhythm, mood, lyrics, and emotional structure to create a music-driven video.
• Narrative Videos and Short Films VidMuse can help create story-based videos, cinematic short films, character-driven scenes, and narrative visual content.
• Product Ads and Commercial Brand Videos VidMuse supports ad-style video creation, including video remixes, viral demos, UGC-style videos, unboxing videos, product showcases, and brand storytelling.
• Cinematic Shots VidMuse supports many shot types, including performance scenes, lip-syncing, dancing, dramatic acting, visual effects, product shots, and stylized cinematic sequences.

In short, VidMuse is not only for music videos. It can help creators, musicians, brands, and storytellers create MVs, narrative videos, product ads, and commercial video content.

Yes.

There are two stages:

• Creation stage Choose Studio Mode for better detail, lighting, and overall quality.
• Export stage Choose your export resolution.

VidMuse currently supports:

• 720p HD
• 1080p Full HD

VidMuse can generate AI music, but it does not currently support detailed editing of existing music.

Supported:

• Generate background music from style or mood prompts
• Create music based on emotional direction

Not currently supported:

• Mixing
• Mastering
• Cutting music
• Editing vocals
• Adjusting existing audio structure

If you need detailed music editing, please use professional audio software before uploading the music to VidMuse.

Text discussion, planning, and prompt editing do not consume credits. However, real audio analysis, image generation, video generation, and full rendering require credits.

If you are new to VidMuse, we recommend:

• testing with a 30s-60s short section first
• using Lite Mode when appropriate
• confirming the storyboard before generation
• generating a few key shots before committing to a full long video

A complete finished video cannot usually be previewed for free because generation uses real model resources.

VidMuse may adjust, soften, or refuse certain violent, unsafe, explicit, or risky content because of model and platform safety rules.

If a scene is blocked or looks too mild, try rewriting it as:

• cinematic tension
• dramatic confrontation
• stylized action
• implied conflict
• non-graphic movement
• fantasy or symbolic action

Avoid instructions that require graphic injury, dangerous real-world behavior, or explicit harm.

Generally, no.

Credits are digital virtual services. Once topped up or consumed, they are non-refundable.

However, if credits were lost due to a confirmed system crash, server error, or technical fault, please contact support for review.

Support channel:

• Feedback & Issues: https://vidmuse.ai/feedback
• Discord: https://discord.gg/vidmuse

The amount of video you can create depends on the model, video length, resolution, project complexity, and how many times you regenerate.

Please note that these are rough estimates, not guaranteed credit quotes.

Rough estimates:

• 300 credits: usually enough for a short product ad around 15 seconds.
• 2,000 credits: usually enough for a standard short video under 2 minutes.
• 4,000 credits: usually enough for a medium-length video or narrative short around 3-4 minutes.

If your project uses higher-quality models, higher resolution, complex scenes, or many regenerations, credits will be used faster.

You can get credits in several ways:

Add-on packs

Add-on credit packs may include bonus credits with certain subscriptions or promotions.

One-time credits for new users

You will receive one-time credits after you sign up.

Referrals

You may receive credits when friends register, generate content, or purchase a plan.

Community events

VidMuse may give out extra credits through Discord events, creator campaigns, or community activities.

Join Discord: https://discord.gg/vidmuse

Credits are deducted based on tool calls, such as audio analysis, image generation, video generation, or audio processing.

For a limited time, VidMuse is waiving LLM costs for video creation, so normal conversation, planning, and prompt editing do not consume credits.

The number of credits consumed may depend on:

• Model type
• Video length
• Resolution
• Task complexity
• Advanced parameters

Yes.

Subscription credits

• Issued monthly based on your plan.
• Valid for one billing cycle.
• Unused subscription credits do not carry over to the next billing cycle.

Add-on credits

• Usually valid for 365 days from purchase.

Bonus credits

• May be granted through events, referrals, or promotions.
• Unless otherwise stated, bonus credits expire after one month.

You can upgrade from Pro to Ultra or Studio, or switch from a monthly subscription to an annual subscription.

When upgrading:

• Remaining credits will carry over.
• The remaining value of your current plan may be applied as a pro-rated discount.
• Benefits will be recalculated based on the new plan.
• For monthly plans, credits are refreshed each month based on your original subscription date.

You can cancel your subscription from your billing page with the following steps:

1. Click your profile avatar in the upper-right corner of the homepage.
2. Go to your personal center.
3. Open the Plan & Usage tab.
4. Click Manage Subscription.
5. On the billing page, click Cancel Subscription.

After cancellation:

• You can continue using paid features until the end of the current billing cycle.
• The subscription will not renew after that cycle ends.

You can delete your account from your account settings:

1. Click your profile avatar in the upper-right corner of the homepage.
2. Go to your personal center.
3. Open the Account tab.
4. Scroll to the bottom and click Delete My Account.

Please note that deleting your account is permanent and cannot be undone.

After deletion:

• Any active subscription will be automatically canceled.
• You will no longer be able to log in to the account.
• You will no longer be able to view previous projects or assets.

If duplicate charges happen due to a payment method change or billing error, please contact us through “Feedback & Issues” for help.

We will respond within 3 business days and complete the refund process within 7 business days after confirmation.

Please read carefully before purchasing.

Once you start using the service, you are considered to have agreed to the refund policy.

• Subscription fees are non-refundable.
• No refunds are provided for partially used billing periods.
• Used credits are non-refundable.
• After cancellation, you can still access paid features until the end of the current billing period.
• After deletion,

VidMuse Operations Team reserves the final right of interpretation.

A code is an exclusive benefit provided to creators, influencers, or special partners.

You can redeem it on the Pricing page by scrolling down to the code redemption section.

Please note:

• Each account can only use a code once.
• A code may provide a top-up discount, extra credits, or other special benefits.
• The exact reward depends on the specific code you receive.

If credits were used for a task that failed due to a confirmed platform, server, or API issue, contact support for review.

Please include:

• the affected project
• which shots or clips failed
• the credit amount shown
• screenshots if available
• whether failure happened before or after generation started

Credits are not automatically restored for every unsatisfactory result, but technical failures can be reviewed.

Some generation tasks may continue in the background after they have been started, especially batch generation or multi-shot workflows.

To reduce unexpected credit use:

• check whether a generation task is still running
• do not leave unfinished batch tasks active
• stop or pause work before closing the page
• avoid giving broad instructions like “finish everything” unless you are ready to spend credits

If credits were deducted without any user-started generation task, contact support with the project/thread ID for investigation.

VidMuse only deducts credits when running real generation or analysis tasks.

Credits may be consumed during:

• Audio analysis
• Video analysis
• Image generation
• Video generation
• Audio generation or processing

Free actions include:

• Chatting with the AI
• Discussing ideas
• Asking for revision suggestions
• Editing scripts in text
• Writing or improving prompts

As long as you do not start a generation task, normal text conversation does not consume credits.

Before storyboard images are generated, you can change the resolution by talking with VidMuse.

After storyboard images have been generated, changing the resolution is not recommended. It may cause a mismatch between the image resolution and the video resolution, which can make the final video look blurry.

After the video has been generated, changing the resolution is not currently supported. An upscale feature is coming soon.

If you need a high-quality final video, we recommend choosing your target resolution when creating the project.

This is strongly not recommended.

When a project is created, VidMuse analyzes the audio rhythm and builds the timeline structure around it. Changing the audio duration, replacing the song, or making major structural changes midway may cause:

• Timeline mismatches
• Rhythm sync issues
• Shot alignment errors
• Generation failures

Best practice:

If you need to change the song, audio duration, or overall structure, adjust the music duration or regenerate the music through conversation at the very beginning of the project. Once music analysis has been completed and the project is underway, we do not recommend making these changes.

Credit estimates are approximate, not a guaranteed final price.

Actual credit use may change based on:

• final shot count
• video duration
• selected model
• resolution
• regenerated shots
• failed or retried clips
• edits requested after the storyboard is created

Before starting a full generation, review the storyboard, model choice, resolution, and estimated cost carefully. For long music videos, test with a shorter section first if you are unsure.

Define the cast before generation.

Include:

• each character’s name
• role or instrument
• who sings which lines
• reference images for each main person
• consistent wardrobe and appearance notes
• clear shot-level instructions

For duet, chorus, or band videos, do not rely only on the lyrics. Tell VidMuse exactly which character should appear and sing in each important shot.

Before any images or videos are generated, you can ask VidMuse to change the project’s aspect ratio through conversation.

After images or videos have been generated, it depends on what kind of change you need.

• If you only want to crop or reframe the existing result, you can ask VidMuse to help adjust it. However, if you want to fully regenerate the project in a different aspect ratio, we recommend creating a new project with the correct format.
• Forcing an aspect ratio change midway may require regenerating reference images and videos, which can waste credits and may affect the final result.
• For best results, choose the target aspect ratio at the beginning of the project and upload source materials that match that format whenever possible.

Style defines the overall visual atmosphere of your video.

It affects:

• Reference images
• First-frame images
• Overall visual direction
• Color, lighting, texture, and artistic mood

VidMuse may recommend several styles based on your project. You can choose one, ask for a new batch, or describe your own style.

Important notes:

• Style only affects the visual look.
• Style does not control editing pace, acting, or shot movement.
• Use broad visual descriptions, not shot-level instructions.
• Avoid private or unclear terms like “VidMuse style” or “Sand AI style.”
• Avoid putting specific camera composition requirements into the global style.

Good style examples:

• “A cinematic cyberpunk style with neon lighting, high contrast, and deep blue-purple tones.”
• “A visual style inspired by Wong Kar-wai, with warm lighting, rich shadows, and emotional close-ups.”
• “A dreamy fantasy style with soft light, misty environments, and pastel colors.”

Bad style example:

• “Wes Anderson style with every shot centered and symmetrical.”

Why this is bad:

If you put “centered and symmetrical” into the global style, the system may force every shot to look that way, which can hurt the storyboard and make the video repetitive.

Use the style step to define the macro visual atmosphere of the whole video.

This usually happens for one of three reasons.

The style description uses private or unclear words

The model understands public visual concepts, not your personal definitions.

Avoid:

• “VidMuse style”
• “Apple style”
• “Like the video I watched yesterday”

Use:

• Public art movements
• Famous directors or artists
• Clear visual adjectives
• Movie references
• Specific color, lighting, and texture descriptions

The style description contains shot-level instructions

Style is global. Shot-level details should stay in the storyboard.

Avoid:

• “Use centered composition in every shot.”
• “Every scene should show the character walking toward the camera.”

Use:

• “High-saturation colors”
• “Vintage film texture”
• “Soft natural lighting”
• “Dark fantasy atmosphere”

You expect one reference image to fully define the style

Reference images are helpful, but they are not strict templates.

Current limitation:

• A single image cannot always produce perfect pixel-level style replication.
• Results may vary across shots.
• Clear text descriptions are still very important.

Style mismatch usually comes from prompt conflicts or incorrect use of reference images.

The shot prompt is too vague

If the shot description is too simple, the style may not attach well to the scene.

Weak prompt:

• “A girl standing outside.”

Better prompt:

• “A girl in a red jacket standing under neon lights on a rainy city street.”

More concrete details help the model apply the style better.

The reference image is used in the wrong place

Global style references are mainly for overall tone and atmosphere.

If you want a specific shot to follow a reference image more closely, upload the reference image directly in the Storyboard Canvas for that shot.

This gives the model stronger guidance for that specific image or shot.

Strong style keywords may affect everything

Some keywords are very strong, such as:

• Watercolor
• Oil painting
• 3D modeling
• Clay style
• Ink wash

If these words are in the global style, they may affect every shot.

If you only want the effect in one scene, put it in that specific shot instead of the global style.

Recommended way to fix issues

Use this format:

Quote Shot + Problem + Target

Example:

• “@Shot 5, the background is too dark. Please change it to a sunny outdoor setting.”
• “@Shot 8, the character should walk past the body without touching it.”
• “@Shot 12, remove the extra person and keep only the main character.”

Use Remix for small local fixes

If most of the image is good and only one small area is wrong, do not regenerate the whole image.

Use Remix to fix a specific area, such as:

• Hands
• Face details
• Clothing color
• Small object changes

Use the image editing page for more control

You can also open the image editing page to make more precise adjustments. There, you can:

• edit the prompt directly
• choose a different model
• control more generation parameters

Review these settings before regenerating the result.

Use visual guidance when text is not enough

If text instructions do not work, try:

• Uploading a reference image
• Switching to another model
• Editing the prompt directly in the storyboard

Avoid vague instructions like:

• “I don’t like this.”
• “Make it better.”
• “Fix everything.”

These are too unclear and may waste credits.

Better character consistency depends heavily on the quality of your reference images.

Best practices

• Upload high-quality reference images from multiple angles

Use clear images that show the target character from different angles, such as front, side, and three-quarter views. Avoid blurry, heavily filtered, or low-resolution images.

• Check the reference stage

Make sure the character and overall look match your expectations before moving forward.

• For image-to-video, check the first frame

If the first-frame image does not look like the character, the video will likely be inconsistent.

• For reference-to-video or multi-shot generation, check the inputs

Make sure the reference images are actually used as video generation inputs.

• Use stronger models when needed

If the reference images, first frame, and inputs are all correct, try stronger models such as Seedance 2.5 or Kling V3, then reroll for better results.

Even with good references, character consistency can still be unstable in some cases.

Common reasons:

• The reference photo is unclear
• The face is too small or blocked
• Lighting is too complex
• The angle is difficult
• The style is too abstract
• The shot has fast motion or extreme camera movement
• Too many characters appear in the same shot

Tips:

• Upload multiple high-quality reference images from different angles, such as front, side, and three-quarter views.
• Use a clear front-facing portrait.
• Avoid complex lighting in reference images.
• Use realistic or 2D anime styles if facial accuracy matters.
• Use the same character description across shots.
• Avoid extreme motion if you need stable faces.
• Use medium or long shots to hide minor facial changes.

Technical note:

Minor face changes, flickering, or limb distortion during motion are still common limitations in current AI video generation.

Credits are often wasted when VidMuse edits the wrong shot or misunderstands the request.

Before video generation

• Review every scene, shot, and camera direction in the storyboard.
• Check the character, look, location, props, composition, and framing.
• Fix storyboard and visual design issues before generating video.
• Fix one shot or one scene at a time.

After video generation

• Fix one video at a time.
• For each video, list all problems at once.
• Do not try to fix the same issue across multiple videos at the same time.
• Align with VidMuse on the editing plan first, then make changes after confirmation.

Better

• “@shot 8, fix the character face, remove the extra person, and change the background to a forest.”

Worse:

• First fix the face.
• Then regenerate.
• Then fix the background.
• Then regenerate again.
• Then fix the action.

This usually costs more credits.

Do not just type “Shot 1” or “Video 1.”

VidMuse may not always locate the correct shot from text alone.

Better ways:

• quote the exact shot or video in chat
• Open the shot or video from the storyboard
• Edit the prompt inside that specific shot or video

This helps VidMuse understand exactly which shot or video you want to fix.

Go back to the storyboard.

Steps:

1. Open the problematic shot.
2. Review the original description.
3. Edit the prompt based on the original description.
4. Be specific about what should change.
5. Generate again only after the instruction is clear.

This is usually more stable than describing the issue only in chat.

Yes.

To reduce unnecessary credit use:

• Check the storyboard before video generation.
• Fix one shot at a time.
• Tag or quote the exact shot.
• Use Remix for small local fixes.
• Avoid vague prompts.
• Avoid repeated full regenerations.
• Choose models based on your quality and cost needs.

Yes.

At project creation

You can choose a generation mode:

• Studio Mode Higher quality and better detail.
• Lite Mode More cost-effective.
• Custom Mode Available for members who want to save preferred parameters.

During editing

In the Canvas, you can ask VidMuse to use a different model for a specific stage or shot, then regenerate the result after confirming what you want to adjust.

Use Remix.

Do not regenerate the whole image if only a small part is wrong.

Use Remix for:

• Changing a hat color
• Fixing hands
• Adjusting a face detail
• Removing a small object
• Editing part of the clothing

Steps:

1. Select the image.
2. Click Remix.
3. Enter a clear instruction.
4. Regenerate the image.
5. Continue editing based on the new image, or use it directly.

Our model library is updated regularly.

We evaluate new models based on:

• Quality
• Stability
• Cost
• User demand
• Compatibility with VidMuse workflows

If you want us to add a specific model, please suggest it in our Discord community or through Feedback&Issues:

https://discord.gg/vidmuse

Lip sync is still a difficult AI video task, especially with long songs, fast vocals, multiple singers, unclear lyrics, or complex camera movement.

For better results:

• provide clean, complete lyrics
• split long songs into clearer sections
• identify who sings each line
• avoid too many singers in the same shot
• use stronger models for close-up singing shots
• regenerate only the shots where lip sync matters most

Minor mouth mismatch, timing drift, or imperfect lyric alignment can still happen.

Reference images and first frames guide the model, but they are not strict pixel-level guarantees.

Common reasons:

• the reference was uploaded to the wrong place
• too many references conflicted with each other
• the face or object in the reference was unclear
• the shot prompt contradicted the reference
• the model had difficulty preserving the reference during motion

For stronger control, upload the reference directly to the specific storyboard shot, use clear character names, and check the first frame before generating video.

If previous versions are still available in the project, select the better version before continuing.

Before regenerating important shots:

• review the current result
• keep the version you like
• regenerate one shot at a time
• avoid broad instructions that may change many shots

If a previous result disappeared or was overwritten unexpectedly, contact support. Restoration is not always guaranteed.

AI video models may sometimes create unwanted text, symbols, subtitles, or incorrect language-like writing, especially if the prompt includes signs, screens, lyrics, captions, or poster-like scenes.

To reduce this:

• say “no visible text, no subtitles, no letters” when text is not needed
• avoid asking for readable text inside the video unless necessary
• use the Remix feature for small unwanted text areas
• regenerate the affected shot if the text is large
• provide exact language requirements if subtitles or captions are needed

First confirm whether the issue appears in the final exported video, not only during preview.

Check:

• the browser/player is not muted
• the uploaded audio file is correct
• the audio duration matches the intended project
• the project was not changed midway with a different song or duration

If the final export has no sound, mismatched duration, or serious audio/video drift, contact support with the project name and export file details.

If a project stays stuck, avoid repeatedly starting new generations. Complex models, long videos, or background tasks can take longer, but if the project is frozen for several hours, the stop button does not respond, or nothing can be edited, contact us through Feedback & Issues for review.

Please include:

• approximate time of the issue
• what step is stuck
• whether credits were deducted

If the issue is confirmed as a system, server, or generation-task failure, support can review credit compensation case by case.

First check that the full project has finished generating. Then try:

• refreshing the page
• waiting a few minutes and trying again
• using a stable network
• trying another browser
• exporting again without changing the project

If download or export still fails, contact support with the project name, export resolution, and the time the issue happened. Do not repeatedly regenerate shots just to solve a download problem.

To help support review the issue faster, include:

• What actually happened
• What you expected

For project-quality issues, describe the exact shot and problem instead of only saying “the video is wrong.”

The same work can only participate in one contest at a time and cannot be entered into multiple contests simultaneously. If you resubmit a work that is already entered in a contest to another contest, the system will automatically switch the contest for that work, retaining eligibility only for the most recently selected one. The previous contest registration will become invalid.

If the work has already won an award in a past contest, it cannot participate in any new contest selection. We encourage you to create new works for submission.

Resubmission is not possible for the same version of a work where neither the title nor the content has been changed. If you only want to modify the cover, we recommend adjusting the title or video content as well before resubmitting.

All contest entries will be simultaneously published to the VidMuse community. Works must pass community content moderation first to enter the candidate pool for the corresponding contest; works that do not pass community moderation are not eligible for awards.

Divided into two main scenarios:

1. If the old work hasn't passed community moderation: Uploading a new version will directly overwrite all original records. Video content and contest selection will all be updated to the latest submission.
2. If the old work has passed moderation and is publicly displayed in the community:

• New version submitted to the original contest: The original online version and contest record remain. After the new version is approved, it replaces the community display and contest entry; if not approved, the original remains unchanged.
• New version submitted to a different contest: The original community version remains, but contest eligibility for the original contest is automatically canceled. The new version joins the new contest after approval.
• New version published to community only (not for contest): The original community version remains, but the original contest eligibility is automatically canceled (if any). Only the online display content is updated after the new version is approved.