How to Use DojoClip Story Music Video: Complete Director V2 Guide

Story Music Video turns one finished song into a directed, full-song visual story. It analyzes the lyrics and musical structure, proposes narrative treatments, builds a timed storyboard, generates still frames, tests representative motion, and then fills the complete song timeline.

This is a guided workflow. It does not make a disposable 30-second preview and silently continue on its own. You review and approve the creative decisions at clear checkpoints, and you see an exact credit quote before image or video generation begins.

Open the app here: DojoClip Story Music Video.

Current output note: Director V2 currently ends with a full-song motion composition that you can inspect and approve in the browser. A downloadable final MP4 export is not available yet. Start a project when the guided planning, frames, motion clips, and in-browser composition are the outputs you need.

Before you start: prepare the right inputs

You need a signed-in account with paid access. Starting the Director workflow uses 0 upfront credits. Credits are requested later, in separate image and motion quotes, after you have reviewed the creative plan.

Prepare these inputs before opening the app:

Input	Required?	Current limit	What it controls
Song	Yes	MP3, M4A, or WAV; up to 50 MB and 3 minutes	Lyrics, timing, structure, mood, and the complete video duration
Character A	No	One portrait image	The lead character's visual identity
Character B	No	One additional portrait image	A co-lead and the relationship between the two characters
Style	No	Up to 500 characters	Art direction, palette, texture, lighting, camera language, and mood
Aspect ratio	Yes	9:16 or 16:9	The format used for every generated frame and motion clip

The audio filename becomes the task name in your history, without its extension and shortened to 20 characters. Rename a file like final_mix_v7.wav to something recognizable, such as Midnight Train.wav, before uploading it.

Choose a song the Director can understand

Use the final mix you actually want on the timeline. A clean vocal, audible lyrics, and a stable master give the transcript and story planner better evidence. Avoid voice notes, unfinished demos with long silent sections, or files that clip heavily.

Instrumental and vocal-free tracks can be uploaded, but the Director has less narrative evidence to work from. In that case, make the Style field more specific about the story, setting, and emotional progression you want.

Only upload audio you own or have permission to use. The same applies to portraits and any recognizable person in them.

Decide whether to upload characters

You can use the app in three ways:

Portraits	Result
None	The Director creates an environment-, object-, or abstract-led story with no recurring character
One	The uploaded person becomes the featured lead
Two	Both people become featured main characters, with relationship direction available at the Cast checkpoint

Each uploaded person is treated as a main character and is intended to appear for at least half of the music video. Do not upload a second portrait merely as a minor extra.

For dependable character-led work, upload at least one good portrait. Choose no portrait only when you want a story led by environments, objects, performance details, or abstract imagery rather than a recurring person.

For the strongest identity consistency, use a clear, recent portrait with one person in frame, even light across the face, visible facial features, and no heavy filter. Avoid group photos, tiny faces, extreme profiles, sunglasses, hands covering the face, and strong motion blur. A JPEG, PNG, or WebP portrait is a dependable choice.

Write a useful style brief

The Style field is art direction, not a place to repeat the lyrics. Describe a coherent visual world and leave room for the Director to stage it.

Rain-soaked neo-noir road story at night. Practical street lighting, deep green
and amber palette, restrained handheld camera, reflective windows, realistic
wardrobe, intimate performances, and a hopeful sunrise in the final chorus.
Avoid text on screen, fantasy creatures, and exaggerated facial expressions.

A good brief usually contains:

the visual medium or level of realism
location and time of day
palette and lighting
camera energy
emotional arc
anything that must not appear

Avoid stacking incompatible directions such as "photorealistic watercolor 3D anime documentary." Pick one primary visual language, then add two or three supporting details.

You do not need to write separate image-provider or video-provider prompts. The Director compiles those from the approved transcript, treatment, storyboard, cast, and style brief.

Pick the aspect ratio before generation

Choose 9:16 for TikTok, Reels, and Shorts. Choose 16:9 for YouTube, television, and landscape players.

The aspect ratio is part of every composition and provider request. Treat it as a project-level decision, not a crop you will casually change at the end.

Step 1: upload and start the Director workflow

Drop the song into Audio file, or click the upload area to browse.
Wait until the song says it is ready. The app checks its type, size, and duration before task creation.
Optionally add Character A and Character B. Wait for every portrait to finish uploading.
Enter a style brief and choose 9:16 or 16:9.
Click Start Director workflow.

Do not leave the page while a file is still uploading. Once the new task appears in Your Music Video History and opens in the workspace, its state is durable. You may then leave the page and reopen the task later.

Starting the workflow creates the task and hands the song to the preparation worker. The first visible work includes speech recognition, transcript preparation, audio analysis, and character observation. The task should show a live stage or handoff state rather than sitting on a generic preview spinner.

The app may show the Cast checkpoint while song preparation continues in parallel. That is normal: portrait review and song analysis do not always finish at the same moment.

Understand the seven checkpoints

The workspace header shows the complete path:

Checkpoint	What the system produces	What you decide
Cast	Character identities and observations	Who appears, who leads, and how two characters relate
Transcript	Timed lyric lines and song evidence	Whether the words are correct
Treatment	Several story concepts	Which narrative direction becomes the plan
Storyboard	Full-song sequences and timed shots	What happens, when it happens, and which details are locked
Frames	Generated start/end frame variants and an animatic	Which still image represents each boundary
Motion	Representative clips, then remaining clips	Whether movement, identity, duration, and continuity work
Final	Full-song motion composition	Which durable clip is used in every range and whether the composition is approved

Processing means the system is working. Awaiting your review means the system has deliberately stopped for you. A review state is not a delay; open the task and complete the visible checkpoint.

Step 2: review the Cast checkpoint

The Cast checkpoint is where you protect character identity before any story plan or image generation depends on it.

For each uploaded person, check the portrait and character-analysis status. You can:

change the display label
describe the story role, such as performer, traveler, witness, or co-lead
make the other person the lead
update the photo while keeping the same person
replace the cast member with a different person
remove a person or add a second portrait
retry character analysis when the observation failed

With two people, also set the relationship: partners, family, friends, rivals, professional contacts, strangers, or another relationship. Add direction notes, list situations to avoid, and decide whether physical contact is allowed. These instructions apply to shots where both characters appear.

Use Update photo when the identity is correct but the reference image is poor. Use Replace person only when you are changing who the character is.

Approve the cast only when the people, labels, roles, and relationship are correct. Changing identity after downstream work exists can make later artifacts stale and require regeneration.

Step 3: correct and approve the Transcript

The Transcript tab displays timed lines extracted from the song. Read it while listening to your original audio, with extra attention to:

names, slang, and uncommon words
repeated hooks that differ slightly
words hidden by effects or backing vocals
line breaks that change the meaning
ad-libs that should or should not influence the story

Edit the text of any incorrect line, then click Save edits. The visible time markers are evidence for where the line occurs; the current editor lets you correct the words, not manually retime each line.

Do not approve a transcript just because it is mostly correct. One wrong name or negation can send an entire treatment in the wrong direction. When the saved lines match the song, click Approve transcript.

Approval starts the creative treatment stage. It does not spend image or video credits.

Step 4: choose a Treatment

A treatment is the high-level story proposal, before the video is divided into shots. The Director can present multiple concepts with a logline, central device, subtext, payoff, narrative mode, and character mapping.

Compare concepts by asking:

Does the idea match the song's emotional change, not just one lyric?
Can the concept sustain the whole song without repeating the same image?
Is the final payoff clear enough to make the ending feel intentional?
Are the uploaded characters being used in roles that make sense?
Can this story be expressed visually, without paragraphs of explanation?

If none is strong enough, choose Revise set before moving on. If a concept is right, select it. With character mappings, you can use the displayed mapping or keep the proposed personas parallel instead of binding them directly.

Treatment generation and selection are planning operations. No frame or motion credits are charged here.

Step 5: inspect and approve the Storyboard

The Storyboard turns the selected treatment into a complete song timeline. It groups the piece into sequences and divides those sequences into timed shots. Each shot carries its cast, visual event, primary action, camera move, environment motion, and pace.

Review the timeline from beginning to end. Look for:

an opening image that establishes the world quickly
visual escalation between verse, pre-chorus, and chorus
enough variation in shot scale and environment
character appearances that match their intended roles
cuts and actions that fit the musical energy
continuity between the exit of one shot and the entry of the next
a final image that pays off the treatment

You can edit an individual shot or a whole sequence. Use a shot edit for a specific action, camera move, or pace. Use a sequence edit for narrative purpose, entry image, exit image, or shared visual grammar.

Lock a shot or sequence when it is correct and you do not want a later revision to replace it. A lock is useful for must-keep moments, but too many early locks can make the rest of the story harder to improve.

Approve the storyboard only after reading the entire timeline. Approval makes the plan eligible for an exact frame quote.

Step 6: get a frame quote, generate variants, and approve the animatic

The Frames checkpoint is the first paid generation stage.

Click Get quote.
Review the number of unique frames, how many can be reused, how many must be generated, and the authorized credit total.
If the quote expires before you use it, refresh it.
Click Generate frames only when the scope and total are acceptable.

The plan may reuse one boundary image between neighboring shots, so the number of unique frames is not necessarily twice the number of shots. The quote, not a rough per-shot estimate, is the source of truth.

Unit pricing can change with the live generation policy. The exact total shown in your quote is what you authorize.

After generation, review every shot's start frame and, when required, end frame. For each boundary:

inspect the full-size selected image
compare durable variants
check the QA state and any identity warnings
select the strongest valid variant
lock a frame you want to preserve
use the included retry only when the button explicitly says Retry without another charge

Do not judge frames only as isolated pictures. Play the Full-song animatic with the original audio. It reveals repetition, weak transitions, poor pacing, and sudden identity changes much faster than a grid does.

Approve the frame set only when every required boundary has an approved durable variant and the full-song animatic tells the intended story.

Step 7: test representative Motion before generating the rest

Video generation is intentionally split into two paid decisions. First, the Director selects a small set of representative shots. These test the hardest or most informative motion before you pay to animate the complete timeline.

Click Get quote under Representative motion.
Check the selected shot count, generated duration, credits, and quote expiry.
Click Generate selection.
Review every resulting clip and its QA status.
Approve the selection only when all representative clips pass.

Trust the exact quote because it uses the current unit price and the actual selected scope.

Watch for more than general attractiveness:

Does the person still look like the approved cast member?
Does the first frame lead naturally into the motion?
When an end frame is supplied, does the clip reach it convincingly?
Does the clip duration match its real song range?
Are hands, faces, and contact between characters believable?
Does camera movement support the shot instead of fighting it?
Does the motion loop, freeze, or jump unexpectedly?

The original audio and approved frame animatic remain visible beside the test clips. Use them to judge the clip in context.

Individual generated motion clips are reviewed as visual clips and may be muted. The full composition places them on the original uploaded song timeline. Story Music Video directs performance and visual narrative; it does not promise phoneme-accurate lip sync.

After approval, request the Remaining motion quote. Representative clips that are already approved are reused without another charge. The quote shows the reused shot count, unresolved shot count, new credit total, and QA progress. Commit it with Generate remaining when you are ready.

Step 8: review the full-song motion composition

When every remaining clip has materialized, the task reaches Full preview ready. The Final workspace aligns the original song with one selected motion variant for every exact shot range.

Move through the full timeline and inspect each shot. You can select another acceptable variant, lock the active motion, or use an authorized retry where available. Then play the full composition and check:

synchronization from the first frame to the end of the song
missing, frozen, or duplicated ranges
abrupt character or wardrobe changes
transitions between representative and remaining motion
continuity at shot boundaries
whether the ending lands on the intended musical moment

Click Approve full motion only when every shot has an accepted durable clip and the full composition is coherent.

Approval records the chosen clip set and composition. At present, this is the end of the Director V2 workflow: the full-song composition is available for in-browser review, but the app does not yet render and offer a downloadable final MP4.

What to do while generation is running

Once the task exists in history, you do not need to keep the browser open during worker or provider generation. You can return to the Story Music Video page, select the task under Your Music Video History, and click Open task.

Use these rules while waiting:

Wait for uploads before leaving. A local file that has not finished uploading is not yet part of a durable task.
Do not submit a duplicate task because a provider is slow. Refresh or reopen the existing task first.
Read the stage label. Speech Recognition, Audio Analysis, planning, frame generation, and motion generation are different kinds of work with different timing.
Open review states. The workflow does not advance past Cast, Transcript, Treatment, Storyboard, Frames, or Motion approval by itself.
Use the manual refresh button when needed. Normal updates are polled automatically, but refresh is useful after returning to an older tab.
Keep the task's Debug ID when an error appears. It identifies the exact failed stage without exposing your media.

There is no reliable universal completion time. Song length, shot count, provider queue time, the number of generated boundaries, motion duration, retries, and how quickly you review checkpoints all matter. A longer song with many cuts will take more time and credits than a short song with a restrained shot plan.

Credits: when you pay and what is preserved

Starting a task, analyzing the song, reviewing cast, correcting the transcript, choosing a treatment, and approving the storyboard use 0 upfront generation credits.

Paid work begins only after you request and commit a quote:

Quote	What it authorizes	How price is set
Frames	New unique start/end frame boundaries	Shown in the live quote
Representative motion	A selected test set of video slots	Shown in the live quote
Remaining motion	Unresolved video slots after approved tests are reused	Shown in the live quote

Always use the total displayed in the live quote. Reuse, shared boundaries, and actual shot scope can change the total. Quotes expire; refreshing an expired quote lets the app recalculate the current scope before you commit.

A completed planning checkpoint remains saved if you do not have enough credits for the next generation batch. Buy or add credits, reopen the same task, refresh the quote, and continue. Do not recreate the project and lose reviewed work.

Some generated items include an authorized retry in the original allocation. The interface says Retry without another charge when that applies. Do not assume every regeneration is free; trust the action and quote shown for that item.

Statuses and recovery

Status	Meaning	What you should do
Processing	A worker, planner, image provider, or video provider is active	Leave it running or return later; refresh the existing task before doing anything else
Awaiting your review	A checkpoint is ready and deliberately paused	Open the task, review the active tab, save changes, and approve when ready
Action required	Processing stopped safely and saved work is waiting for recovery	Open the red banner, read the stage, use the recovery action when one is shown, and keep the Debug ID
Failed	The task or a required stage ended unsuccessfully	Read the saved error and retry only through the task's available action
Cancelled	The task was stopped	Start a new task only if you intentionally want a new project

If a handoff says it is initializing, queued, sending, accepted, running, retry scheduled, or uncertain, do not immediately resubmit the whole music video. The system keeps a durable dispatch record and reconciles worker state. Refresh the task first. If it becomes Action required, use the recovery button when the banner provides one. If no recovery is available, keep the Debug ID and contact support instead of repeatedly restarting the project.

Some two-character or provider-capability pauses can offer a choice such as waiting, continuing with Character A only, or cancelling. Read the consequence shown for each option before deciding; continuing with one character changes the approved creative scope.

When reporting a problem, include:

the task name
the checkpoint or stage label
the time the problem appeared
the Debug ID from the error banner
whether you had just saved, approved, quoted, or generated

Do not send your original song or private portrait unless support specifically needs it.

Troubleshooting common problems

The song will not upload

Confirm that it is MP3, M4A, or WAV, no larger than 50 MB, and no longer than 3 minutes. If the browser cannot read its duration, export a fresh copy from your audio editor instead of only changing the filename extension.

I clicked Start, but no task appears

Check whether the song or a portrait still says Uploading. The button cannot create a durable task until the song is ready and all portrait uploads have finished. If every upload is ready, try once more and read the error shown above the form.

The task looks stuck at Cast

Open it. Cast is a real approval checkpoint, not only a progress label. Check character analysis, correct labels and roles, save any relationship direction, then click Approve cast. Song analysis may also still be running; the live handoff banner will say so.

There is no separate ASR task in the history

Speech recognition is an internal preparation phase of the Music Video task. You follow the single Story Music Video task, then review its Transcript checkpoint. You do not need to create or track a second ASR project.

The transcript is wrong

Correct the affected lines before approval. For many errors, re-export cleaner audio with the vocal more audible and begin a new project. Do not approve bad lyrics and hope the treatment will ignore them.

The quote expired

Click Refresh quote or Get quote again. An expired quote cannot be committed because the storyboard, reuse scope, balance, or provider conditions may have changed.

A generated face does not match

Compare variants and read the QA warning. Select a stronger valid variant or use an included retry. If the reference itself is ambiguous, correct it at Cast before approving expensive downstream work in a new revision.

The task says Action required

Your completed checkpoints are still saved. Open the task, use the recovery action shown in the error banner, and copy the Debug ID if the retry fails. Do not create a duplicate unless the interface explicitly says the existing task cannot continue.

I cannot download the final video

That is the current V2 product boundary, not a hidden button. The Final checkpoint provides the full-song browser composition and durable approved clips. Downloadable MP4 render/export is not wired yet.

A practical quality checklist

Before each approval, ask one short set of questions:

Cast: Are these the correct people, with the right lead and relationship?

Transcript: Are the important words, names, and repeated hooks exact?

Treatment: Does this idea develop across the whole song and earn its ending?

Storyboard: Does every shot add something, and do transitions make visual sense?

Frames: Is identity stable, is the composition strong in the chosen aspect ratio, and does the animatic flow with the song?

Representative motion: Do the hardest sample clips preserve identity, duration, and frame continuity?

Final motion: Is every range present, is the timeline synchronized, and is the selected variant truly the best available one?

The most expensive mistake is approving weak planning and trying to repair it with more generation. Spend your attention on Transcript, Treatment, and Storyboard; spend credits only after those decisions are solid.

The short version

Upload a final MP3, M4A, or WAV under 50 MB and 3 minutes.
Optionally upload one or two clear character portraits.
Write a focused style brief and choose 9:16 or 16:9.
Start the Director workflow and wait until the task appears in history.
Approve Cast, correct Transcript, select a Treatment, and inspect the full Storyboard.
Review the exact frame quote, generate variants, and approve the full-song animatic.
Generate representative motion first; approve it before paying for the remaining clips.
Review and approve the full-song browser composition in Final.
If work stops, recover the same saved task with its visible action and Debug ID instead of submitting a duplicate.

Start a guided music video in DojoClip Story Music Video