dreamina-cli-image2video

Jimeng CLI image-to-video

⚠️ Maintenance Discontinuation Notice: Legacy Dreamina CLI (dreamina, v1.4.18) will from November 2026 cease maintenance. For new tasks, switch to the Canvas replacement skill dreamina-canvas-cli-ref2video(install:npx skills add full-aigc-skills/dreamina-skills --skill dreamina-canvas-cli-ref2video). The content of this skill is retained as a historical contract reference.

First select the command based on the input intent, then run the corresponding dreamina <command> -h：

When to use and boundary

For CLI video tasks with at least one reference image or one reference video. Should not be used for text-only video, image generation, or pure prompt polishing; for these scenarios, switch to the corresponding dreamina-cli-* or dreamina-prompt-* skill.

Collaboration routing:dreamina-prompt-image2video handles single image, first/last frame, multi-frame, and multimodal prompts; this Skill handles local dreamina CLI execution; when the user explicitly chooses OpenCLI transport, switch to dreamina-opencli-image2video, do not submit two execution chains at the same time.

Scenario	Command	Key input
Single-image animation	image2video	--image, required --prompt
First/last frame transition	frames2video	--first、--last
2–20 storyboard images	multiframe2video	--images
Image/video/audio all-purpose reference	multimodal2video	At least one image, one video, or one audio

v1.4.18 hard constraints

• All four modes must explicitly pass --video_resolution。
• Resolution token uses lowercase 480p/720p/1080p/4k；480p only seedance2.5 supports.
• multiframe2video Fixed model, does not accept seedance2.5, only accepts 720p/1080p.
• seedance2.5 Supports 480p/720p/1080p, does not support 4k; for other model combinations, refer to runtime help.
• image2video Supports seedance1.0fast、seedance1.5pro, Seedance 2.0 family and seedance2.5。
• frames2video Supports seedance1.5pro, Seedance 2.0 family and seedance2.5。
• multimodal2video Supports Seedance 2.0 family and seedance2.5；seedance2.5 allowspure audioinput.
• seedance2.5 Duration 4–30 seconds, total duration of reference audio/video 2–30 seconds.
• Aspect ratio must be discovered at runtime via dreamina <command> --help :image2video、frames2video、 multimodal2video exposes --ratio，multiframe2video does not expose and follows the first image.
• image2video and frames2video when using seedance2.5 it is forbidden to explicitly pass --ratio, aspect ratio follows the first frame; multimodal2video can explicitly pass aspect ratio, defaults to 16:9 when omitted.

For the complete matrix, see the v1.4.18 contract of the dreamina-cli skill(if not installed:npx skills add full-aigc-skills/dreamina-skills --skill dreamina-cli）。

Minimal correct example

dreamina image2video --image=./photo.png --prompt="镜头缓慢推近" --video_resolution=720p --poll=0
dreamina frames2video --first=./start.png --last=./end.png --prompt="季节自然变化" --video_resolution=720p --poll=0
dreamina multiframe2video --images=./a.png,./b.png --prompt="人物转身走向远处" --video_resolution=720p --poll=0
dreamina multimodal2video --image=./subject.png --audio=./music.mp3 --prompt="按音乐节奏运镜" --video_resolution=720p --poll=0

For a multi-frame story of 3 or more images, provide N−1 transitions for every N images; single segment duration 1–8 seconds:

dreamina multiframe2video \
  --images=./a.png,./b.png,./c.png \
  --transition-prompt="从 A 走向 B" \
  --transition-prompt="从 B 走向 C" \
  --transition-duration=3 \
  --transition-duration=3 \
  --video_resolution=1080p \
  --poll=0

Asynchronous closed loop

Step 1: Routing and validation

Select one of the four commands based on the input intent, verify that files are readable, then use the corresponding --help to validate model, duration, aspect ratio, and resolution at runtime.

Step 2: Submission

Check credits, explain the consumption impact, submit asynchronously, and save submit_id。

Step 3: Final-state closed loop

For querying continuously call query_result, until success or fail. On failure, report fail_reason. When encountering AigcComplianceConfirmationRequired the user is prompted to complete authorization on the Web side first.

Gotchas

1. Mode misrouting: the parameter names differ for single image, first/last frame, storyboard, and omni-reference.
2. Missing resolution: all four commands must pass video_resolution；seedance2.5 supports 480p/720p/1080p.
3. Multi-frame privilege escalation: multiframe does not accept model_version, cannot use 4k, and cannot use seedance2.5。
4. transition count: N images require N−1 transitions.
5. Aspect ratio misconfiguration：seedance2.5 image/frames mode prohibits explicit aspect ratios; multiframe has no --ratio。
6. Audio submitted separately：multimodal2video 's seedance2.5 allows pure audio input (reference audio/video total duration 2–30 seconds).
7. Duration out of range：seedance2.5 4–30 seconds; Seedance 2.0 family 4–15 seconds; exceeding it will be rejected.

References

• dreamina-cli skill's v1.4.18 parameter contract(if not installed:npx skills add full-aigc-skills/dreamina-skills --skill dreamina-cli）
• Mode selection
• Parameter reference
• VIP guide
• Workflow mode
• Examples

When to use

Load this skill when the user needs to execute generation or write operations after clarifying inputs, budget, and delivery constraints . First extract the goal, inputs, constraints, delivery format, and acceptance criteria from the request; the description summary is: Use when the user provides image, video, or audio references and wants Dreamina 即梦 video generation through image2video, frames2video, multiframe2video, or multimodal2video. Covers CLI v1.4.18 runtime ratio discovery, required video resolution, Seedance model constraints, input limits, and async terminal statuses.。

Rules

• Read before write: first confirm the current state and real capabilities, then perform actions that change external state.
• Least privilege: only use the files, tools, accounts, and network scope required to complete the current step.
• Evidence first: when run results, resource IDs, versions, hashes, or test outputs are missing, clearly mark them as NOT_VERIFIED。
• Idempotency first: preserve request identifiers and phase status; when results are unclear, query first rather than blindly retrying.
• Privacy and security: logs, examples, receipts, and error messages must not contain tokens, cookies, keys, or personally sensitive data.

Workflow

Step 1: Clarify intent

Confirm whether this skill matches the goal; if it is only an adjacent need, hand it off to a more precise skill.

Step 2: Run prechecks

Validate inputs, model/tool capabilities, output paths, budget limits, and approval status; if any critical condition is unknown, stop at the read-only stage.

Step 3: Form a plan

List the tools to be called, the objects that will change, the success criteria, and the safe exit method after failure.

Step 4: Execute actions

Execute once approved and record the request identifier; for ambiguous results, query first instead of resubmitting; preserve an associable status or receipt for every external call.

Step 5: Verify delivery

Verify artifact existence, format, hash/identifier, cost status, and quality gates, and state facts, inferences, and unverified items separately.

Validation checklist

• The skill trigger conditions are consistent with the user's intent, and adjacent tasks have not been misrouted to this skill.
• Inputs, target objects, versions, and output locations are all clear, and no guessed values have been used in place of required values.
• All write, paid, publish, or irreversible actions are within the scope of user authorization.
• Results have been verified with independent checks; only "command succeeded" or "file exists" does not count as complete acceptance.
• Output includes actual evidence, failed/skipped items, remaining risks, and actionable next steps.

Gotchas

1. Treating a plan as a result: Documentation or prompts are not the same as actual execution; the actual execution level must be indicated.
2. Retrying on error: A timeout or lost response may already have produced remote state; query first before deciding whether to retry.
3. Implicit scope expansion: Batch operations, full-volume operations, publishing, overwriting, and paid actions are not natural extensions of ordinary reads and writes.
4. Version drift: When referencing external resources, record the version, tag, or commit; do not treat a mutable branch as release evidence.
5. Stale evidence: Caches, old screenshots, and historical tests cannot prove the current environment; refresh key evidence before delivery.

Non-applicability and boundaries

Paid actions, publishing, overwriting, uploading, or external writes must use the explicit authorization of the current task; do not automatically expand the number of times or the budget. If the request requires another skill, do not copy its body text; hand off by skill name and preserve the current task context.

Progressive disclosure

• When inputs/outputs, states, and authorization points need to be determined, read references/workflow-contract.md。
• When pre-delivery self-checks are needed, read references/validation-checklist.md。
• When encountering timeouts, partial success, or recovery scenarios, read references/error-recovery.md。
• For first runs, refusal of privilege escalation, and failure recovery, refer respectively to examples/happy-path.md、examples/boundary-refusal.md、examples/failure-recovery.md。