SonoVision is now live on SeaArt! It supports both text-to-video and image-to-video. Beyond high video quality, it natively generates audio—so you can produce voice-over videos from a given audio file, or control dialogue, background music, and sound effects directly via prompts.

Click to try Sonovision now:

SeaArt SonoVision Img2Vid

SeaArt SonoVision Text2Vid

Keynode Intro

	SeaArtSonoVision_T2V	SeaArtSonoVision_i2V
Inputs	audio: connect to a specific audio file  optional	image：connect to your upload image required audio：connect to a specific audio file  optional
Core Parameters	prompt / negative_prompt duration：support 5s or 10s prompt_extend：true / false (prompt enhancement/rewriting) size：resolution（width × height） support 1280 × 720 / 960 × 960 / 720 × 1280 / 1088 × 832 / 832 × 1088 seed：radom seeds control after generate：seeds control	prompt / negative_prompt duration：support 5s or 10s prompt_extend：true / false (prompt enhancement/rewriting) resolution： support 480p / 720p / 1080p seed：radom seeds control after generate：seeds control

Below we'll explain the workflow in detail, following the sequence of text-to-video and image-to-video generation.

Let's begin!

SeaArtSonoVision-T2V-Workflow

Click to try Sonovision-T2V-Workflow

node preview

Price (credits)

Sample

Camera Shot: A high-angle, third-person perspective shot following a seagull in one continuous take; the camera position is slightly above the seagull, steadily tracking it. Time slows down (100% to 40% to 100%) as the shark emerges from the water.

Scene/Time: Open sea on a clear day, strong winds and large waves, obvious splashes and spray, bright reflections on the sea surface.

Subject/Action: The camera follows a seagull as it skims over a wave crest; a giant shark, approximately 7 meters long, suddenly bursts through the water behind it, its mouth opening wide. Time slows down, showing the tense moment of "passing by"; the seagull quickly rises its wings to escape, and the shark plunges back into the sea, creating a huge splash.

Style: Cinematic realism, IMAX scale, HDR, natural motion blur, volumetric water mist and spray.

Lighting/Tone: High-altitude sunlight, cool blue sea color, warm-toned rim lighting on the seagull's feathers; obvious highlights on the water surface and refraction from the spray.

SeaArtSonoVision-I2V-Workflow

Click to try Sonovision-T2V-Workflow

node preview

Price (credits)

Sample

A ladybug flew past a lizard, and the lizard instantly flicked its tongue, trapping the ladybug and then retracting it just as quickly.

Tips:Prompt Framework

Compose prompts with:

• Subject:  who/appearance/quantity
• Action: what they are doing
• Scene/Time:  location/time/weather
• Camera:  one motion + camera position
• Visual Style:  style/materials/lens language
• Lighting & Tone: key/fill, color temp, primary color
• Mood:  atmosphere keywords
• Audio: dialogue (language/gender/tone/lines), music (style/instruments), SFX (source)

Example structure:

• Subject + Action + Scene/Time + Camera + Visual Style + Lighting/Tone + Mood + Audio

Extension—Native Audio Support

You can describe audio directly in the prompt (background music, ambient SFX, or spoken lines), or specify an external audio file

Tips：When producing videos with audio, please emphasise the following elements in your prompt: (Subject + Facial Expression + Tone of Voice + Character Dialogue)

T2V with Designated Audio

• Add a LoadAudio node and connect it to SeaArtSonoVision_T2V.
• Supports human voice, pure music, and environmental sounds.

📃Tips:

• Even with a voice file, describe who is speaking and their expression/tone.
• Keep audio length aligned with video duration.

🔍Q&A:

• Q: If audio length ≠ video duration, will the task fail?
• A: No. The duration (5s/10s) defines how much audio is read. For a 5s video, only the first 5s of longer audio is used; shorter audio behaves similarly. Best practice: match lengths.

I2V with Designated Audio

Same usage as T2V. With a voice clip, you can make the character in the image speak.

📃Tips: Explicitly state the character is speaking and describe their expression, otherwise lip‑sync may not trigger.

Sample

🔍FAQ

• Lip‑sync off:
• Use shorter, clearer lines; natural speaking pace.
• Reduce environmental noise in descriptions; ensure the audio contains human voice.
• Face drift (I2V):
• Use a clearer, frontal reference image.
• Soften camera‑motion instructions in the prompt.
• Optionally turn off prompt_extend for better identity consistency.

-------------------------

That’s the basic guide to the SonoVision node. Hope it helps—give it a try!🔥