VPick Video Creator#

All-in-one AI video production studio — from image generation to video creation, voiceover, music, lip-sync, and final export — all on a visual canvas. Powered by VPick.

Security & Data Handling#

• Authentication: Your MCP link contains an embedded token — no separate API keys or credentials are needed or sent. Treat your MCP link like a password.
• Data flow: All generation requests go through VPick's server ( vpick.10xboost.org on Google Cloud). VPick routes requests to third-party AI model providers (Kling, Veo, Runway, Sora, ElevenLabs, Suno) on your behalf. Your prompts and uploaded media are sent to these providers for processing.
• Storage: Generated files are stored in Google Cloud Storage under your VPick account. Uploaded images/videos are used only for generation and stored in your project.
• No local credentials: This skill does not require any local API keys, environment variables, or secrets. All auth is embedded in the MCP link.
• Billing: Generation costs are charged to your VPick credit balance, not directly to any third-party service.

Prerequisites#

1. Sign up at vpick.10xboost.org (Google OAuth — new users get $1 free credit)
2. Get your MCP link: Go to Settings → copy your MCP Server URL (contains your auth token)
3. Add to Claude: Paste the MCP link into Claude settings as a Connector — no install, no API keys needed
4. Top up credits if needed — generation costs are deducted from your VPick balance

Supported Models#

Video Generation Models#

ModelDurationSoundCostMultiShotBest ForKling 3.0 Standard3-10sYes~$0.30/secYesMultiShot scenes, character consistencyKling 3.0 Professional3-10sYes~$0.405/secYesHigher quality MultiShotVeo 3.1 Fast8s fixedYes$0.90/videoNoQuick high-quality clipsSora 210-15sYes$0.525-$0.60NoCreative, artistic videosRunway 720p/1080p5-10sNo$0.18-$0.45NoFast iterationGrok Imagine6-15sYes$0.15-$0.60NoBudget-friendly with audioMidjourney Video5sNo$0.90NoStylized, artistic clips

Image Generation Models (for references, storyboards, thumbnails)#

ModelCostOutputBest Fornano-banana-2$0.16/image1 imageDefault, fast, multi-referenceMidjourney (relaxed/fast/turbo)$0.045-$0.24/grid4 imagesArtistic, stylizedGrok Imagine$0.06/call6 imagesBulk, budgetSeedream 5.0 (Lite/HD)$0.0825/image1 image (2K-3K)High resolution

Audio & Voice Models#

ModelTypeCostFeaturesElevenLabs V3Text-to-Speech$0.21/1000 chars29+ voices, multi-language, stability controlSuno V4.5Music Generation$0.10/songCustom style, instrumental toggle, vocal genderKling AI AvatarLip Sync$0.12/secFace animation from image + audioDemucsVocal Separation$0.30/callIsolate vocals/accompaniment from audioElevenLabs STS v2Voice ChangerFree (user API key)Speech-to-speech, noise removal

Production Pipeline Overview#

VPick covers the entire video production workflow in one place:

text

Image Gen → Video Gen → Voiceover/Music → Lip Sync → Vocal/Voice Edit → Combine & Export

1. Pre-production: Generate character designs, environments, storyboard frames
2. Video generation: Single shot or MultiShot with character consistency
3. Audio production: Voiceover (ElevenLabs TTS), music (Suno), vocal separation (Demucs), voice changing
4. Lip sync: Animate a face image to speak with generated audio (Kling AI Avatar)
5. Post-production: Combine video clips, mix audio tracks, export final video

Core Workflow#

Step 1: Set Up the Canvas#

Start by understanding the user's project:

• Call get_canvas to see the current state
• Call list_projects to check existing projects
• Use create_project if starting fresh

Step 2: Prepare Assets#

Create input nodes on the canvas:

Text prompts:

text

add_node(type: "text", name: "Scene 1 Prompt", data: { content: "A samurai walking through rain, cinematic lighting" })

Reference images (for start/end frames or character consistency):

text

upload_image(url: "https://example.com/character.jpg")

Generate images first if needed:

text

run_image_generator(nodeId: "<image_node_id>", prompt: "samurai portrait, white background", model: "nano-banana-2")

Step 3: Generate Video#

Simple Video (Single Shot)

text

add_node(type: "video-generator", name: "Scene 1")
connect_nodes(sourceId: "<prompt_node>", targetId: "<video_node>", sourceHandle: "text-out", targetHandle: "prompt-in")
connect_nodes(sourceId: "<image_node>", targetId: "<video_node>", sourceHandle: "image-out", targetHandle: "start-image-in")
run_video_generator(nodeId: "<video_node>", model: "kling-3.0", duration: 5, sound: true)

MultiShot Video (Multiple Connected Shots — Kling 3.0 Only)

MultiShot generates 3-15 seconds of video with multiple camera angles and character consistency in a single API call.

text

run_video_generator(
  nodeId: "<video_node>",
  model: "kling-3.0",
  multiShot: true,
  multiPrompt: [
    { "prompt": "@character walks into frame, wide shot", "duration": 4 },
    { "prompt": "@character looks at camera, medium close-up", "duration": 3 },
    { "prompt": "@character turns away, slow dolly out", "duration": 3 }
  ],
  elements: [
    {
      "name": "character",
      "description": "Main protagonist, male samurai",
      "imageUrls": ["https://.../char-front.jpg", "https://.../char-side.jpg"]
    }
  ],
  sound: true
)

MultiShot Rules:

• Minimum 2 reference images per element (white background, isolated figure)
• Element name must exactly match @name in prompts (case-sensitive)
• Total duration: 3-15 seconds
• Max 5 shots per group
• Sound is forced ON in MultiShot mode

Step 4: Audio Production#

Audio is a core part of video production. VPick supports 5 audio tools:

4a. Voiceover — ElevenLabs Text-to-Speech

Generate natural narration or dialogue from text. 29+ built-in voices, multi-language support.

text

add_node(type: "audio-generator", name: "Narration")
run_audio_generator(
  nodeId: "<audio_node>",
  prompt: "The samurai stood alone in the rain, waiting for dawn.",
  model: "elevenlabs",
  voiceId: "<voice_id>",
  stability: 0.5
)

You can connect a Text node as input:

text

connect_nodes(sourceId: "<text_node>", targetId: "<audio_node>", sourceHandle: "text-out", targetHandle: "text-in")

4b. Music Generation — Suno

Create original background music, theme songs, or jingles.

text

add_node(type: "music-generator", name: "BGM")
run_music_generator(
  nodeId: "<music_node>",
  prompt: "epic cinematic orchestral, tension building, dark atmosphere",
  model: "suno",
  instrumental: true,
  style: "cinematic orchestral"
)

Set instrumental: false to include AI-generated vocals with lyrics from the prompt.

4c. Lip Sync — Kling AI Avatar

Animate a character's face to speak with any audio. Turns a still image into a talking head video.

text

add_node(type: "lipsync-generator", name: "Talking Character")
connect_nodes(sourceId: "<face_image>", targetId: "<lipsync_node>", sourceHandle: "image-out", targetHandle: "image-in")
connect_nodes(sourceId: "<audio_node>", targetId: "<lipsync_node>", sourceHandle: "audio-out", targetHandle: "audio-in")
run_lipsync_generator(nodeId: "<lipsync_node>")

Cost: ~$0.12/sec. Great for dialogue scenes, explainer videos, or virtual presenters.

4d. Vocal Separation — Demucs

Isolate vocals from background music in any audio/video file. Outputs: vocals track, accompaniment track, and original.

text

add_node(type: "vocal-separator", name: "Separate Audio")
connect_nodes(sourceId: "<video_or_audio>", targetId: "<separator_node>", ...)
run_vocal_separator(nodeId: "<separator_node>")

Use cases: Extract dialogue from a scene, remove background music, remix audio.

4e. Voice Changer — ElevenLabs Speech-to-Speech

Transform any voice recording into a different voice while preserving speech patterns and emotion.

text

add_node(type: "voice-changer", name: "New Voice")
connect_nodes(sourceId: "<original_audio>", targetId: "<voice_changer_node>", sourceHandle: "audio-out", targetHandle: "audio-in")
run_voice_changer(nodeId: "<voice_changer_node>", voiceId: "<target_voice_id>", removeBackgroundNoise: true)

Requires user's own ElevenLabs API key (free, no credit charge).

4f. Audio Mixing

Combine multiple audio tracks (e.g., voiceover + BGM) into one:

text

add_node(type: "audio-combine", name: "Mixed Audio")
connect_nodes(sourceId: "<voiceover>", targetId: "<mix_node>", sourceHandle: "audio-out", targetHandle: "audio-in")
connect_nodes(sourceId: "<bgm>", targetId: "<mix_node>", sourceHandle: "audio-out", targetHandle: "audio-in")
run_audio_combine(nodeId: "<mix_node>")

Supports up to 10 audio inputs.

Step 5: Combine & Export#

Combine multiple video clips:

text

add_node(type: "combine", name: "Final Video")
connect_nodes(sourceId: "<video_1>", targetId: "<combine_node>", sourceHandle: "video-out", targetHandle: "videos-in")
connect_nodes(sourceId: "<video_2>", targetId: "<combine_node>", sourceHandle: "video-out", targetHandle: "videos-in")
connect_nodes(sourceId: "<bgm_node>", targetId: "<combine_node>", sourceHandle: "audio-out", targetHandle: "audio-in")
run_combine(nodeId: "<combine_node>")

Mix audio tracks:

text

run_audio_combine(nodeId: "<audio_combine_node>")

Step 6: Organize Canvas#

Keep the canvas clean:

text

auto_layout(nodeIds: ["<id1>", "<id2>", ...], direction: "horizontal", spacing: 200)
create_group(nodeIds: ["<id1>", "<id2>", ...], overrides: { label: "Scene 1", color: "#4A90D9" })

Workflows (Automated Pipelines)#

For repeatable processes, create workflows:

text

create_workflow(nodes: [...], edges: [...])
run_workflow(workflowId: "<id>")

Checking Results#

• list_nodes — See all nodes with their generation status and output URLs
• get_node(id) — Get specific node details including generated video/audio URLs
• list_generated_files(limit: 10) — Recent generation history
• get_generation_stats — Usage breakdown by model and cost

Tips#

• Start with image generation to create reference images for characters/environments before video
• Use MultiShot for scenes with character consistency — it's the most powerful feature
• White background reference images work best for Elements
• Check credits before large generations — use get_generation_stats to see spending
• Element name mismatch is the most common bug — always verify @name matches exactly
• Aspect ratio is set once and affects all generations — confirm with user first (16:9, 9:16, 1:1)

Error Handling#

ErrorSolutionGeneration timeoutAuto-retries up to 2 times; check node status with get_nodeInsufficient creditsPrompt user to top up at vpick.10xboost.orgElement name mismatchVerify @name in prompts matches element name exactlyInvalid media formatVideos: MP4 recommended; Images: JPG/PNGNode not foundUse list_nodes to get current node IDs

Documentation#

• VPick App: vpick.10xboost.org
• Available models: Call list_models tool for current pricing and capabilities