DX.GL

MCP for Video

Markdown MCP for Datasets

Automate turntable video production for e-commerce, marketing, and content teams using AI agents

Contents
What You Can DoSetupYour First PromptPrompt ExamplesQuality TiersAvailable ToolsTipsLearn More

What You Can Do

With a single natural-language prompt, your AI agent can:

No scripting required. Describe what you want in plain English — the agent handles the API calls.


Setup

Add this to your MCP client configuration (Windsurf, Claude Desktop, Cursor, or any MCP-compatible host):

{
  "mcpServers": {
    "dxgl": {
      "serverUrl": "https://mcp.dx.gl/",
      "headers": {
        "Authorization": "Bearer dxgl_sk_..."
      }
    }
  }
}

Replace dxgl_sk_... with your API key from the Portal → API Keys.


Your First Prompt

Paste this into your AI agent to verify everything works:

List my models and tell me how many I have.

The agent will call list_models and return a summary. If this works, you're connected.


Prompt Examples

Render a single model in multiple formats

Render model Ab3kF9x2qL1m in all three aspect ratios (16:9, 1:1, 9:16) with a white background, 9 seconds, shadows enabled.

The agent calls create_batch_renders with 3 variants and polls until all complete.

Render all new models

Render every model that doesn't have any video renders yet. Use 16:9, white background, 6 seconds with shadows.

The agent will:

  1. Call list_models to get all models
  2. Call get_model for each to check render history
  3. Filter to models with no video renders
  4. Call create_batch_renders for the unrendered models
  5. Poll and return results

Full catalog in multiple variants

I just uploaded 30 shoe models tagged "ss26". Render each one in 16:9 white, 1:1 black, and 9:16 with shadows. Quote the total cost first.

The agent will:

  1. Call list_models with tags: "ss26" to find the 30 models
  2. Call quote with 90 renders (30 models × 3 variants) to show total cost
  3. Wait for your confirmation
  4. Call create_batch_renders to submit all 90 at once
  5. Credits are deducted atomically — if you don't have enough, nothing is charged
  6. Poll and return video URLs when complete

E-commerce product page assets

For all models tagged "hero-products", create a 16:9 video with reflection on black background for the product page, plus a 1:1 video with shadows on white for social media. 9 seconds each, with easing.

The agent creates 2 variants per model across the tagged set.

4K renders for large displays

Render my top 10 furniture models in 4K quality, 16:9, white background with reflection. These are for in-store displays.

The agent uses quality: "4k" (4 credits each) for high-resolution output suitable for 4K screens and kiosks.

ProRes for video editing

Create Pro renders of models tagged "campaign-q2" in 16:9 with transparent background. I need ProRes 4444 for compositing in DaVinci Resolve.

The agent uses quality: "pro" (16 credits each). The .mov files have alpha channels — the background color only applies to the web/thumbnail/poster variants.

Import and render in one go

Import this GLB from https://example.com/chair.glb, then render it in 16:9 and 1:1 with a white background.

The agent chains ingest_model → waits for system preview → create_batch_renders with 2 variants → polls → returns video URLs.

Custom backgrounds

Render all models tagged "luxury" with my gradient background image (overlay ID abc123DEF456) in 16:9 and 9:16.

The agent uses bgImageId in render settings to apply your uploaded background image instead of a solid color.

Rotation and tilt

Render my shoe models with a 45° rotation so the toe faces the camera, and a 15° forward tilt to show the sole. 1:1, white background, shadows.

The agent uses rotateY: 45 and tiltX: 15 to control the starting camera angle and product tilt.

Budget check

I have 150 models. How many credits would it cost to render each one in 16:9 white and 1:1 black at standard quality?

The agent calls quote with 300 renders (150 × 2 variants × 1 credit each), checks your balance, and reports the total.

Download all videos

Give me the download URLs for all completed renders of models tagged "fall-collection".

The agent calls list_models with the tag filter, then get_model for each, and compiles download URLs from download_render for every completed render.


Quality Tiers

Tier Resolution Codec Alpha Credits Best for
standard 1920×1080 H.264 MP4 No 1 Web, social media, product pages
4k 3840×2160 H.264 MP4 No 4 In-store displays, kiosks, high-DPI screens
pro 3840×2160 ProRes 4444 Yes 16 Video editing, compositing, broadcast

Render Settings

Setting Options Default
Aspect 16:9, 1:1, 9:16 16:9
Background Any hex color (e.g. #ffffff), or bgImageId for uploaded images #ffffff
Length 3–30 seconds 6
Surface shadows: true or reflector: true or neither Shadows
Easing Smooth start/stop On
Rotation rotateY: −180° to 180° (starting angle)
Tilt tiltX: −90° to 90° (forward/back lean)

Output Files

Every completed render produces:


Available Tools

Tool What it does
ingest_model Import a 3D model from a URL
list_models List models with tag and pagination filters
get_model Get model details and all its renders
create_render Create a single video render
create_batch_renders Create multiple renders atomically
get_render Poll render status until complete
download_render Get the download URL for a completed video
get_account Check your credit balance
quote Estimate credit cost before committing

Tips

Tag everything

Tags are the key to efficient batch operations. Tag models on upload or via the portal, then target them in prompts:

Render all models tagged "new-arrivals" in 16:9 white with shadows.

Quote first on large batches

The quote tool prevents surprises. Ask the agent to quote before any batch:

Quote 16:9 + 1:1 + 9:16 renders for my 80 shoe models, then proceed only if it's under 250 credits.

Atomic credit deduction

Batch renders deduct all credits at once. If you don't have enough for the full batch, nothing is charged — the batch fails cleanly and you can adjust.

Use the right quality tier

Combine with background images

Upload branded backgrounds via the portal, then reference them in batch renders. The background image is composited during encoding, so all variants in a batch share the same GPU render pass — efficient even at scale.

Chain operations

Your agent can orchestrate multi-step workflows:

Import all GLBs from this spreadsheet of URLs, tag them by category, render each in our standard 3-variant format, and compile a CSV of model names and video URLs.


Learn More