rw-integrate-uploads（rw 集成上传）

Integrate Uploads

PREREQUISITE: Run +rw-check-compatibility first. Run +rw-fetch-api-reference to load the latest API reference before integrating. Requires +rw-setup-api-key for API credentials.

Help users upload local files (images, videos, audio) to Runway's ephemeral storage for use as inputs to generation models.

When to Use Uploads

Use the Uploads API when:

• The user has a local file (not a public URL) they want to use as input
• The file exceeds data URI size limits (5 MB for images, 16 MB for video/audio)
• The file's URL doesn't meet Runway's URL requirements (HTTPS, proper headers, no redirects)

You do NOT need uploads when:

• The asset is already at a public HTTPS URL with proper headers
• The asset is small enough for a data URI (< 5 MB image, < 16 MB video)

How It Works

1. Request an ephemeral upload slot → get a presigned upload URL and form fields
2. Upload the file to the presigned URL
3. Use the returned runway:// URI as input to any generation endpoint

runway:// URIs are valid for 24 hours.

SDK Upload (Recommended)

Node.js

import RunwayML from '@runwayml/sdk';
import fs from 'fs';

const client = new RunwayML();

// Upload from a file stream
const upload = await client.uploads.createEphemeral(
  fs.createReadStream('/path/to/image.jpg')
);

// Use the runway:// URI in any generation call
const task = await client.imageToVideo.create({
  model: 'gen4.5',
  promptImage: upload.runwayUri,
  promptText: 'The scene comes to life',
  ratio: '1280:720',
  duration: 5
}).waitForTaskOutput();

The Node.js SDK accepts:

• fs.ReadStream — file streams
• File objects — from web APIs
• Blob objects
• Buffer / ArrayBuffer / typed arrays
• Response objects — from fetch()
• Async iterables

Python

from runwayml import RunwayML
from pathlib import Path

client = RunwayML()

# Upload from a file path
upload = client.uploads.create_ephemeral(
    Path('/path/to/image.jpg')
)

# Use the runway:// URI
task = client.image_to_video.create(
    model='gen4.5',
    prompt_image=upload.runway_uri,
    prompt_text='The scene comes to life',
    ratio='1280:720',
    duration=5
).wait_for_task_output()

The Python SDK accepts:

• pathlib.Path objects
• IOBase objects (file-like objects)
• Two-tuples of (filename, content)

REST API Upload (Manual)

If not using the SDK, the upload flow has three steps:

Step 1: Create an upload slot

const response = await fetch('https://api.dev.runwayml.com/v1/uploads', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${process.env.RUNWAYML_API_SECRET}`,
    'X-Runway-Version': '2024-11-06',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    filename: 'image.jpg',
    type: 'ephemeral'
  })
});

const { uploadUrl, fields, runwayUri } = await response.json();

Step 2: Upload the file using the presigned URL

const formData = new FormData();

// Add all presigned form fields first
for (const [key, value] of Object.entries(fields)) {
  formData.append(key, value);
}

// Add the file last
formData.append('file', fileBuffer, 'image.jpg');

await fetch(uploadUrl, {
  method: 'POST',
  body: formData
});

Step 3: Use the runway:// URI

const task = await fetch('https://api.dev.runwayml.com/v1/image_to_video', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${process.env.RUNWAYML_API_SECRET}`,
    'X-Runway-Version': '2024-11-06',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: 'gen4.5',
    promptImage: runwayUri,
    promptText: 'Animate this scene',
    ratio: '1280:720',
    duration: 5
  })
});

Upload Constraints

Integration Pattern

Express.js — Upload Endpoint with File Generation

import RunwayML from '@runwayml/sdk';
import express from 'express';
import multer from 'multer';

const client = new RunwayML();
const app = express();
const upload = multer({ storage: multer.memoryStorage() });

app.post('/api/image-to-video', upload.single('image'), async (req, res) => {
  try {
    // Upload the user's file to Runway
    const runwayUpload = await client.uploads.createEphemeral(req.file.buffer);

    // Use the uploaded file for video generation
    const task = await client.imageToVideo.create({
      model: 'gen4.5',
      promptImage: runwayUpload.runwayUri,
      promptText: req.body.prompt || 'Animate this image',
      ratio: '1280:720',
      duration: 5
    }).waitForTaskOutput();

    res.json({ videoUrl: task.output[0] });
  } catch (error) {
    console.error('Generation failed:', error);
    res.status(500).json({ error: error.message });
  }
});

Next.js — Upload + Generate

// app/api/image-to-video/route.ts
import RunwayML from '@runwayml/sdk';
import { NextRequest, NextResponse } from 'next/server';

const client = new RunwayML();

export async function POST(request: NextRequest) {
  const formData = await request.formData();
  const imageFile = formData.get('image') as File;
  const prompt = formData.get('prompt') as string;

  try {
    // Upload file to Runway
    const upload = await client.uploads.createEphemeral(imageFile);

    // Generate video from the uploaded image
    const task = await client.imageToVideo.create({
      model: 'gen4.5',
      promptImage: upload.runwayUri,
      promptText: prompt || 'Animate this image',
      ratio: '1280:720',
      duration: 5
    }).waitForTaskOutput();

    return NextResponse.json({ videoUrl: task.output[0] });
  } catch (error) {
    return NextResponse.json(
      { error: error instanceof Error ? error.message : 'Failed' },
      { status: 500 }
    );
  }
}

FastAPI — Upload + Generate

from fastapi import FastAPI, UploadFile, Form, HTTPException
from runwayml import RunwayML

app = FastAPI()
client = RunwayML()

@app.post("/api/image-to-video")
async def image_to_video(image: UploadFile, prompt: str = Form("Animate this image")):
    try:
        # Upload to Runway
        content = await image.read()
        upload = client.uploads.create_ephemeral((image.filename, content))

        # Generate video
        task = client.image_to_video.create(
            model="gen4.5",
            prompt_image=upload.runway_uri,
            prompt_text=prompt,
            ratio="1280:720",
            duration=5
        ).wait_for_task_output()

        return {"video_url": task.output[0]}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

Tips

• Always upload local files before passing them to generation endpoints. Don't try to pass local file paths — they won't work.
• runway:// URIs expire after 24 hours. If you need to re-use an asset, upload it again.
• The SDK handles the presigned URL flow automatically — prefer the SDK over manual REST calls.
• For models requiring image/video input (image-to-video, video-to-video, character performance), upload the asset first, then pass the runway:// URI.
• Maximum 200 MB per file via uploads — larger than URL (16 MB) or data URI (5 MB) limits.