UI はローカライズ済みです。スキルの紹介と説明文書は選択した言語へ AI が自動翻訳し、コード・コマンド・パス・専門用語は原文のままです。原作者の文書が優先されます。
スキルを探す
生産性・共同作業 · SKILL.md

reve-fast-remix

reve-fast-remix

Reve | Fast | Remix. Remix images using reference images and text prompts. Triggers: image remix, reve, fast, reference images, style remix

Independent GitHub snapshot · Automated classification · Original documentation governs

スキルの説明

GitHub

name: reve-fast-remix description: "Reve | Fast | Remix. Remix images using reference images and text prompts. Triggers: image remix, reve, fast, reference images, style remix" allowed-tools: Bash(curl *), WebFetch

Reve | Fast | Remix

Remix images using 1-4 reference images combined with text prompts. Use XML img tags to reference specific images in the prompt for precise control.

Quick Start

Requires an each::labs API key. Get one at eachlabs.ai.

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "reve-fast-remix",
    "version": "0.0.1",
    "input": {
      "prompt": "combine the style of <img>0</img> with the composition of <img>1</img>",
      "image_urls": ["https://example.com/style-ref.jpg", "https://example.com/composition-ref.jpg"],
      "aspect_ratio": "16:9"
    }
  }'

Parameters

ParameterTypeDefaultDescription
aspect_ratiostring16:9The aspect ratio of the generated image. enum: 16:9, 9:16, 3:2, 2:3, 4:3, 3:4, 1:1
image_urlsarrayList of URLs of reference images. 1 to 4 images.
num_imagesinteger1Number of images to generate.
output_formatstringpngOutput format. enum: png, jpeg, webp
promptstringText description. May include XML img tags like <img>0</img> to reference images.

Examples

Style remix from two references:

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "reve-fast-remix",
    "version": "0.0.1",
    "input": {
      "prompt": "create a new artwork combining the color palette from <img>0</img> with the subject from <img>1</img>",
      "image_urls": ["https://example.com/color-ref.jpg", "https://example.com/subject.jpg"],
      "aspect_ratio": "1:1",
      "output_format": "png",
      "num_images": 2
    }
  }'

Multi-image composition:

curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "reve-fast-remix",
    "version": "0.0.1",
    "input": {
      "prompt": "blend all reference images into a harmonious landscape scene",
      "image_urls": ["https://example.com/sky.jpg", "https://example.com/mountains.jpg", "https://example.com/lake.jpg"],
      "aspect_ratio": "16:9",
      "output_format": "jpeg"
    }
  }'

Related Models

Documentation

保存したスキル