Skip to main content

​
Overview

Soul (by Higgsfield) is a character-focused image generation model with 80+ style presets. It’s designed for creating consistent, stylized character images that can be reused across video scenes.
Model IDCredits~Cost
soul15$0.15

​
Quick start

import { createVarg } from "vargai/ai"

const varg = createVarg({ apiKey: process.env.VARG_API_KEY! })

const result = await varg.imageModel("soul").generate({
  prompt: "friendly tech influencer, casual hoodie, ring light, warm smile",
  aspectRatio: "9:16",
  providerOptions: {
    varg: { styleId: "REALISTIC" },
  },
})

console.log(result.image.url)

​
Style presets

Soul supports 80+ style presets. Common ones:
Style IDDescription
REALISTICPhotorealistic rendering
ANIMEAnime/manga style
EDITORIAL_90S90s editorial photography
Y2KY2K aesthetic
GRUNGEGrunge style
Explore all styles through the Higgsfield API — each produces distinctly different character aesthetics.

​
Parameters

​
prompt
string
required
Character description. Focus on appearance, clothing, expression, and setting.
​
aspect_ratio
string
default:"1:1"
Supported: 9:16 (PORTRAIT_1152x2048), 1:1 (SQUARE_1024x1024), 16:9 (LANDSCAPE_2048x1152).
​
provider_options
object
styleId — style preset string (e.g., "REALISTIC", "ANIME"). enhance_prompt — boolean to auto-enhance the prompt. quality — "HD" for high definition. batch_size — 1, 2, or 4 images per request.

​
Pricing

ModelCreditsUSD
soul15$0.15

​
Tips

  • Create a character once, reuse everywhere. Generate a Soul character image, then use it as a reference for video models (image-to-video) across multiple scenes.
  • Style presets are powerful. Each preset dramatically changes the aesthetic. Experiment with different styles for the same character.
  • Batch generation (2 or 4 images) gives you options to pick the best result at the same per-image cost.
  • Portrait (9:16) is ideal for talking head videos and social media content.

​
Composition example

Create a character and animate them across scenes: 
const character = Image({
  model: varg.imageModel("soul"),
  prompt: "young woman, professional, confident, modern office background",
  aspectRatio: "9:16",
  providerOptions: { varg: { styleId: "REALISTIC" } },
})

// Animate the character
const video = Video({
  model: varg.videoModel("kling-v3"),
  prompt: { text: "she turns and walks toward the camera", images: [character] },
  duration: 5,
})

Phota

Profile-based identity control. Better for consistent faces across many images.

Nano Banana

More versatile generation and editing at lower cost.

Kling V3

Animate Soul characters with image-to-video.