EternalChamponD Image Generation

Discover EternalChamponD Image Generation, an innovative model for transforming text into high-quality images. See what makes this AI model special!

Platform: Replicate

Text to ImageImage RefinementPrompt-Based GenerationAI Art Creation

854 runs

L40S

License Check Required

🚀Function Overview

Generates images from textual descriptions with extensive customization options including prompt weighting, negative prompts, detail enhancers, and post-processing controls.

Key Features

Compel prompt weighting syntax support
Multi-object handling via conjunction syntax
ADetailer modules for face/hand/person enhancement
Configurable CFG/PAG scales for prompt adherence
Refiner functionality for detail correction
CLIP layer skipping for stylistic control

Use Cases

•Creating photorealistic scenes from text descriptions
•Generating character art for gaming/media
•Producing advertising visuals
•Enhancing anatomical details in AI-generated images

⚙️Input Parameters

model

string

The model to use

vae

string

The VAE to use

prompt_conjunction

boolean

Force Compel conjunction syntax on prompt, separating linebreaks into different items on the Compel conjunction function

prompt

string

The prompt, uses Compel weighting syntax

negative_prompt

string

The negative prompt (for things you don't want), uses Compel weighting syntax

negative_prompt_conjunction

boolean

Force Compel conjunction syntax on prompt, separating linebreaks into different items on the Compel conjunction function

cfg_scale

number

CFG scale defines how much attention the model pays to the prompt when generating, set to 1 to disable

guidance_rescale

number

Amount to rescale CFG generated noise to avoid overexposed images, set to 0 or 1 to disable

pag_scale

number

PAG scale is similar to CFG but improves the result; set to 0 to disable

clip_skip

integer

How many CLIP layers to skip; 1 means no skip

width

integer

The width of the image

height

integer

The height of the image

upscale

string

Upscale factor

refiner

boolean

Enable image refiner using model

refiner_prompt

string

The prompt of refiner (leave empty for default correct hands and details)

refiner_strength

number

Refiner strength (0.0 to 1.0)

prepend_preprompt

boolean

Prepend preprompt (Prompt: "score_9, score_8_up, score_7_up, score_6_up, photorealistic," | Negative prompt: "score_6,score_5,score_4,from behind,pornpics-com style watermark,, worst quality, low quality, bad anatomy, bad hands, missing fingers, fewer digits, source_furry, source_pony, source_cartoon,3d, blurry,anime,not Realistic,messy face,")

scheduler

string

The scheduler to use

steps

integer

Number of steps when generating

seed

integer

The seed for generation; set to -1 for random seed

adetailer_face

boolean

Use ADetailer face_yolov9c.pt

adetailer_face_prompt

string

ADetailer face prompt (if blank, will use the image prompt)

adetailer_face_negative_prompt

string

ADetailer face negative prompt (if blank, will use: blurry, deformed face, extra features, unrealistic skin texture, oversmoothed, low resolution, artifacts)

adetailer_hand

boolean

Use ADetailer hand_yolov9c.pt

adetailer_hand_prompt

string

ADetailer hand prompt (if blank, will use: detailed hands, realistic anatomy, well-defined fingers, natural pose, soft shadows, high resolution, intricate details)

adetailer_hand_negative_prompt

string

ADetailer hand negative prompt (if blank, will use: extra fingers, fused or missing fingers, blurry, distorted, low detail, poorly drawn, unnatural proportions)

adetailer_person

boolean

Use ADetailer person_yolov8m-seg.pt

adetailer_person_prompt

string

ADetailer person prompt (if blank, will use: best quality, detailed, realistic shadows)

adetailer_person_negative_prompt

string

ADetailer person negative prompt (if blank, will use: deformed, unrealistic proportions, extra limbs, blurred, low quality, distorted pose, exaggerated features)

💡Usage Examples

Example 1

Input Parameters

{
  "vae": "NeptuniaXL-VAE-ContrastSaturation",
  "seed": 1,
  "model": "EternalChamponD",
  "steps": 30,
  "width": 1024,
  "height": 1024,
  "prompt": "street, 1girl, dark-purple short hair, purple eyes, medium breasts, cleavage, casual clothes, smile",
  "refiner": false,
  "upscale": "Original",
  "cfg_scale": 3,
  "clip_skip": 0,
  "pag_scale": 0,
  "scheduler": "DPM2 a Karras",
  "adetailer_face": false,
  "adetailer_hand": false,
  "refiner_prompt": "",
  "negative_prompt": "nsfw, naked",
  "adetailer_person": false,
  "guidance_rescale": 1,
  "refiner_strength": 0.4,
  "prepend_preprompt": true,
  "prompt_conjunction": false,
  "adetailer_face_prompt": "",
  "adetailer_hand_prompt": "",
  "adetailer_person_prompt": "",
  "negative_prompt_conjunction": false,
  "adetailer_face_negative_prompt": "",
  "adetailer_hand_negative_prompt": "",
  "adetailer_person_negative_prompt": ""
}

Output Results

https://replicate.delivery/xezq/Z0izZE5vNpYdMV0e1ovfdWAf5bPIT0VoqRHOqeZmEeMHypZjC/0.png

Quick Actions

Use NowView Documentation

Technical Specifications

Hardware Type: L40S
Run Count: 854
Commercial Use: Unknown/Restricted
Platform: Replicate

Related Keywords

Text to ImageImage RefinementPrompt-Based GenerationAI Art CreationPhotorealistic Image CreationCharacter ArtAdvertising VisualsAnatomical Detail Enhancement

Related Models

ProjectIL v3 Image Generation

A model for generating images based on text prompts with detailed control over generation parameters.

Ziba AI Advertising Persona

Ziba is a young girl advertising persona

Sasha Advertising Persona

Sasha is a young boy advertising persona