⚡ Daily GPU quota exhausted — generations paused.Upgrade to PRO →
NVIDIA H200 ZeroGPU — Live

The Anime AI
Creative Platform

Professional-grade anime generation, editing and upscaling — powered by SDXL with ControlNet, ADetailer, and Compel long-prompt support. Free with any Hugging Face account.

Please note: The site is free and I am not getting any money. Any PRO subscription mentioned is bought directly through Hugging Face, not this site. If you'd like to support development, you can donate at Ko-fi!
Platform Overview
2
SDXL Models
4K
Max Output
4
ControlNet Modes
Free
Base Tier
T2II2IOpenPoseDepthCannyScribbleADetailerCompelReal-ESRGANAnime2Real
Workspace
Seven integrated tools

Each tool connects to Hugging Face ZeroGPU. Outputs forward between tools in one click.

Text to Image

Prompt-driven SDXL generation with Compel weighting, resolution presets, and ADetailer face+hand enhancement.

SDXLCompelADetailer
Image to Image

Transform any image with a prompt. Adjustable strength from subtle refinement to full reimagining.

StrengthAuto-resSDXL
ControlNet

OpenPose skeleton, Intel DPT depth, Canny edge and Scribble sketch guidance. Live preview before generating.

OpenPoseDepthCannyScribble
Character Lab

IP-Adapter reference-driven character consistency. Keep identity while changing pose, scene and style.

IP-AdapterReferenceStandalone
Inpaint

Targeted masked editing. Replace only selected regions while preserving the rest of the image composition.

MaskRegional EditSDXL
4K Upscaler

Real-ESRGAN anime-optimized model. Preserves lineart perfectly. 4× output with lossless PNG export.

Real-ESRGAN4×PNG
ControlNet
Structural guidance modes

Four SDXL-grade ControlNets — prompt controls style, control map controls composition.

🦴

OpenPose

Full body skeleton with hand+face keypoints via xinsir/controlnet-openpose-sdxl-1.0. Lock pose exactly.

🌊

Depth

Intel DPT-hybrid monocular depth estimation. Preserves 3-D scene structure via diffusers/controlnet-depth-sdxl-1.0.

✏️

Canny Edge

OpenCV Canny edge detection extracts lineart from any image via xinsir/controlnet-canny-sdxl-1.0.

🖊️

Scribble

Otsu binarization turns rough sketches into guidance maps via xinsir/controlnet-scribble-sdxl-1.0.

Access
ZeroGPU tiers

All tools run on NVIDIA H200 ZeroGPU. Free with any Hugging Face account.

Free
Free Tier
~3.5 min GPU time per day
All 4 tools accessible
NVIDIA H200 access
Shared queue — may wait
PRO — Billed by HF
HF PRO Tier
25 min GPU time per day
Priority queue — skip wait
Faster inference
Get PRO at Hugging Face →
Studio
Character ↗
Tools
Get HF PRO
Sign in to unlock
waiNSFW Illustrious · Realmix · ADetailer · Compel
Text to Image
SDXL · Compel · ADetailer
Model Selection
Positive Prompt
Negative Prompt
Resolution
Quick Ratio
CFG 6.0
Steps 25
Sampler
Seed 🎲 Randomize
Output

Generated image
appears here

Sign in to unlock
Image-to-Image · Strength control · ADetailer
Image to Image
Transform any image with a prompt
Model
Input Image
Click or drag image
Upload image to begin
Positive Prompt
Negative Prompt
✨ Enable ADetailer
Face and hand enhancement
CFG 6.0
Steps 25
Strength 0.75
Sampler
Seed 🎲 Randomize
Output

Result appears here

Sign in to unlock
OpenPose · Depth · Canny · Scribble · xinsir SDXL ControlNets
ControlNet
Structural guidance · OpenPose · Depth · Canny · Scribble
Mode & Model
🦴 OpenPose — Body skeleton keypoints · xinsir/controlnet-openpose-sdxl-1.0
Reference Image
Click or drag image
Upload image to begin
Positive Prompt
Negative Prompt
✨ Enable ADetailer
Face and hand enhancement
Conditioning Scale 1.0
0.5–0.8 subtle · 0.8–1.2 balanced · 1.2–2.0 strict
CFG 6
Steps 25
Sampler
Seed 🎲
W override
H override
Control Map Preview

Upload image
then Preview Map

Control map
Final Output

Result appears here

ControlNet output
Sign in to unlock
4-step Lightning · Natural language · Anime2Real LoRA
Qwen Lightning Editor
Natural language · 4-step · Anime2Real
Input Image
Click or drag image to edit
Upload image to begin
Aspect Ratio
Resolution
Edit Instruction
🌸 Anime2Real LoRA
Converts anime to realistic style
Strength 0.85
CFG 1.0
Steps 4
Seed 🎲
Output

Edited image appears here

Sign in to unlock
Real-ESRGAN Anime 4× · Lossless PNG
4K Upscaler
Real-ESRGAN · Anime-optimized · Lossless PNG
Input Image
Click or drag image to upscale
Upload image to begin
Upscale Model
Output Scale 4.0×
Output

Upscaled image appears here

Sign in to unlock
SDXL Inpaint · Regional masked editing
Inpaint
Masked regional image editing
Model
Base Image
Upload base image
Upload base image first
Paint Mask (brush directly on image)
Brush Size 32
Upload base image, then paint white mask areas with brush
Positive Prompt
Negative Prompt
✨ Enable ADetailer
Face and hand enhancement
CFG 6.0
Steps 25
Strength 0.90
Sampler
Seed 🎲 Randomize
W override
H override
Output

Result appears here