Stable Diffusion @ CloudDock

Sweet Japanese vibes • zero-stress setup • We’re #1 at the craft, #2 at memes—and no one dares claim #1.

Start creating Read guides

What makes our SD different?

You handle the pretty. We tame dependencies, VRAM, and settings.

learn more →

CloudDock Universal Usagi 4.0 Beta 4 — Overview

Click and draw. If it crashes, it’s not on you.

3060 > 3090 (for most)

Save VRAM first, then talk upscaling.

Training Center — DreamBooth & LoRA on CloudDock

Lock silhouettes, then color—composition wins first.

Quick Start

Queue to canvas in 60s. At 61s, you’re choosing colors.

learn more →

Your First Image in CloudDock

Step-by-step: launch Stable Diffusion in CloudDock.

Prompt Basics (Sweet Anime)

Preset snippets = instant sweet; skip the “mystic” tweaks.

img2img Basics

Use img2img in CloudDock Stable Diffusion to make cleaner versions.

ControlNet Magic

Bodies behave, faces stay free. Layered Pose/Hand/Face out of the box.

learn more →

Pose · Hand · Face

Hands=strong, face=medium—avoid mannequin mode.

Lineart + Depth

Shape first, color later—composition stays clean.

Cleanup & Inpaint

Smudged or off? Mask, nudge—style stays intact.

LoRA Toolkit

Preloaded styles/characters/lighting/color. Sensible weights, no mysticism.

learn more →

LoRA Training — Step-by-Step (Portrait)

Train a portrait LoRA with Kohya on CloudDock.

Dataset & Captioning Guide

Build a clean dataset for Kohya-based Training Center on CloudDock.

Recreate This Look

Reverse-engineer from history and relaunch.

App Store

App Store: One-click Model Pull

learn more →

Model Store — Install Checkpoints & ControlNet Models

Use CloudDock’s Model Store to find Stable Diffusion checkpoints and ControlNet models.

No Fake Models

Hash-verified provenance, no fakes.

Using Your Custom Model in A1111

How to load your own checkpoint or LoRA in CloudDock Stable Diffusion (A1111).

CloudDrive Quick Import

Mount your drive—models/assets on tap. Exports include params.json for instant replay.

learn more →

Mount & Go

Index into the container—no duplicate storage.

Tidy Structure

Clean paths /models & /inputs for fewer mistakes.

Export w/ Params

Image plus params.json = guaranteed replay.

Hires Fix

Hires Fix: Dual-Stage, No-Mush

learn more →

Fixing Blurry Faces & Bad Hands

Three tweaks that instantly clean up faces, hands, and detail.

Stage Two

1.5× upscale + low-noise repair—sharp, not harsh.

VRAM-Friendly

VRAM meter warns: +0.25× risks mush.

Privacy & Safety

Your waifu stays yours. No root shells. Token-gated access. Normal shutdown resets containers.

learn more →

Upload to cloud drive

One-click export to your storage.

Save your style

Keep prompts, LoRAs & models tidy.

Make your anime waifu

References → consistent characters.

Featured Guides

Step-by-step tutorials for SD, ControlNet, prompts and more.

back to top ↑

Launch SD in 60 seconds

A tiny checklist that always works.

DreamBooth Training — Step-by-Step (Full-Body/Mascot)

Train a DreamBooth checkpoint for full-body characters or mascots with Kohya on CloudDock.

Training Center — Tuning & Troubleshooting

Tune Kohya-based training on CloudDock: VRAM strategy ladder.