Before starting

This website contain ads provided by an adnetwork, it means that redirections and misleading stuff will appear outside of my control. This exists because is a way to fund the website and keep it running for free. I would appreciate if you let ads run here.

If you want to use the website without ads, you can become a supporter, it's just 5$/month. If you don't want to pay, you can still use an adblock, but this will prevent some tools and options to be used.

No ads can harm you or your device, just don't download and execute stuff on your pc/phone, neither write your personal data anywhere.

AI Hentai Generation Tutorial

Welcome to the Hentai Generator tutorial! This guide will walk you through creating AI-generated hentai artwork with powerful tools. Basic generation is free and unlimited, with no account, tokens or credits needed. Follow these steps to bring your fantasies to life.

Writing Effective Prompts

The key to great AI art is in your prompt. Here's how to craft the perfect description:

Prompt Structure

Start with the main subject, then add details in order of importance:

1. Character description (gender, hair, eyes, etc.)
2. Clothing or lack thereof
3. Pose and expression
4. Background and setting
5. Special effects or style

Example Prompts

An Advice

The AI can't do everything on its own, so some things might not work as expected. Don't give up, if you're still struggling, ask on Discord how to get closer to what you want. If it's completely impossible, I can try to add it for you.

To check more prompt examples or if you need help, join the Discord Server

Tag Examples

These are common tags but it's not limited to them:

Image Settings

Size Selection

Choose the dimensions for your artwork. Common sizes:

• 512x512 - Standard square
• 768x512 - Wide format
• 512x768 - Portrait format

I recommend you to use sizes lower than 768 with "Upscale" or "Improve" activated, it have less deformities and better quality.

Style Selection

Choose an art style for your generation by clicking on it:

The chosen model affects the look of everything you generate, including image-to-image, sketches, masks and the other tools.

Tip: click and hold a style button to add it to your prompt as a special tag. You can mix several styles at once and add weights to control how strong each one is (e.g. plus:SuperMixEtheris:0.8).

Sharpness Slider

Control how sharp and correct you want the image:

Fidelity Slider

Control how closely the AI follows your prompt:

Detail Slider

Control how many details you want in the image:

Mix Slider

Control how strong you want the style:

The default values are fine, but I recomend to test different values with different prompts and styles.

Advanced Features

Negative Prompt

You can control the AI to prevent something from appearing by using the negative prompt. For example, you need the image to be just a landscape, but it keeps getting you a character, you can write: "person, human, character", etc on negative to get rid of it.

Weight

To increase weights you can use parentheses (tag)

The more parenthesis you use, more emphasis a tag it will have. (((tag))) = 3x stronger

You can use numbers as well (tag:2.3) = 2.3x stronger

Obs: Close every one that you open or it will mess all your prompt; If you exagerate your weights the images will be distorted; If you dont want something to appear write it on negative, it works with weights as well; If something doesn't appear even after a strong weight, check if the tag is written correct, or maybe the AI can't do it.

Nobias

Adding the tag "Nobias" to your prompt will remove hentai bias from the generations, allowing you to create different things, like landscapes, sfw characters, objects, concepts, etc.

Special Tags and Special Characters

Most concepts, styles and characters should work across the different styles just by describing, for example:

"Tsunade, Tsunade from naruto, large breasts, blonde hair, brown eyes, green coat"

But sometimes the AI doesn't know these. That's why there is the Special Tags and Characters.

These are focused tags to guarantee that something will appear in the image. Click on it to show the list of available tags, hover your mouse over it to see a example on the right.

They work better for Supporters, for the main page it is limited and sometimes won't work together properly.

Inventing plus: or character: tags won't work, if it is not on the list, you can try to describe the character or concept using it's name and traits with weights, like the Tsunade example. If nothing of it works, and there is no special tag for it, you can request on Discord.

Image Reference

Upload an image or provide a URL to guide the AI generation, THE LINK NEEDS TO BE A PUBLIC AND AVAILABLE IMAGE OR ELSE THE GENERATOR WON'T READ IT:

Or drag and drop an image file to upload, (uploading of questionable material may lead you to a restriction on this feature):

There are four reference modes in the Reference Tools menu: Image, Sketch, Mask and Sketch+Mask. Whatever you pick, always write a prompt, since the AI won't read all the context from the reference on its own.

Sketch Free

Sketch something and describe how it should look:

Obs: Make it with COLOR, reminding what you want, like a frame, just a scribble will be translated to a black and white scribble image (stick figures stay stick figures, so draw colored shapes representing what you want).

Mask Supporter

Draw over the part that you want to change; only the covered area is regenerated, the rest stays intact:

Write a prompt for it, like the example above:

(((no skirt, red panties, ass, big ass, naked ass, thong, naked))) /// Negative: (((skirt, green skirt, clothes)))

Sketch+Mask Supporter

A mix of both: the AI changes only the part you drew over, but it also takes into account what you drew there together with your prompt. Great for adding a new object or detail exactly where you want it.

Mask² & Sketch+Mask² Advanced Controls Supporter

On the Zenith model the mask tools get three extra sliders (each with an ON/OFF switch; turn it off to use the default behavior) for much cleaner, more controlled edits:

Tip: pair a high Mask Sustain with a low Mask Warp for delicate fixes, or the opposite for a full repaint.

Freedom Slider

Control how closely the AI follows your reference image. This controls Image, sketches and masks:

Example of levels of freedom. (Scroll to the right)

Prompt by Region

Write a prompt for each part individually:

Tip: If you want multiple characters, like 2 girls, write "2girls" on every prompt to enhance it. For best results, pair it with a reference image that already has the right colors in each place.

On Zenith the region tool works differently: instead of a grid, you paint areas of the canvas with different colors and write a separate prompt for each color. Also describe where things are in each prompt ("girl on the left", "tree in the background") and try not to overlap colors, since overlapping colors mix the prompts. A Weight slider biases how strongly the image leans toward a region. Note that Zenith's region tool is softer, since it prioritizes following the prompt over the exact painted boundaries.

Fix Face

Detects faces in an image and re-draws them to look better. Very useful for full-body or distant shots where the face often comes out distorted. You can also write a small prompt to guide how the face should look (e.g. "pretty face, blue eyes, blushing").

• Freedom controls how much the new face may differ from the original: low keeps it close, high lets the AI redraw it freely.
• The AI can't recognize faces that are too blurred or far away (there's a minimum quality), and it can occasionally have pareidolia (see faces where there aren't any).

3D Pose Supporter

The old pose maker was replaced by a full 3D Pose tool (available on the Zenith model). Open it and pose a 3D character in a draggable floating window, then use it as a pose reference for your generation:

• Drag bones to pose the body, or use the IK handles: drag the hand or foot marker and the whole limb follows naturally.
• Mirror button copies a whole limb from one side to the other, great for symmetric poses.
• Ready-made hand gestures (fist, open hand, thumb out…) and full-body poses you can apply in one click and then tweak.
• The pose window is locked to the aspect ratio of the size you picked, so what you frame is what you generate.
• Pose Strength slider controls how strictly the AI follows the pose.
• Skeleton mode shows an OpenPose skeleton over the character; turn it on to send the skeleton (instead of the rendered figure) as the pose guide.

Pose your character, then generate; the pose is used as a reference alongside your prompt.

Some considerations:

The image ASPECT RATIO will follow the size that you choose not the image that you are using, if your image is 1:2 and you choose a size for 512x512 = 1:1, the generated image will squish the uploaded image to generate a new one.

Make sure that the image that you are linking is available for everyone to see, or else the generator won't be able to see it too.

Zenith: Natural Language & Smart Tags Supporter

Zenith is our newest and most capable model. Pick it from the model menu (it's unlocked for supporters). Everything you already learned still works, but Zenith adds a few big upgrades:

Natural Language Prompts

Zenith understands full sentences, not just tags. You can literally write "a shy girl with long silver hair sitting by a window while it rains outside, soft warm light" and it will understand the scene. You can still mix in tags and weights whenever you want more control, and the two styles work together.

Smart Auto-Complete: Artists, Characters & Shows

As you type, Zenith suggests matching artists, characters and shows/series from a huge database, so you don't have to guess the exact spelling:

• Start typing a show or character name and pick it from the list; it gets inserted in the correct, AI-friendly format automatically.
• Recognized tags glow in the prompt: characters, series and artists each get their own color so you can see at a glance what the AI understood.
• A toggle button at the top-right of the prompt box turns the suggestion lists on or off (your choice is remembered). The glows stay on either way.

This makes it far easier to nail a specific character or copy an artist's style without hunting for the right tag by hand.

Zen Edit Supporter

Zen Edit lets you edit images by describing the change in plain language. Upload 1 to 3 images and tell the AI what to do with them.

Example:

• Image 1: a character
• Image 2: a piece of clothing
• Prompt: "the character from image 1 wearing the clothes from image 2"

Video Generation Supporter

Animate an image into a short video with a prompt. Select the Video tile and the interface simplifies to just a start-image slot, a prompt and a negative prompt.

How to use it

• Start from an image: you must upload or generate an image first. The video keeps that image's aspect ratio (at a small size / low resolution, since this server runs on modest hardware).
• Prompt like a director: describe the action. What are the hands doing? What is being exposed or covered? How hard is something gripped? The more specific the motion, the better the result.
• Length: each generation makes a ~3 second clip. To reach ~6 s, after it finishes open the options and click "Continue from last frame"; the last frame loads into the image slot and the next clip continues seamlessly. Change the prompt before continuing to shift the action mid-video.

Limits & cooldown

• Supporters: up to 2 videos every 25–30 minutes.
• Supporters ×3: up to 4 videos per cooldown window.
• A countdown timer on the video tile shows when you can generate again.

Generation Process

After clicking GENERATE, you'll enter a queue:

• Free users: 3 concurrent generations max
• Supporters: 8 concurrent generations max
• Supporters have priority in the queue

Typical generation time: 20s-3minutes depending on server load.

After Generating: Result Tools

Once an image is ready, a row of tools appears so you can save it or keep working on it without starting over. (Each button also has its own ⓘ help icon.)

Frequently Asked Questions

Support Hentai Generator

Your support helps keep this site running and improves it for everyone!