A good prompt will allow you to create magnificent generated images.
You can find prompt examples here.
Welcome to the Yiff AI Guide! Whether you want to bring your fursona to life, explore your wildest creative fantasies, or simply create stunning furry art, this guide will walk you through step by step.
Ready to unleash your imagination? Let's get wild! 🔥
💡 Tip for your prompt: You don't need to add quality keywords like "masterpiece, 4K, details,..." or in the negative "ugly, dismorphed,..." - the system already includes these improvements automatically.
Each word carries importance and power, which is why we advise you to test their impact gradually.
In this documentation, we will explain each parameter to help you achieve your desired results.
Parameters
Model Selection
Yiff AI offers several generation models. Here is our recommendation:
🥇 Yiff AI Advanced (Recommended)
Our best model, updated regularly. It offers the best results with minimal effort and faithfully reproduces the style of your favorite artists.
🥈 BB95 Furry
An excellent model equivalent to Furry Fandom quality. Generates beautiful images without requiring complex prompts.
🥉 Yiff AI Basic
Our very first model, still available. Requires more detailed and solid prompts to avoid blurry or distorted results. Reserved for experienced users.
💡 Tip: If you're a beginner, start with Yiff AI Advanced for optimal results without any hassle.
Positive Prompt
Here, you should specify what you wish to see appear. Build your prompt with these elements in this recommended order:
- Artist style - by [artist name] (e.g.: by Chunie)
- Number of characters - solo, duo, trio, group
- Character type - anthro (anthropomorphic) or feral (four-legged animal)
- Species and gender - male wolf, female fox, etc.
- Physical characteristics - fur color, musculature, etc.
- Pose or action - sitting, lying, bent over, etc.
- Background - forest background, bedroom, etc.
- Rating - safe (SFW) or explicit (NSFW)
Example of a structured prompt:
by Chunie, solo, anthro, male wolf, grey fur, muscular, sitting, forest background, explicit
Autocompletion
💡 Use the autocompletion built into the generator!
When you type in the prompt field (positive or negative), tag suggestions appear automatically. This feature helps you:
- Discover tags recognized by our models
- Avoid typos or made-up words
- Find variations you might not have thought of
⚠️ Important: A prompt made of tags (keywords separated by commas) works better than a complete sentence.
✅ Good: male wolf, blue eyes, muscular, sitting, forest
❌ Less effective: A muscular male wolf with blue eyes sitting in a forest
Negative Prompt
Elements you do not wish to appear if the generation includes them in previous results.
You don't need to add "ugly" tags like: bad anatomy, bad hands, missing head, etc. - the system already handles these aspects.
Main use: Specify the opposite gender or unwanted elements.
Example: If you generate a male character and a female appears in the result, add female to the negative prompt then regenerate.
Image Size
The size of the image might seem trivial but it is essential for creating quality images according to your character and their pose.
The aspect ratio is as important as the resolution. For example, choosing a landscape format for a standing anthro character might result in them not fitting entirely within the frame.
Recommended formats:
- Standing anthro - Portrait (vertical)
- Lying/bent anthro - Landscape (horizontal)
- Feral - Landscape (horizontal)
- Scene with multiple characters - Landscape (horizontal)
💡 Tip: Test different formats if your character has a particular pose (bent, in motion, etc.).
Image Reference
🖼️ New feature!
Image Reference allows you to paste an image URL to guide the generation. This feature is ideal for:
- Reproducing a specific pose - The AI will draw inspiration from the character's position
- Keeping a character design - Maintain the visual characteristics of an existing character
- Starting from a sketch or rough drawing - Even a rough drawing with thick lines can serve as a base! The engine will take into account the main lines of your sketch to understand the pose and scene composition
⚠️ Important:
- Image Reference does not copy the artistic style - use the by [artist] tag for that
- The text prompt remains mandatory to describe what you want
- Works by image URL only
Example of use:
You have an image with a pose you love → paste its URL → add your prompt describing your character → the AI generates your character in that pose.
BREAK Control
💡 BREAK helps you separate different parts of your prompt to prevent them from mixing with each other.
How does it work?
The generator reads your prompt in blocks. Without BREAK, it can confuse attributes between different elements (for example: you ask for a wolf with blue eyes and an orange background, and you get a wolf with orange eyes).
The BREAK keyword creates a clear separation between the elements of your prompt.
⚠️ Important: BREAK must always be written in UPPERCASE to work.
Simple example (avoid color mixing):
male wolf, blue eyes, grey fur, red scarf
BREAK
female fox, green eyes, orange fur, blue dress
→ Blue eyes stay on the wolf, green eyes stay on the fox!
Example with multiple characters (duo):
For scenes with multiple characters, repeat the common elements (artist, number of characters) in each section:
by Chunie, duo, male wolf, muscular, brown fur, standing
BREAK
by Chunie, duo, female fox, slim, orange fur, sitting
Advanced example (duo + action + background):
by Chunie, duo, male wolf, muscular, brown hair, passionate embrace
BREAK
by Chunie, duo, female fox, blond hair, long hair, spread legs
BREAK
forest background, sunset, detailed lighting, warm colors
When to use BREAK?
- A single simple character - ❌ Not necessary
- A character + detailed background - ❌ Not necessary
- Two or more characters - ✅ Highly recommended
- Multiple specific colors not to mix - ✅ Highly recommended
Emphasis / De-emphasis
⚠️ This parameter should be used with care to avoid distorted details: increase it gradually.
If you wish to strengthen or weaken the importance of a term in your prompt, you can do so on a scale between 0.1 and 2 (e.g.: keyword:1.5). Words have a default value of 1.
Examples:
- feet:1.5 → more prominent feet
- feet:0.8 → less prominent feet
Emphasis scale:
- 0.1 - 0.5 - The word is almost ignored
- 0.6 - 0.9 - Reduced presence
- 1.0 - Default value
- 1.1 - 1.3 - Slightly reinforced ✅ Recommended
- 1.4 - 1.6 - Strongly reinforced ⚠️ Risk of distortion
- 1.7 - 2.0 - Very risky ⚠️ Often distorted
💡 Tip: Always start with values close to 1 and increase gradually: feet:1.2 → feet:1.4 → feet:1.6...
Upscaler
✨ The Upscaler allows you to enlarge your image (x2) with more details.
This feature is applied during generation. You directly receive the upscaled image - there is no "before/after" version, only the enhanced final result.
Advantages:
- Larger and more detailed image
- Better quality for printing or wallpapers
- Improves textures (fur, eyes, etc.)
⚠️ Warning: The Upscaler can sometimes alter the requested artist's style. If artistic rendering is important to you, test without Upscaler first.
💡 Tip: The Upscaler is ideal for final images you want to keep in high quality.
Guideline Keywords
Here are popular keywords you can use in generation to help you achieve various styles or poses. You can combine multiple if it makes sense.
💡 Tip: Use the autocompletion built into the generator to discover even more tags while you type!
Artists (non-exhaustive list)
by Angellsview3, by Bebebebebe, by Braeburned, by Chunie, by Dimwitdog, by Doxy, by F-r95, by Fluff-Kevlar, by Funkybun, by Hioshiru, by Jay Naylor, by Jasonafex, by Jishinu, by Kabier, by Kalahari, by Kluclew, by Kuroodod, by Meesh, by Mercurial64, by Raaz, by Reallynxgirl, by Redrusker, by Roanoak, by Ruaidri, by Sabuky, by Seibear, by Skweekers, by Smileeeeeee, by Tsampikos, by Zackary911, by Zaush, by Zica, ...
💡 Tip: If the artist's style doesn't appear clearly in your results, it's probably because that artist has a subtle style or few referenced posts. In that case, use emphasis on the artist tag: (by artistname:1.3) or (by artistname:1.5) to strengthen their influence.
Scene Type / Characteristics
solo, duo, trio, group, male, female, andromorph, futanari, anthro, feral, anthro on feral, feral on feral, [color] fur, [color] penis, [color] pussy, [color] butt, clothed, naked, safe, explicit
Species / Animals
Canines (popular)
wolf, dog, fox, husky, german shepherd, shiba inu, coyote, jackal, fennec
Felines
cat, lion, tiger, leopard, snow leopard, panther, cheetah, lynx, cougar
Other mammals
rabbit, bunny, horse, deer, bear, raccoon, otter, skunk, badger, red panda, hyena, bat
Reptiles & Dragons
dragon, kobold, lizard, snake, crocodile, gecko, sergal
Birds
eagle, owl, raven, crow, parrot, phoenix, gryphon
Mythical creatures
dragon, gryphon, phoenix, unicorn, werewolf, kitsune
Pokémon / Digimon
lucario, eevee, umbreon, vaporeon, pikachu, charizard, renamon, lopunny
Original furry species
protogen, sergal, dutch angel dragon
💡 Tip: Also specify the animal family for better results. Example: wolf, canine, canid
Poses
Basic poses (SFW)
standing, sitting, lying, kneeling, crouching, leaning, stretching, walking, running, jumping, flying, waving, pointing, arms crossed, hands on hips, looking back, looking at viewer
Sensual / suggestive poses
lying on back, lying on side, lying on stomach, spread legs, bent over, on all fours, presenting, arched back
Explicit NSFW poses
presenting pussy, presenting penis, presenting anus, spread pussy, spread anus, masturbating, pawing off
Sexual actions
blowjob, fellatio, licking, kissing, anal penetration, vaginal penetration, riding, cowgirl, reverse cowgirl, doggystyle, missionary, spooning, 69 position, knotting
Duo/group interactions
hugging, cuddling, mounting, pinned down, dominance, submission
Style / Camera Angle
front view, back view, side view, three-quarter view, hindquarters view, bottom view, top view, worm's-eye view, dutch angle, close up, portrait, full body
If you need more help or support for your prompt, you can join our community server on Discord here (#yiff-ai).