Adam Stewart’s Guide to Crafting Effective Prompts for Juggernaut X & SDXL
Update a Sequel to this guide for prompting Juggernaut XI and XII is now available here.
Before we dive into this guide, I would like to extend my gratitude to Kandoo for his exceptional dedication to the development of Juggernaut X NSFW + RunDiffusion Photo and Juggernaut X SAFE + RunDiffusion Photo. A heartfelt thank you is also due to RunDiffusion for their significant contributions to RunDiffusion Photo and their collaborative efforts with Team Juggernaut. Additionally, I must acknowledge the generous support of my private sponsor, whose assistance in navigating the complexities of AI has been invaluable. This sponsor has continually encouraged me to craft community guides, enhancing the collective knowledge and skills of our community. This guide, much like our models, is a service aimed at enriching the art creation experience for all, exemplifying the transformative power of decentralized and open-source AI. If you find it helpful please be sure to follow Team Juggernaut, RunDiffusion and me please.
Introduction
Juggernaut X is tailored for professionals who demand precise control over image generation. This advanced model thrives on detailed, specific prompts, enabling users to meticulously adjust every element to align with their vision. While this guide is written with Juggernaut X in mind the components of a prompt are applicable to other models too.
Audience: Juggernaut X is ideal for artists, designers, and visual professionals who require control over their image outputs.
Prompting Style: To leverage Juggernaut X's capabilities fully, your prompts should be detailed and specific. Unlike its predecessors, Juggernaut 8 or 9, where general prompts might suffice, Juggernaut X requires explicit, impactful trigger words. This precision ensures the AI comprehends and executes the desired elements of your image accurately.
Juggernaut X Versions (As of 4/27/2024)
- Juggernaut X SAFE(Suitable AI for Everyone): RunDiffusion
- Juggernaut X NSFW: Available on CivitAI and Huggingface.
- Juggernaut X NSFW Hyper: Available on CivitAI and Huggingface
Key Trigger Words for Juggernaut X
To optimize image quality and detail, consider incorporating these trigger words into your prompts:
- Skin Textures: Essential for detailed depictions of human or animal skin.
- High Resolution: Ensures images are generated with utmost clarity. This token has three versions: “High Resolution”,"High-Resolution" or "High-Resolution Image."
- Cinematic: Perfect for creating dynamic, narrative-rich images with a film-like quality.
At the end of the Guide is a large list of tokens.
Understanding the Importance of Detailed Prompting
Example Prompt:high-resolution portrait of a Renaissance noblewoman, intricate lace collar, holding an ancient book, dominant colors deep red and gold, style reminiscent of Vermeer's lighting techniques, mood thoughtful, lighting soft, natural window light, perspective close-up, texture rich fabrics and aged paper, cultural elements European Renaissance elegance
Components of a Prompt
Craft each part of your prompt to be direct and descriptive, avoiding unnecessary verbosity:
Subject
- Definition: The primary focus of the image (e.g., person, animal, object).
- Example: "Renaissance noblewoman", "vampire queen".
Action
- Definition: Describes what the subject is doing, adding dynamism or narrative.
- Example: "holding an ancient book".
Environment/Setting
- Definition: The background or scene surrounding the subject.
- Example: "in a dimly lit Gothic castle".
Object
- Definition: Secondary items that enhance the subject or story.
- Example: "wearing an intricate lace collar".
Color
- Definition: Dominant colors or color schemes.
- Example: "shades of deep red and gold", "monochrome palette with stark contrasts".
Style
- Definition: The artistic style or method of rendering.
- Example: "in the style reminiscent of Vermeer's lighting techniques", "emulating a noir film".
Mood/Atmosphere
- Definition: The emotional or atmospheric quality.
- Example: "atmosphere of mystery", "serene mood".
Lighting
- Definition: Specific lighting conditions or effects.
- Example: "bathed in soft, natural window light", "dramatic shadows under a spotlight".
Perspective/Viewpoint
- Definition: The angle or perspective from which the scene is viewed.
- Example: "bird’s eye view", "from a low angle".
Texture/Material
- Definition: Prominent textures or materials visible in the image.
- Example: "textures of rich velvet and rough stone".
Time Period
- Definition: A specific era or historical period.
- Example: "Victorian Era", "futuristic 22nd century".
Cultural Elements
- Definition: Elements that reflect specific cultures or traditions.
- Example: "inspired by Norse mythology", "traditional Japanese setting".
Emotion
- Definition: The expressed emotion if the subject is sentient.
- Example: "expression of deep contemplation", "joyful demeanor".
Medium
- Definition: Specifies the artistic medium or level of detail.
- Example: "resembling a watercolor painting", "crisp digital rendering".
Clothing
- Guidance: Ensure prompts clearly specify subjects are clothed by describing the types and styles of garments, particularly when avoiding inappropriate content.
- Example: "clothed in a Victorian gown", "dressed in modern casual wear".
Text
- Capabilities: Juggernaut has the ability to generate text, though accuracy decreases with sentence length and complexity. Use short, clear phrases enclosed in quotes to improve output.
- Example:A Sign with the text “Peace” .
NSFW Content
- Avoidance: Explicitly include terms to exclude inappropriate content due to the model's training data. Place sensitive terms in the negative prompts section.
- Integration: Adding these terms in a prompt will create NSFW content.
- Sensitive Tokens: "Naked", "Nude", "Vagina", "Penis (big, Long)", "Pussy", "Breasts (Small, Medium, Large)", "Testicles".
Token Bleed and Multiple Subjects
- Challenges: When prompts combine multiple subjects or themes, there may be blending of elements, known as token bleed.
- Solution: Be aware that multiple attempts might be needed to achieve desired separations in images depicting complex interactions like "Angel and Devil". While it may not always be possible to completely prevent token bleed, experimenting with various seeds and the arrangement of tokens in your prompt can enhance the likelihood of success.
Juggernaut X Hyper NSFW
- Challenges: Managing nudity is particularly challenging with the Hyper model as it lacks support for traditional strong negative prompts.
- Solution 1: To address this issue, it's crucial to include specific clothing items in your prompts. Despite these efforts, please be aware that accidental nudity may still occur. It's important to meticulously craft your prompts and remain cautious of this limitation when using the model.
- Solution 2: By setting the CFG options to level 2, you can achieve limited influence using negative prompts. This allows for a slight adjustment in the direction of the content.
- Solution 3: Use the Juggernaut X SAFE version available on RunDiffusion
Juggernaut X NSFW and Juggernaut X SAFE Settings
- Sampling Method: DPM++ 2m Karras
- Sampling Steps: 30-40
- CFG: 6-7
- Resolution: 1024x1024, 832x1216, 1216x832 (Others work but these are my favorites)
- Possible Negative Prompt: Naked, Nude, fake eyes, deformed eyes, bad eyes, cgi, 3D, digital, airbrushed
- Prompt Size: Do not exceed 75 tokens
Juggernaut X NSFW Hyper Settings
- Sampling Method: DPM++ SDE
- Sampling Steps: Recommend 6 but 4-6 possible.
- CFG: Recommend 2 but 1-2 possible.
- Negative Prompt: None
- Prompt Size: Do not exceed 75 Tokens
- Tips: Use more tags with Hyper than the standard model.
Hires. Fix Settings
- Denoising Strength: .25-.32
- Upscaler: 4x_NMKD-Siax_200k
- Hires steps: 0-20
- Upscale by: 1.5 - 2
Conclusion
By strategically incorporating specific trigger words and detailed descriptions into your prompts, you can fully leverage the capabilities of Juggernaut X. This method not only clarifies the best prompting practices but also substantially improves the quality of the generated images, ensuring they align with or surpass your expectations. While Juggernaut X is adaptable to various prompting techniques, combining this detailed prompting method with tagging provides a solid foundation to embark on your creation journey.
Examples Prompts
Golden retriever in a lush green park, fur glistening under the soft afternoon sunlight, colors vivid green and golden, style realistic, mood serene, lighting natural and diffused, perspective close-up, texture soft fur and fresh grass, cultural elements joyful outdoor activity.
Fluffy Maine Coon cat in a sunlit forest, surrounded by vibrant greenery, colors deep greens and rustic browns, style natural habitat, mood peaceful, lighting filtered through trees, perspective ground level, texture thick fur and mossy ground.
Young woman reading a book in a cozy coffee shop, brunette hair cascading over her shoulders, colors warm browns and soft whites, style candid photography, mood relaxed, lighting natural light streaming through a nearby window, perspective over-the-shoulder, texture soft wool sweater and glossy wooden table.
High resolution Portrait of a stylish African woman in urban setting, short brunette hair, bold red lipstick, colors striking red and deep blacks, style modern fashion, mood confident, lighting high contrast with sharp shadows, perspective frontal view, texture leather jacket and smooth skin
Common Token list for Juggernaut X
This is not a complete list but some of the more common tokens in the captioning.
- alloy wheels
- active
- animal portrait
- automotive photography
- beard
- blonde
- blonde hair
- blue sky
- blurred background
- bokeh
- bokeh background
- brunette
- brunette view
- business
- business attire
- candid
- candid photography
- candid shot
- casual clothing
- casual style
- caucasian
- cinematic
- cityscape
- clean lines
- clear sky
- close-up
- cloudy
- confidence
- contemplative
- cultural attire
- curly hair
- daylight
- daytime
- depth of field
- depth of field
- dramatic lighting
- dusk
- dynamic pose
- earrings
- earth tones
- elegance
- facial hair
- fashion
- fashion
- fashion photography
- flexibility
- food photography
- forest
- freckles
- frontal view
- full body
- full body
- gaze
- greenery
- headshot
- high resolution
- high-resolution
- high-resolution image
- indoor
- intense gaze
- interior design
- jewelry
- joyful expression
- leisure
- leisure activity
- long hair
- looking at camera
- looking away
- makeup
- male
- man
- midshot
- minimalistic
- modern
- natural habitat
- natural light
- natural lighting
- natural makeup
- nature
- neutral background
- neutral color palette
- neutral colors
- outdoor
- outdoor setting
- overcast sky
- park
- parked car
- peace
- portrait
- reflection
- reflective
- selective focus
- serene
- serene
- serenity
- serious expression
- serious windows
- setting
- shadow
- sharp details
- sharp focus
- side profile
- side view
- smiling
- soft focus
- standing
- standing pose
- street
- studio
- studio portrait
- studio shot
- stylish
- stylish female
- subtle makeup
- sunglasses
- sunlit
- sunny
- traditional
- travel destination
- trees
- urban setting
- vertical composition
- vibrant colors
- warm color palette
- warm lighting
- warm tones
- white background
- white shirt
- wildlife
- woman
- wooden table
- wristwatch
- young adult
