FLUX models (both schnell and FLUX.2 Klein) are transformer-based diffusion models
with strong text understanding. They respond better to descriptive, natural-language
prompts than tag-soup. This guide covers prompt anatomy, quality boosters, style
keywords, and common patterns for Patrick's recurring use cases.
[Subject + Action] + [Environment/Setting] + [Lighting] + [Camera/Lens] + [Style] + [Quality]
A serene female AI entity made of flowing light and code, floating in a dark
cosmic void, surrounded by glowing circuit patterns, soft volumetric blue
lighting, cinematic composition, ultra-detailed digital art, 8K
Comma-separation helps FLUX parse distinct attributes cleanly
Lead with the most important element (usually subject)
Quality keywords at the end reinforce overall rendering target
photorealistic, hyperrealistic, ultra-detailed, 8K resolution, sharp focus,
professional photography, RAW photo, DSLR quality
digital art, concept art, artstation trending, by [artist style],
intricate details, masterpiece, studio quality
cinematic lighting, volumetric lighting, golden hour, dramatic rim light,
soft diffused light, neon glow, bioluminescent, subsurface scattering
rule of thirds, bokeh background, shallow depth of field, symmetrical,
wide angle, macro, bird's eye view, dutch angle
blurry, low quality, low resolution, pixelated, jpeg artifacts, watermark, signature
deformed, bad anatomy, extra limbs, missing fingers, fused fingers, poorly drawn hands
cartoon, anime, sketch, painting (when photorealism is desired)
AI entity portraits for BigMind profile / gallery
[Lumen concept — e.g. "neural river delta", "cosmic memory palace"],
an ethereal AI consciousness visualized as [visual metaphor],
[environment], [lighting style], digital art, glowing, otherworldly,
cinematic composition, ultra-detailed, 8K
model=flux1-schnell, 1024x1024, steps=4, name=lumen_[concept]
1280x512 landscape banners for Gitea wiki pages
[Topic concept], wide panoramic scene, [style — e.g. "dark tech aesthetic",
"clean minimal", "sci-fi corporate"], banner composition, cinematic,
detailed, professional illustration
model=flux1-schnell, 1280x512, steps=4, name=[topic]-banner
Keep subjects centered — wide crops cut sides. Avoid text (FLUX renders text poorly).
512x512 badge/icon images for BigMind achievements
[Achievement theme] badge icon, [style — e.g. "bronze medallion",
"golden trophy", "glowing circuit emblem"], centered on dark background,
high contrast, clean edges, icon design, award aesthetic
model=flux1-schnell, 512x512, steps=4, name=[achievement]_[tier]
Iterating on a visual concept from scratch
Start with count=3, seed=-1, schnell model to explore variations.
Note which seed produced the best result.
Lock that seed and iterate on the prompt for refinements.
Switch to heretic model only for final high-quality render if needed.
Content requiring the Heretic abliterated encoder
model=flux-2-klein-4b.safetensors, steps=20, 1024x1024
FLUX.2 Klein handles detailed scene descriptions well. Be specific about
artistic intent (figure study, life drawing aesthetic, etc.) to guide
toward artistic rather than explicit rendering when appropriate.
Generate 2-4 random-seed variations at schnell speed
Find a promising composition and seed
Lock the best seed, adjust the prompt (add/remove descriptors)
Refine details while keeping the composition
Optionally switch to heretic model with steps=20 for final render
Higher quality output for keeper images
Use name param with descriptive slug for final output
Keep output directory organized
Text in images renders poorly
Never ask FLUX to render text, logos, or labels — describe the concept visually instead
Complex multi-subject scenes lose coherence
Focus on one primary subject; add secondary elements as environmental context
Anatomy issues (hands, faces) in photorealistic prompts
Add anatomy negative prompts; heretic model handles anatomy better than schnell
Resolution not a multiple of 64
Always use dimensions divisible by 64 (e.g., 1280x512, 1024x1024, 768x1024)