★ steely dan × liminal hotel room × 35mm film ★ brutalist architecture sunset vaporwave ★ 1970s rock album × medium format ★ renaissance cyberpunk samurai ★ macro honey gold leaf ★ tokyo aerial rain cinematic ★ surrealist collage editorial ★ analog grain portrait studio ★ neon botanical illustration ★   ★ steely dan × liminal hotel room × 35mm film ★ brutalist architecture sunset vaporwave ★ 1970s rock album × medium format ★ renaissance cyberpunk samurai ★ macro honey gold leaf ★ tokyo aerial rain cinematic ★ surrealist collage editorial ★ analog grain portrait studio ★ neon botanical illustration ★   
STEAL THIS PROMPT

SIX MODELS.
ONE LIBRARY.

Every model has its own dialect. We speak all of them.

Flux 1.1 Pro Midjourney v7 Nano Banana Pro SD3.5 Large Seedance 2 Veo 3.1
01

Flux 1.1 Pro

The photorealism king.

Flux 1.1 Pro is Black Forest Labs' flagship. It's the model you use when the brief says 'make it look real.' Exceptional at natural lighting, skin textures, and complex scenes with multiple subjects. The prompt dialect is clean and descriptive – no special tokens needed.

Quirks & tips

  • Loves descriptive adjectives – "warm amber dusk light" beats "golden hour"
  • Handles long prompts well (150+ words) without breaking
  • Negative prompts optional but useful for denoising artifacts
  • Aspect ratio: native 1:1, but 16:9 and 3:4 work beautifully
Best for: PhotorealismPortraitsProduct shotsEditorial
Browse Flux 1.1 Pro prompts
Sample output from Flux 1.1 Pro
Flux 1.1 Pro / SAMPLE OUTPUT
[ TOP PROMPT FOR FLUX 1.1 PRO ] Loves descriptive adjectives – "warm amber dusk light" beats "golden hour"
02

Midjourney v7

Still the dreamiest.

Midjourney v7 generates images with a distinct painterly quality that no other model has fully replicated. Cinematic, slightly surreal, compositionally confident. The model has its own aesthetic fingerprint – lean into it.

Quirks & tips

  • Style references work incredibly well: "in the style of Gregory Crewdson"
  • Use :: to weight concepts: "fog::2 architecture::1"
  • --ar flag for aspect ratio (e.g. --ar 16:9)
  • Avoid hyper-literal prompts – MJ interprets, it does not copy
Best for: Cinematic scenesSurrealismLandscapeAlbum art
Browse Midjourney v7 prompts
Sample output from Midjourney v7
Midjourney v7 / SAMPLE OUTPUT
[ TOP PROMPT FOR MIDJOURNEY V7 ] Style references work incredibly well: "in the style of Gregory Crewdson"
03

Nano Banana Pro

Character consistency that actually works.

Nano Banana Pro is the model for anything involving consistent characters across multiple shots. It's a Google-backed architecture and it shows – exceptional instruction following, consistent facial features, strong typography rendering.

Quirks & tips

  • JSON prompt format preferred (separate style, subject, setting, mood)
  • Best-in-class for typography in images
  • Character reference images dramatically improve consistency
  • Does not require negative prompts – positive-only workflow
Best for: Character designTypographyBrand shootsStoryboarding
Browse Nano Banana Pro prompts
Sample output from Nano Banana Pro
Nano Banana Pro / SAMPLE OUTPUT
[ TOP PROMPT FOR NANO BANANA PRO ] JSON prompt format preferred (separate style, subject, setting, mood)
04

SD3.5 Large

Open source and actually good now.

Stable Diffusion 3.5 Large finally delivers on the promise of open-source image generation. Run it locally, fine-tune it, merge it. The prompt dialect evolved significantly from SD1.5 – shorter prompts with clear subject-first structure work best.

Quirks & tips

  • Subject first, always: "a woman in a red dress" not "red dress woman"
  • Use quality boosters sparingly: "8K, hyperdetailed" can over-sharpen
  • LoRA fine-tunes are its superpower – check CivitAI for style packs
  • Handles inpainting better than any closed model
Best for: IllustrationFantasyFine-tune workflowsInpainting
Browse SD3.5 Large prompts
Sample output from SD3.5 Large
SD3.5 Large / SAMPLE OUTPUT
[ TOP PROMPT FOR SD3.5 LARGE ] Subject first, always: "a woman in a red dress" not "red dress woman"
05

Seedance 2

Video prompting without the motion blur mess.

Seedance 2 from ByteDance is the most controllable video generation model we have tested. Camera movements follow instructions reliably, motion is smooth without that AI-blur quality, and 5-10 second clips hold together narratively.

Quirks & tips

  • Include camera instruction in every prompt: "slow dolly left", "static shot"
  • Keep subject count to 1-2 for clean motion
  • Reference frame helps enormously for style consistency
  • Sound sync is not native – plan for post-processing
Best for: Short video loopsProduct videoMotion graphicsSocial content
Browse Seedance 2 prompts
Sample output from Seedance 2
Seedance 2 / SAMPLE OUTPUT
[ TOP PROMPT FOR SEEDANCE 2 ] Include camera instruction in every prompt: "slow dolly left", "static shot"
06

Veo 3.1

Google's cinematic video model.

Veo 3.1 is Google DeepMind's video generation model and it produces genuinely cinematic footage. Lighting is exceptional, human movement is natural, and the model handles complex scene transitions. The prompt format borrows from screenwriting.

Quirks & tips

  • Write prompts like a shot list: "EXT. LISBON STREET - DUSK. A woman walks toward camera."
  • Lens specs improve output: "shot on anamorphic 35mm, 1.85:1 ratio"
  • Environmental details matter: time of day, weather, season
  • Longer prompts (50-100 words) outperform short ones
Best for: Cinematic footageHuman subjectsBrand filmsScene transitions
Browse Veo 3.1 prompts
Sample output from Veo 3.1
Veo 3.1 / SAMPLE OUTPUT
[ TOP PROMPT FOR VEO 3.1 ] Write prompts like a shot list: "EXT. LISBON STREET - DUSK. A woman walks toward camera."
✦ copy the prompt ✦ remix this ✦ drop into flux ✦ steal this look ✦ open the moodboard ✦ crack it open ✦ send to nano banana ✦ go wild ✦ copy the prompt ✦ remix this ✦ drop into flux ✦ steal this look ✦ open the moodboard ✦ crack it open ✦ send to nano banana ✦ go wild ✦