Model
Trends
.ai
Model
Trends
.ai
nejlepší open source AI modely

Ghibli Diffusion: the ghibli style token and how to use it

Téma: Modely
Autor: Captain
Publikováno 2026-09-30
Nitrosocke's Ghibli Diffusion was one of the first style models for Stable Diffusion. Learn its trigger phrase, settings, how to describe scenes it can draw, and creative ideas for painterly animation-style landscapes.
Tato stránka zatím nebyla přeložena, proto je zobrazena v angličtině.

Přehled

Ghibli
Diffusion by Nitrosocke is a Stable Diffusion 1.x fine-tune trained on stills in the style of the Japanese animation studio. It is triggered by the phrase ghibli style and excels at lush landscapes, cozy interiors, skies and gentle characters.
It knows the look, not the characters: there is no training on specific names, so describe appearances and settings. It is a 512-pixel model, so upscale with an anime-friendly
upscaler
for finals.

Reference

Název
Typ
Co to je
Trigger
ghibli style
Put it at the start of the prompt.
Prompt
text + tags
Short descriptive phrases work well; describe the scene, time of day and weather.
CFG
7
Standard
SD 1.5
value.
Steps
25-30
Euler a or DPM++ 2M Karras.
Resolution
512x512 to 512x768
Upscale afterward; do not render large directly.
AD
Bez instalace - spouštějte AI modely v cloudu
BitVector Prism je nejsnazší způsob, jak začít: vyberte model, napište prompt a generujte v přehledné webové aplikaci, bez jakéhokoli nastavování. BitVector je k dispozici také na Discordu a na webu (SpyGlass).

Krok za krokem

  1. Begin with ghibli style, then the scene in one or two phrases.
  2. Add time of day and weather words: golden hour, after the rain, summer haze.
  3. Keep the
    negative prompt
    simple: lowres, blurry, text, watermark, bad anatomy.
  4. Generate at 512x768 for vertical scenes or 768x512 for landscapes.
  5. Upscale 2x with an anime upscaler and a light
    img2img
    pass at 0.35 denoise.

Příklady

Landscape

ghibli style, a grassy hill with a lone tree, a girl in a yellow dress flying a kite, cumulus clouds, summer afternoon, distant sea

Interior

ghibli style, cozy bakery kitchen at dawn, bread in a wood oven, copper pots, flour dust in sunlight, an old baker smiling

Creative idea: sky harbor

ghibli style, a harbor town built on clouds, propeller boats docked at wooden piers, lanterns, sunset, children waving

Tipy

  • Landscapes with one small figure are its signature; describe the figure briefly.
  • Flying machines, trains and seaside towns produce especially convincing results.
  • Interiors want warm light and clutter: kitchen, pots, hanging herbs, open window.
  • For characters, describe hair, clothing and expression; avoid celebrity or franchise names.

Řešení problémů

Faces look distorted

Proč k tomu dochází
512-pixel base and small faces in wide shots.

Jak to opravit
Make the figure small and distant, or inpaint the face at a higher resolution after upscaling.

Output looks like generic SD 1.5

Proč k tomu dochází
Trigger phrase missing or buried late in the prompt.

Jak to opravit
Put ghibli style first and raise
CFG
to 8.

AD
PirateDiffusion
Bez instalace - spouštějte AI modely v cloudu
PirateDiffusion je pouze na Telegramu a je stavěný pro profesionály: tisíce modelů, LoRA a workflow ovládaných chatovými příkazy, s neomezeným generováním za pevnou cenu.

Otázky

Can it draw specific film characters?

No. It learned the visual style only; describe the character's look instead.

Does it work with ControlNet?

Yes, as an SD 1.5 model it works with all SD 1.5 ControlNets.

Odkazy a zdroje

Modely v tomto návodu

Napsal
Captain

Související návody

Modely
DreamShaper 8 Inpainting: fixing and changing parts of an image
Lykon's DreamShaper 8 inpainting checkpoint replaces masked regions cleanly in a versatile semi-realistic style. Learn mask strategy, denoise values, prompting for the masked area, and ideas for creative edits.
Autor: Captain
2026-10-01
Modely
Waifu Diffusion: Danbooru-tag prompting for a classic anime model
Waifu Diffusion is the anime base model that taught a generation how to prompt with Danbooru tags. This guide covers tag order, quality tags, negatives, the right resolution and ideas for clean anime illustrations.
Autor: Captain
2026-10-01
Modely
Stable Diffusion v1.1: the original release, used creatively
Stable Diffusion v1.1 is the first public Stable Diffusion checkpoint, trained at 256 then 512 pixels. This guide shows what it is good for today: dreamlike textures, abstract compositions and deliberately rough, early-AI aesthetics.
Autor: Captain
2026-09-29
Modely
Realistic Vision 5.1 Inpainting: photoreal repairs and edits
The inpainting build of Realistic Vision 5.1 (with baked VAE) fills masked regions with convincing photographic detail. Learn mask and denoise strategy, the RAW photo prompt style, and ideas for realistic edits.
Autor: Captain
2026-10-02

Další návody

Modely
FLUX.1 Dev: how to prompt it, best settings and creative ideas
A practical guide to FLUX.1 Dev by Black Forest Labs: natural-language prompting, guidance and step settings, LoRA stacking, text rendering and prompt ideas that play to its strengths.
Autor: Captain
2026-09-30
Modely
Stable Diffusion XL 1.0: prompts, settings and what it still does best
How to get the most out of the official SDXL 1.0 base model: native resolutions, CFG and sampler settings, the refiner, negative prompts and ideas for styles where SDXL still shines.
Autor: Captain
2026-10-01
Modely
Stable Diffusion 1.5: the classic model, prompted properly
Stable Diffusion 1.5 is still worth knowing: the right resolution, CFG and sampler, how to use its huge library of LoRAs and embeddings, and creative prompt ideas suited to a 512-pixel model.
Autor: Captain
2026-10-02
Modely
FLUX.2 Dev: prompting the newest Black Forest Labs model
FLUX.2 Dev brings larger prompts, better text, stronger realism and multi-reference editing. This guide covers the turbo workflow, guidance and resolution settings, prompt structure and ideas that use its new strengths.
Autor: Captain
2026-10-03
Modely
Qwen-Image: long prompts, perfect text and bilingual posters
Qwen-Image is the model to reach for when the words in the picture matter. Learn how to write its long descriptive prompts, render English and Chinese text accurately, set CFG and steps, and explore layout-heavy creative ideas.
Autor: Captain
2026-09-29
Modely
SDXL-Lightning: four-step generation on any SDXL checkpoint
ByteDance's SDXL-Lightning LoRA turns a 30-step SDXL render into a 4-step one. Learn the step counts, the CFG you must use, the sampler, and how to combine it with your favorite checkpoints and style LoRAs.
Autor: Captain
2026-09-30