V5 Curated chalkboard示例 01nai-diffusion-5-curated
NAI Diffusion 5 Curated 用于动漫插画与角色创作,支持图中文字和参考图创作。
更新于:
V5 Curated chalkboard示例 01
Six-armed cosmic being示例 02
Snowfield explorer postcard示例 03
Japanese two-panel comic示例 04从提示词到成图
1girl, cat girl, cat ears, pink hair, short hair, blue eyes, school uniform, holding chalk, chalk, writing, classroom, chalkboard, text, english text. Writing on chalkboard can be seen saying "V5 Curated" along with a heart, right next to the girl's head., location, very aesthetic, masterpiece, no text, teXt: V5 Curated
1girl, blonde hair, short hair, blunt bangs, green scarf, red cardigan, purple eyes, smile
1girl, multiple boys, very aesthetic, best quality, amazing quality, absurdres,, wide shot, medieval, historically accurate, mountainous horizon, wind, spear, An army of soldiers holding spears can be seen standing behind the girl.
year 2025, meta:novel era, drawn, anime character style, happy atmosphere, millipen (medium), painting (medium), papercraft (medium), cinematic lighting, rim lighting, backlighting, dramatic shadows, soft shadows, high contrast, bloom, volumetric lighting, depth of field, very aesthetic, masterpiece, high complexity, depthness, res_mult:2.5x, background is out of this world space A faceless, bald, no ears, ethereal slender being with black galaxy skin with six arms and six hands with five fingers each proudly showing off objects upper left arm raised high displaying a clear glass of water middle left arm extended forward holding a blue ball, lower left arm bent at the elbow holding a gold crown, upper right arm lifted holding an ace of spades card face, forward middle right arm stretched outward holding a orange cube, lower right arm lowered slightly gripping a fresh red apple, , transparent background
1girl, 1other, very aesthetic, best quality, amazing quality, -1::blur, mosaic::, 2::watercolor (medium)::, illustration, -1::realistic::,, light source, the main light source in the image comes from the glowing magic runes on the broadsword, magic, medieval, forest, first person view, dialogue options, text, english text, There is a dialogue box titled "Elf Ranger" next to the girl's head that says "I can't believe a drunken oaf like you was able to wield the Sword of Legends...". On the bottom part of the image are three dialogue options in boxes which say: "(Str 10) It was ez cuz I'm strong af LMAO", "The sword of what?", "What is this place?", teXt: Elf Ranger I can't believe a drunken oaf like you was able to wield the Sword of Legends... (Str 10) It was ez cuz I'm strong af LMAO The sword of what? What is this place?
1girl, blonde hair, short hair, blunt bangs, green scarf, red cardigan, purple eyes, smile, very aesthetic, masterpiece, no text
solo, 1girl, explorer girl, year 2026, very aesthetic, absurdres, best quality, res_mult:3x from side, Snowfield, outdoors, wide shot, solo focus walking, from behind, spread_arms, nature complex background, high complexity huge text, polaroid_photo, handwritten text, Postmark text:Come, see the cliffs and the desert!
2人の女の子、カラー漫画、2コマ、 , very aesthetic, masterpiece, no text, teXt: この画像は全て日本語のプロンプトで書かれています。 V5は日本語も記載ができます。
1girl, blonde hair, short hair, blunt bangs, green scarf, red cardigan, purple eyes, smile, very aesthetic, amazing quality, no text
nai-diffusion-5-curated 能做什么
根据提示词生成
提交简洁提示词,通过一个 TTAPI 任务获取图像结果。
常用参数
设置提示词、模型、输出参数和回调即可接入。
异步查询
通过轮询或 Webhook 接收结果,避免耗时生成阻塞界面。
接入产品
NovelAI 的输出可以展示、保存或交给后续流程处理。
按操作计费,单位为 quota
按模型、操作和输出设置扣除 quota。
请求头与提交参数
使用 TTAPI 密钥,按下方字段向接口发送请求。
/novelai/text-to-image提交 NovelAI 文生图任务。
查看接口文档/novelai/image-to-image用 NovelAI 修改参考图。
查看接口文档/novelai/fetch查询 NovelAI 图像任务结果。
查看接口文档请求头
TT-API-KEYstring必填填写 TTAPI API 密钥。
Content-Typestring必填使用 application/json。
请求体
promptstring必填描述想在插画中出现的内容。
modelenum必填可选 nai-diffusion-5-full、nai-diffusion-5-curated、nai-diffusion-4-5-full、nai-diffusion-4-5-curated,默认 nai-diffusion-5-curated。
ninteger选填生成图像数量,范围 1 至 4,默认 1。
widthinteger选填输出宽度,单位为像素,默认 1024。
heightinteger选填输出高度,单位为像素,默认 1024。
seedstring选填复现或改变生成结果时使用的随机种子。
stepsinteger选填扩散步数,默认 23。
scalenumber选填提示词引导强度,默认 5。
negative_promptstring选填描述要避免的元素或效果。
quality_toggleboolean选填启用文档所述的质量预设,默认 true。
samplerenum选填可选 Euler Ancestral、Euler、DPM++ 2S Ancestral、DPM++ 2M SDE、DPM++ 2M、DPM++ SDE,默认 Euler Ancestral。
noise_scheduleenum选填可选 karras、exponential 或 polyexponential,默认 karras。
character_promptsobject[]选填角色提示词与位置设置。
vibe_transferobject[]选填氛围参考图及影响强度。
character_referenceobject[]选填角色和风格参考设置。
hook_urlstring选填任务完成时调用的回调 URL。
发送请求并获取结果
复制当前模型的请求地址、鉴权请求头和请求体。
curl --request POST \
--url https://api.ttapi.io/novelai/text-to-image \
--header "TT-API-KEY: $TTAPI_KEY" \
--header 'Content-Type: application/json' \
--data '{"prompt":"anime character portrait, detailed eyes","model":"nai-diffusion-5-curated","n":1,"width":1024,"height":1024,"steps":23,"scale":5,"quality_toggle":true,"sampler":"Euler Ancestral","noise_schedule":"karras","hook_url":"https://example.com/webhooks/novelai"}'