CyaniModel
创建属于同一故事的图片
探索图片生成、参考驱动的角色设计和漫画转视频规划。看看身份、场景连贯性和细致审核如何把单张图片串成完整故事。
模型
这个创作工具生成的是静态图像。漫画和视频请继续使用链接中的工具;研究策略并不是可选的生产模式。
值得围绕它讲述故事的画面

使用 Vertex AI / Nano Banana 2.1 创建的原创作品。这些是示意静帧,不是 CyaniModel 的输出,也不是完成的动画。
图片记录
动漫电影感画面:一名系着深红色围巾的信使在蓝调时刻的海港上方与灯塔看守人相遇
Google Vertex AI · gemini-nano-banana-2.1 · · 1800 × 1005 px
原始生成提示词
Create one extraordinary original hand-painted 2D anime FEATURE-FILM FRAME, widescreen 16:9, not promotional key art. A quiet, emotionally charged encounter in a fantastical coastal city built on weathered sea cliffs and arched bridges: an ADULT woman courier, age about 29, dark short windswept hair, charcoal travel coat and a single crimson scarf, has arrived at an old lighthouse in the last minute of blue hour. An elderly lighthouse keeper in a faded mustard coat holds the door open for her. The two people are modest in scale, both fully visible, separated by several steps; their posture and eye line imply a long-awaited message, with no lettering. Foreground weathered dark stone and a moored narrow boat lead diagonally into the frame. Beyond them, layered luminous harbor windows, elegant tram cables, soft ocean haze and distant low islands form a deeply observed, believable world. The lighthouse lamp has just kindled: one warm amber pool of light crosses the cool indigo slate walkway and catches the crimson scarf. Sophisticated restrained color script of indigo, desaturated sea-green, smoky violet, small amber highlights; nuanced hand-painted clouds, natural atmospheric perspective, delicate cel linework, rich watercolor-gouache background brushwork, controlled texture, dramatic but gentle chiaroscuro, coherent grounded perspective. The composition feels like a held story moment in an expensive auteur animated film, with confident negative space and fine environmental storytelling. Anatomy and hands natural, clothing folds intentional, original character and architecture designs. NO imitation of a named artist or franchise. Entire image is the finished artwork edge to edge. NO photograph, desk, paper mockup, device, human hands holding artwork, book, frame, contact sheet, collage, UI, logos, typography, subtitles, speech bubbles, watermark, generic glowing particles, glossy 3D or game render. Output a native detailed 4K image.
查看完整作品动漫特写:一位穿海军蓝雨衣的成年女性站在雨水打湿的铁路月台上
Google Vertex AI · gemini-nano-banana-2.1 · · 1800 × 2683 px
原始生成提示词
Create one original, superbly art-directed 2D anime cinema character close-up in portrait 2:3. ADULT East Asian woman, about 32, a rain-soaked investigative reporter with a practical dark bob tucked behind one ear, tired alert dark eyes, small healed eyebrow scar, plain navy raincoat over a cream shirt. She has just recognized someone across an empty elevated railway platform; her expression is subtle surprise and resolve, lips slightly apart, not smiling for a camera. Frame from upper chest to just above her head in three-quarter view; leave delicate breathing room around hair silhouette and keep both eyes readable. Rain tracks on her cheek, damp hair strands, translucent water at the collar and the weight of soaked cloth are carefully observed rather than decorative. Behind her: compressed painterly night city, a train sliding away in soft focus, one warm carriage window reflected in her eyes and an atmospheric teal-green signal reflected on wet steel. Premium theatrical animation craftsmanship: confident economical cel drawing, delicate imperfect ink contours, sophisticated layered painted shadows, luminous but restrained skin tones, dark forest teal and cool cobalt against muted golden light, fine gouache grain, filmic depth, a believable 85mm portrait composition. Beautiful because of acting, light and emotional specificity, not glamour. No oversized doll eyes, pin-up pose, shiny porcelain skin, plastic 3D, generic neon cyberpunk excess or sexualization. Entire image is finished original anime artwork edge to edge, with no imitation of any named artist or franchise. NO mockup, photographic model, tablet, paper, desk, hands holding art, montage, UI, writing, subtitles, speech bubbles, logo or watermark. Native highly detailed 4K image.
查看完整作品动漫电影感画面:一艘小型有人驾驶飞艇正接近一座由石桥和铜穹顶组成的云端之城
Google Vertex AI · gemini-nano-banana-2.1 · · 1800 × 1005 px
原始生成提示词
Create one breathtaking original hand-painted 2D animated feature-film establishing shot, widescreen 16:9. A very small open-cockpit brass-and-linen airship, piloted by one ADULT aviator in a dark flight jacket and pale scarf, glides from the lower left into an immense cathedral-like architecture made of sunlit stone bridges, ancient observatories and hanging gardens rising from a vast sea of clouds. The world has a distinct inventive identity: sweeping limestone ribs and oxidized copper roofs, long eroded stairways and tiny climbing plants; not a recognizable franchise city. The single tiny airship and readable adult silhouette anchor scale. A giant soft cloud shadow crosses the architecture while the first warm light of sunrise reaches the highest terrace. Air currents slightly bend the scarf and fabric wings; no battle, explosions or random flying objects. Use a high wide camera with a strong foreground arch framing the scene, a clearly connected middle-distance bridge and distant silhouettes gradually dissolving into lavender atmosphere. A sophisticated cinematic color script: dusty peach, warm ivory, pale jade, deep blue-grey shadows, minute copper accents. Meticulous architectural perspective and thoughtful structural design, tactile gouache and dry-brush watercolor backgrounds, delicate cel line art for the airship, layered cloud edges, beautiful controlled light, atmospheric scale and an unmistakable sense of wonder and narrative destination. Premium theatrical animation still, restrained and mature, not a generic fantasy poster. Entire image is finished original artwork edge to edge. NO named-artist or franchise imitation, glossy 3D, photorealistic render, game UI, text, logo, watermark, map, collage, panels, book, tablet, hand-held print, desk or mockup. Native detailed 4K output.
查看完整作品四格黑白日漫:一位成年人在雨中的铁路月台上发现一封信
Google Vertex AI · gemini-nano-banana-2.1 · · 960 × 1431 px
原始生成提示词
Create a finished original museum-quality BLACK-AND-WHITE MANGA PAGE, portrait 2:3, exactly FOUR distinct panels in clear TOP-TO-BOTTOM reading order, as direct flat artwork filling the image. The four panels stack vertically, separated by three clean thin white horizontal gutters; the first and fourth panels are wide cinematic views and the middle two are closer shots. One coherent quiet mystery with the SAME ADULT male archivist, age about 40, wavy dark hair, thin round spectacles, worn dark overcoat, leather satchel: Panel 1, at the top, an establishing view of a deserted old railway platform in driving rain with the archivist alone under the canopy and an envelope on a bench. Panel 2, closer, his single anatomically correct hand lifting the unmarked sealed envelope from the bench, wet leather satchel edge and sleeve visible, no other hands. Panel 3, intense close-up of his three-quarter face and spectacles as he recognizes the silhouette of a woman at the far end of the platform reflected faintly in the lens. Panel 4, at the bottom, a wide moving train partly obscures the receding adult woman while the archivist stands still holding the envelope, one emotionally decisive departure moment. Expert traditional manga brush-ink line hierarchy, white-space composition, rich spotted blacks, elegant architectural perspective, subtle crosshatching, crisp dry-brush rain, fine screentone used sparingly, strong cinematic staging and continuity. Exactly four rectangular panels, no inset panels, no repeated scene, no dialogue or written content anywhere. The visual story should read through acting and composition alone. Premium mature graphic storytelling, no imitation of any named mangaka or existing IP. Pure monochrome black ink and ivory-white ground, NO color. NO lettering, speech balloons, sound effects, numbers, symbols, logos or watermark. NO visible physical page edges, book binding, photograph, desk, artist tools, tablet, mockup or hand holding the page. Native 4K detailed artwork.
查看完整作品三格彩绘漫画:一位探险者在被淹没的天文台内发现发光的海洋生物
Google Vertex AI · gemini-nano-banana-2.1 · · 1800 × 1208 px
原始生成提示词
Create a single finished original prestige painted GRAPHIC NOVEL strip in landscape 3:2, exactly THREE vertical comic panels side by side with two narrow ivory gutters. A coherent silent story with the SAME ADULT female deep-sea explorer, about 35, short black hair, strong practical build, ochre diving suit, round brass helmet carried at her hip, a compact lamp in her right hand. This is a richly painted 2D narrative illustration, not photographed pages or a movie screenshot. Left panel, a medium-wide shot: the explorer cautiously wades through knee-deep dark water into a half-submerged abandoned astronomical observatory, broken dome ribs framing night sky. Center panel, a nearer over-shoulder shot: she directs the small lamp toward a worn bronze orrery as tiny living blue organisms emerge from the water around its rings. Right panel, a wider reveal: she stands small in the chamber as a graceful luminous constellation of the same bioluminescent organisms climbs the orrery and connects with stars visible through the broken dome. The sense is discovery and awe, not supernatural lasers. Maintain exact outfit, lamp hand and brass helmet across the three moments; distinct plausible compositions and a consistent old stone location. Outstanding European-album-like craftsmanship expressed without copying a named artist: expressive fine ink, textured gouache, translucent watercolor washes, tactile patinated bronze, sophisticated blue-black water, muted ochre, moss green and restrained cyan bioluminescence, warm lamp highlights, confident large shadow masses and small precise material details. Clear silhouettes, believable adult anatomy, coherent space and story transitions. Finished artwork edge to edge. EXACTLY three equal-width tall panels, no fourth panel or inset, no dialogue, no lettering, no subtitles, no glyphs, no map, no logos, no watermark. NO desk, tablet, frame, paper mockup, book, photograph, hands holding artwork, glossy 3D or game interface. Native 4K detailed image.
查看完整作品动漫电影感画面:两位成年人在黄昏时分的屋顶上修理一台复古投影仪
Google Vertex AI · gemini-nano-banana-2.1 · · 1800 × 1208 px
原始生成提示词
Create one intimate, exquisitely art-directed original 2D anime feature-film frame, landscape 3:2. Two ADULT siblings in their early thirties repair their late father's vintage 16mm film projector on an old apartment rooftop at twilight before an outdoor neighborhood screening. The woman on the left has short textured dark hair, a loose rolled-sleeve cream work shirt and indigo trousers; she steadies the metal projector housing with one natural hand. Her brother on the right has slightly longer tied-back dark hair, plain muted red T-shirt, thin glasses; he leans in to carefully align a single film reel with one hand. Exactly two adult characters, two natural visible hands doing the work, other hands plausibly resting out of frame; no extra limbs. Their faces carry quiet concentration and shared affection, exchanging a small glance rather than facing the viewer. Camera at table height in a medium-wide three-quarter view: faces, hands, reel and the warm-lit mechanism all legible, not obscured. Behind them, rooftop plants, a thin clothesline and layered lived-in city roofs against a deepening periwinkle sky; one empty cream projection screen softly out of focus in the distance, no displayed image or text. One warm task lamp lights their skin and the weathered aluminum machine; the evening sky adds a cool lavender rim. Premium hand-drawn cinema treatment: restrained realistic adult proportions, graceful confident linework, nuanced character acting, tactile watercolor-gouache background, delicate cloth and metal highlights, controlled painterly edges, sophisticated indigo, dusty rose and amber color harmony, fine film grain. Emotional slice-of-life storytelling with outstanding craftsmanship, not a generic people-at-desk stock composition and not glossy 3D. Artwork itself fills the whole frame. Entirely original characters, no named artist or franchise imitation. NO UI, words, subtitles, speech bubbles, logos, watermarks, montage, hand-held artwork, tablet, comic print or physical artwork mockup. Native 4K output.
查看完整作品探索创意工作流
从简报到审核草稿

图片记录
使用 Vertex AI / Nano Banana 2.1 创建的原创作品。这些是示意静帧,不是 CyaniModel 的输出,也不是完成的动画。
Google Vertex AI · gemini-nano-banana-2.1 ·
原始生成提示词
Create a single finished original prestige painted GRAPHIC NOVEL strip in landscape 3:2, exactly THREE vertical comic panels side by side with two narrow ivory gutters. A coherent silent story with the SAME ADULT female deep-sea explorer, about 35, short black hair, strong practical build, ochre diving suit, round brass helmet carried at her hip, a compact lamp in her right hand. This is a richly painted 2D narrative illustration, not photographed pages or a movie screenshot. Left panel, a medium-wide shot: the explorer cautiously wades through knee-deep dark water into a half-submerged abandoned astronomical observatory, broken dome ribs framing night sky. Center panel, a nearer over-shoulder shot: she directs the small lamp toward a worn bronze orrery as tiny living blue organisms emerge from the water around its rings. Right panel, a wider reveal: she stands small in the chamber as a graceful luminous constellation of the same bioluminescent organisms climbs the orrery and connects with stars visible through the broken dome. The sense is discovery and awe, not supernatural lasers. Maintain exact outfit, lamp hand and brass helmet across the three moments; distinct plausible compositions and a consistent old stone location. Outstanding European-album-like craftsmanship expressed without copying a named artist: expressive fine ink, textured gouache, translucent watercolor washes, tactile patinated bronze, sophisticated blue-black water, muted ochre, moss green and restrained cyan bioluminescence, warm lamp highlights, confident large shadow masses and small precise material details. Clear silhouettes, believable adult anatomy, coherent space and story transitions. Finished artwork edge to edge. EXACTLY three equal-width tall panels, no fourth panel or inset, no dialogue, no lettering, no subtitles, no glyphs, no map, no logos, no watermark. NO desk, tablet, frame, paper mockup, book, photograph, hands holding artwork, glossy 3D or game interface. Native 4K detailed image.
查看完整作品
图片记录
使用 Vertex AI / Nano Banana 2.1 创建的原创作品。这些是示意静帧,不是 CyaniModel 的输出,也不是完成的动画。
Google Vertex AI · gemini-nano-banana-2.1 ·
原始生成提示词
Create one intimate, exquisitely art-directed original 2D anime feature-film frame, landscape 3:2. Two ADULT siblings in their early thirties repair their late father's vintage 16mm film projector on an old apartment rooftop at twilight before an outdoor neighborhood screening. The woman on the left has short textured dark hair, a loose rolled-sleeve cream work shirt and indigo trousers; she steadies the metal projector housing with one natural hand. Her brother on the right has slightly longer tied-back dark hair, plain muted red T-shirt, thin glasses; he leans in to carefully align a single film reel with one hand. Exactly two adult characters, two natural visible hands doing the work, other hands plausibly resting out of frame; no extra limbs. Their faces carry quiet concentration and shared affection, exchanging a small glance rather than facing the viewer. Camera at table height in a medium-wide three-quarter view: faces, hands, reel and the warm-lit mechanism all legible, not obscured. Behind them, rooftop plants, a thin clothesline and layered lived-in city roofs against a deepening periwinkle sky; one empty cream projection screen softly out of focus in the distance, no displayed image or text. One warm task lamp lights their skin and the weathered aluminum machine; the evening sky adds a cool lavender rim. Premium hand-drawn cinema treatment: restrained realistic adult proportions, graceful confident linework, nuanced character acting, tactile watercolor-gouache background, delicate cloth and metal highlights, controlled painterly edges, sophisticated indigo, dusty rose and amber color harmony, fine film grain. Emotional slice-of-life storytelling with outstanding craftsmanship, not a generic people-at-desk stock composition and not glossy 3D. Artwork itself fills the whole frame. Entirely original characters, no named artist or franchise imitation. NO UI, words, subtitles, speech bubbles, logos, watermarks, montage, hand-held artwork, tablet, comic print or physical artwork mockup. Native 4K output.
查看完整作品
图片记录
使用 Vertex AI / Nano Banana 2.1 创建的原创作品。这些是示意静帧,不是 CyaniModel 的输出,也不是完成的动画。
Google Vertex AI · gemini-nano-banana-2.1 ·
原始生成提示词
Create one breathtaking original hand-painted 2D animated feature-film establishing shot, widescreen 16:9. A very small open-cockpit brass-and-linen airship, piloted by one ADULT aviator in a dark flight jacket and pale scarf, glides from the lower left into an immense cathedral-like architecture made of sunlit stone bridges, ancient observatories and hanging gardens rising from a vast sea of clouds. The world has a distinct inventive identity: sweeping limestone ribs and oxidized copper roofs, long eroded stairways and tiny climbing plants; not a recognizable franchise city. The single tiny airship and readable adult silhouette anchor scale. A giant soft cloud shadow crosses the architecture while the first warm light of sunrise reaches the highest terrace. Air currents slightly bend the scarf and fabric wings; no battle, explosions or random flying objects. Use a high wide camera with a strong foreground arch framing the scene, a clearly connected middle-distance bridge and distant silhouettes gradually dissolving into lavender atmosphere. A sophisticated cinematic color script: dusty peach, warm ivory, pale jade, deep blue-grey shadows, minute copper accents. Meticulous architectural perspective and thoughtful structural design, tactile gouache and dry-brush watercolor backgrounds, delicate cel line art for the airship, layered cloud edges, beautiful controlled light, atmospheric scale and an unmistakable sense of wonder and narrative destination. Premium theatrical animation still, restrained and mature, not a generic fantasy poster. Entire image is finished original artwork edge to edge. NO named-artist or franchise imitation, glossy 3D, photorealistic render, game UI, text, logo, watermark, map, collage, panels, book, tablet, hand-held print, desk or mockup. Native detailed 4K output.
查看完整作品研究方向
工作流背后的研究
研究方向
角色一致性基准
在镜头、姿势、服装和场景变化下,角色的可识别性有多稳定?
我们会如何评估: 分别评估身份、服装、角色数量、故事一致性和每个已接受分镜的总成本。
研究方向
自适应参考选择
哪种身份、全身、服装或镜头视角参考更适合特定镜头?
我们会如何评估: 使用相同的故事和验收规则,对比固定参考集与镜头匹配集。
研究方向
质量与成本路由
更简单的镜头能否使用更便宜、允许的模型,同时满足同样的审核标准?
我们会如何评估: 统计已接受分镜时,要把生成、检查、修复和重试成本都算进去。
研究方向
视觉审核级联
哪些检查能尽早发现身份或故事错误,哪些则需要更仔细地看?
我们会如何评估: 跟踪漏检的严重错误和误拒的分镜,同时记录审核成本和延迟。
研究方向
定向角色修复
能否在保留已通过场景的前提下,只修正一个错误的角色细节?
我们会如何评估: 每次拟定修复后,都要检查编辑区域、周边像素和故事内容。
研究方向
可复用角色素材
哪些身份视图、服装和场景素材能帮助后续镜头,而不必重复旧的故事图?
我们会如何评估: 跟踪素材版本、可复用性、准备成本,以及因过期参考导致的错误。
研究方向
多角色身份
在互动过程中,每个角色都能保持正确的脸、服装、比例和位置吗?
我们会如何评估: 检查每个镜头中的身份与位置对应、角色数量以及相对大小。
研究方向
身份适配器与角色 LoRA
在什么情况下,共享身份适配器或角色专属训练步骤能提升重复角色工作流?
我们会如何评估: 在选择方法前,先对比新姿势和新身份、准备时间、训练成本以及预期复用率。
研究方向
高效漫画模型
更小或经过调整的模型,能否用更少算力达到所需的视觉标准?
我们会如何评估: 在适用的模型许可下,衡量已接受输出的质量、延迟、内存、吞吐量和训练摊销。
常见问题
什么是 CyaniModel?
这里的 CyaniModel 是 LlamaGen 创意工作流和视觉叙事研究所使用的名称。可用工具可能会调用多个模型提供商。这些页面并不意味着 LlamaGen 为每个工作流都拥有独立训练的基础模型。
可以把研究实验室当作正式功能使用吗?
每个实验室都描述了我们想研究的问题,以及评估方式。当前可用功能请使用关联的 LlamaGen 创作工具;研究说明不代表已启用功能,也不保证结果。
重复出现的角色应该从哪里开始?
准备一张原创身份图、当前服装,以及一份需要保持稳定的细节清单。先从一个小序列开始,一起检查分镜,并记录哪些镜头需要调整。
这些图片能证明模型表现吗?
这些图像展示的是原创作品和示意性使用场景。真正的基准测试需要固定输入、可比较的设置、明确的验收标准,以及完整结果——包括失败案例。仅靠宣传场景无法提供这些证据。
继续你的故事
创建属于同一故事的图片
继续在 LlamaGen 中查看可用工具、模型选择、账户要求和积分后再生成。
Original Cyani Model introduction · preserved
Cyani Model: ACG writing, roleplay and branching stories
This retains the earlier LlamaGen introduction to Anime, Comic and Game writing: model overview, roleplay, educational dialogue, branching fiction, how-to steps, FAQ, posters and creator examples. It complements the new visual-storytelling workflows rather than replacing them.
Historical marketing copy, not a verified statement of current model architecture or performance. The original called Cyani an AI-native LLM; this alone does not prove separately trained weights, ownership of a foundation model or a current runtime choice. Read the original English content below; verify current model access and credits in the linked tools.
Cyani Model: Free ACG Writing Guide
Explore the Cyani Model guide and ACG writing ideas for free. Plan character dialogue, roleplay, and branching stories, then check the access and credits for the creation workflow you choose.
Original model description: An AI-native LLM designed for ACG (Anime, Comic, Game) creation — optimized for creative roleplaying, narrative branching, education scenarios, and consistent character voice.
Powered by Cyani Model
Explore comic creation
ACG with AI — AI Story & Comic Creation for Young Creators
Roleplaying First
Creative dialogue branching from Games/Roleplaying and Arts & Entertainment domains.
Education Ready
Curriculum-aligned prompts from Jobs & Education/Education for learning scenarios.
Creative Writing
Narrative branches sourced from Books & Literature and Arts & Entertainment.
Original ACG tools
AI Manga Maker
Design manga panels with crisp B/W line art and speech bubbles.
AI Animation
Storyboard frames with motion guidance to visualize animation beats.
OC Maker
Create character sheets with poses, color palettes, and accessories.
Creators’ Stories — ACG with AI
How creators use Cyani Model for AI Anime, AI Comic, and AI Game
Interactive Roleplay
Design character-driven storylines with flexible responses, context retention, and emotional nuance.
Educational Dialogue
Create tutoring sessions, quizzes, and guided lessons with structured branches.
Fiction Authoring
Outline chapters, refine prose, and explore plot variations with consistent character voices.
How It Works
Choose a Scenario
Select roleplay, education, or storytelling templates to start.
Define Characters & Context
Provide goals, personas, and world constraints for consistency.
Branch & Iterate
Explore dialogue branches, refine tone, and publish final drafts.
Build a brief for character dialogue and branching stories
Cyani focuses on text-based ACG planning: define a character, describe the setting, and explore possible dialogue or story branches. Review continuity and facts before using a draft.
Help Center FAQs · original
What is Cyani Model intended for?
Cyani is presented as an ACG-oriented language workflow for character dialogue, roleplaying ideas, and narrative planning. It is separate from the photo-to-manga image filter.
Does a text draft guarantee consistent final artwork?
No. A character brief can guide later image work, but artwork consistency depends on the image workflow, references, and manual review.
Can I use it for education or roleplay?
Use the drafts as a starting point for lessons, character dialogue, or roleplay. An adult should review classroom material and verify factual claims.
Can I explore Cyani Model for free?
Yes. Read this Cyani Model guide and explore its ACG writing examples for free. Connected story, comic, and image-generation workflows have their own account, model-access, and credit requirements; check the chosen tool before generating.
Original posters, creator personas and gallery (preserved)
ACG with AI — Poster Showcase



Original fictional creator profiles
Emi · Manga Artist
Indie manga artist refining long arcs with consistent character voices · Warm, detail-focused, keeps consistent voice
Kenji · Teacher
High school teacher running classroom roleplays in literature · Patient, structured, educational storyteller
Sora · Indie Creator
Indie creator prototyping character sheets and manga conversions · Thoughtful, iterative, indie creator
ACG with AI — Visual Showcase












