捉妖师

作者:翻山计划发布日期:2026-04-14观看:176
捉妖师

妖魔横行,帝斩天职。

01/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 8
捉妖师|图片节点 8图片节点 8

Cinematic game character splash art. a young woman with long disheveled black hair and empty. hollow eyes. kneeling on a blood-soaked battlefield. wearing a tattered and torn white kimono stained with crimson. in her trembling hands she loosely grips a massive cursed odachi. the blade is longer than her entire body. pulsing with an ominous malevolent red glow. the red light creeps up her arms like veins. her face shows a flicker of horror and remorse. yet the corner of her mouth twitches with a ghost of a cruel smile. cherry blossom petals caught mid-air frozen in the bloody mist. hyperrealistic. hyperdetailed. unreal engine 5 style. --ar 16:9

02/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 9

Cinematic game key art. colossal Yamata no Orochi. an eight-headed eight-tailed giant serpent deity filling an entire mountain valley. each head as large as a shrine. sixteen crimson eyes glowing like pools of blood staring from different angles. scaly dark body coiling through misty ravines and shattered torii gates. tiny human figures fleeing in the foreground for scale. oppressive ancient evil atmosphere. storm clouds and red lightning overhead. unreal engine 5 render. hyperrealistic. hyperdetailed. --ar 16:9

03/ 36图片生成mj-v7image2imageratio=16:9 quality=auto图片节点 10
捉妖师|图片节点 10图片节点 10

Cinematic game key art. Fei. a colossal bull-like beast with a single eye. its body emitting a faint aura of death. standing in a once-lush forest that is now a withered wasteland. trees dead. riverbed cracked and dry. an embodiment of blight and decay. oppressive atmosphere of silent doom. unreal engine 5 style. hyperrealistic. hyperdetailed. --ar 16:9

04/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 12
捉妖师|图片节点 12图片节点 12

Cinematic game character splash art. Onikiri. a white-haired demon with piercing crimson eyes and a face twisted in cold fury. originally the legendary demon-slaying blade of the Genji clan. now a vengeful spectral swordsman. countless shattered sword fragments and broken blades swirl around him in a violent metallic storm. some pieces floating. others embedded in the ground around his feet. his tattered white and black robes whip in the gale. one hand outstretched. fingers curled as if commanding the shards. dark crimson energy trails from the fragments. hyperdetailed. hyperrealistic. unreal engine 5 style. --ar 16:9

05/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 11 - 副本

Cinematic game key art. Hundun. one of the Four Perils of Chinese mythology. a colossal bloated bestial entity. resembling a bear or dog but with no facial orifices. a body of pure chaotic flesh. standing in a barren landscape. the ground around it warped and twisted by its presence. an embodiment of primordial confusion and moral inversion. deeply unsettling and oppressive. unreal engine 5 style. hyperrealistic. hyperdetailed. --ar 16:9

06/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 11 - 副本
捉妖师|图片节点 11 - 副本图片节点 11

Cinematic game key art. epic scale battle. Otengu a handsome arrogant tengu warrior with massive black feathered wings. tiny figure suspended in the stormy sky. facing Yamata no Orochi. a colossal eight-headed eight-tailed serpent deity that fills the entire mountain valley below. each serpent head is larger than a shrine. sixteen crimson eyes glowing like pools of blood staring from all directions. the tengu's black wings are fully spread but appear as a speck against the monster's immense coiled body. wind and purple lightning clashing. shattered torii gates crushed under the serpent's scales. overwhelming scale. extreme size contrast. hyperdetailed. hyperrealistic. unreal engine 5 style. --ar 16:9

07/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 6

Cinematic game key art. a barefoot young man in torn white robes dragging a massive rusted iron coffin by chains. walking along a colossal rib bone of a giant skeleton mountain. translucent ghostly hands reaching out from the coffin cracks. resting on his shoulders. misty white abyss below. pale golden light breaking through clouds. hyperrealistic. unreal engine 5. --ar 16:9

08/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 7
捉妖师|图片节点 7图片节点 7

{{magic:247f7d76-f949-4c78-9b67-1ccebfa3bbe6_1:0}}图片中有四只翅膀,去掉前面收起来的两只翅膀

09/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 5

根据 {{Image 2}} 的衣服给 {{Image 1}} 扩图,变成全身照,场景氛围保持 {{Image 1}} 不变

10/ 36图片生成nebula-ultratext2imageratio=16:9 quality=2K图片节点 16

16:9 wide composition divided into three vertical panels. Left panel: extreme close-up portrait of face. Center panel: full body front view. showcasing complete outfit. standing pose. Right panel split horizontally into two smaller sections. Top right: full body side view from left. Bottom right: full body back view. White background. clean studio lighting. concept art style. character reference sheet. high resolution. --ar 16:9

11/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 6

只保留张开的翅膀

12/ 36图片生成nebula-ultratext2imageratio=16:9 quality=2K图片节点 7
捉妖师|图片节点 7图片节点 7

Cinematic game character splash art. a devastatingly handsome young man with delicate facial features but a cold. condescending glare. wearing pristine white and gold Heian-era robes. holding a closed ornate fan in one hand. behind him massive jet-black feathered wings unfurl to fill the sky. whipping up a violent tempest of wind and leaves. storm clouds spiraling overhead. traditional Japanese temple courtyard in the background being torn apart by the gale. elegant yet terrifying presence. unreal engine 5 style. hyperdetailed. hyperrealistic. --ar 16:9

13/ 36图片生成nebula-ultratext2imageratio=16:9 quality=2K scene=camera_focus图片节点 2

脸部瑕疵,战斗感,少年感

14/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 10
捉妖师|图片节点 10图片节点 10

{{Image 1}} 中侧身的人物换成 {{Portrait 2}} 中的角色,氛围保持 {{Image 1}} 的感觉

15/ 36图片生成mj-v7image2imageratio=16:9 quality=auto图片节点 8 - 副本

参考图中的人物和龙战斗的画面,俯视,200mm镜头,龙在人物身后,对称构图,巨物感,压迫感

16/ 36图片生成mj-v7image2imageratio=16:9 quality=auto图片节点 8
捉妖师|图片节点 8图片节点 8

参考图中的人物和龙战斗的画面

17/ 36图片生成mj-v7image2imageratio=16:9 quality=auto图片节点 12

以参考图的形象生成夸张的战斗动作

18/ 36图片生成image2image多角度
捉妖师|多角度

dutch angle,tilted frame

19/ 36图片生成mj-v7text2imageratio=16:9 quality=auto图片节点 11

Cinematic game key art. Yinglong. the only winged dragon in Chinese mythology. colossal ancient divine dragon soaring above mountains. massive feathered wings spread wide. scales shimmering with celestial light. muscular body coiling through clouds. benevolent yet overwhelming divine power. awe-inspiring majesty. unreal engine 5 style. hyperrealistic. hyperdetailed. --ar 16:9

20/ 36视频生成star-video2mixed2videoratio=21:9 resolution=720p duration=15 enableSound=on视频节点 11
捉妖师|视频节点 11视频截图视频节点 11

{{Portrait 1}}的男性角色快速飞行,趁机飞到 {{Portrait 2}} 龙的身后,从尾巴开始跑到龙的颈部部,踩住颈部高高跃起,拔出腰间佩剑(子弹时间),重重的插住龙的颈部,龙受伤,摔倒地上,大场面,电影感,可以生成音效,不要生成音乐

21/ 36视频生成star-video2text2videoratio=16:9 resolution=720p duration=15 enableSound=on视频节点 9 - 副本
捉妖师|视频节点 9 - 副本视频截图视频节点 9

场景参考 ,第一个镜头是 的眼神特写(又淡淡的惊讶转为坚毅),镜头快速切远景, 的男性和 的中国神话龙 “应龙”在空中激烈的缠斗。 可以有音效,不要有音乐。

22/ 36视频生成star-video2mixed2videoratio=21:9 resolution=720p duration=14 enableSound=on视频节点 9

场景参考 {{Portrait 4}} ,第一个镜头是 {{Portrait 3}} 的眼神特写(又淡淡的惊讶转为坚毅),镜头快速切远景, {{Portrait 1}} 的男性和 {{Portrait 2}} 的中国神话龙 “应龙”在空中激烈的缠斗,缠斗过程体现打击感,关键打斗处使用慢动作特写镜头。 可以有音效,不要有音乐。

23/ 36视频生成star-video2mixed2videoratio=16:9 resolution=720p duration=15 enableSound=on视频节点 19
捉妖师|视频节点 19视频截图视频节点 19

{{Portrait 2}} 是场景参考, {{Portrait 5}} 是人物面部细节参考, {{Portrait 3}} 是角色设定图, {{Portrait 4}} 是 龙的参考图。 根据 {{Portrait 1}} 的站位,镜头1:男性角色飞行中的半身俯视视角,螺旋上升的运镜 镜头2: 龙从男性后面追了上来(男性主角回头的视角) 镜头3:两人对打被互相弹开一大段距离 可以生成音效,不要生成音乐

24/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 15

把 {{Image 2}} 的中国神话“应龙”放到 {{Image 1}} 中的云层位置,人物保持 {{Image 1}} 动作构图

25/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 18
捉妖师|图片节点 18图片节点 18

{{Portrait 2}} 是脸部参考 **Role:** Cinematic Director & Storyboard Artist **Task:** Create a 3x3 "Cinematic Montage Grid" based on the input image. **Goal:** Depict a **progressive storyline** (Time t=0s to t=5s) while simultaneously using **diverse camera angles and focal lengths**. **Directive:** Do NOT just show the same moment from different angles. Do NOT just show a static camera with moving subjects. **YOU MUST combine "Time Progression" with "Camera Movement".** **The Narrative Arc (3x3 Grid Structure):** **Row 1: The Setup (Time: Start | Wide & Dynamic)** * **Panel 1 (Extreme Long Shot):** Establishing the scene. The subject is entering the space or starting the action. * **Panel 2 (Long Shot / Low Angle):** The subject moves closer. Camera looks up to emphasize presence. Action is initiating. * **Panel 3 (Full Shot / Tracking):** Subject is in full motion. Camera tracks alongside them. **Row 2: The Conflict/Action (Time: Middle | Mid-Range & Interaction)** * **Panel 4 (Medium Shot):** The core action occurs (e.g., interacting with an object, talking, fighting). Focus on body language. * **Panel 5 (Over-The-Shoulder / POV):** A perspective shift. Seeing what the subject sees or facing the subject directly during the peak of action. * **Panel 6 (Medium Close-Up):** The action intensifies. Camera pushes in. Focus shifts to the upper body/reaction. **Row 3: The Resolution/Detail (Time: End | Close & Intimate)** * **Panel 7 (Close-Up):** Tight focus on the face or main focal point. Capturing the emotional reaction or effort. * **Panel 8 (Extreme Close-Up / Macro):** A specific detail of the action (e.g., a hand gripping, an eye widening, a foot landing). * **Panel 9 (High Angle / Top Down):** The aftermath or final state. Looking down at the subject as the action concludes. **Visual Consistency Rules:** 1. **Subject Identity:** The character's face, costume, and props must remain identical across all shots. 2. **Lighting Continuity:** The direction and mood of the light (e.g., sunset, neon, diffuse) must match the original image exactly. 3. **Montage Flow:** The sequence should feel like a professionally edited movie trailer where cuts are fast but logically connected. A photorealistic 3x3 storyboard grid. - **Top Row:** Wide shots, movement starting. - **Middle Row:** Medium shots, action peaking. - **Bottom Row:** Close-ups/Angles, action concluding. High contrast, cinematic composition, distinct depth of field for each shot type. no text,just picture

26/ 36图片生成nebula-ultratext2imageratio=16:9 quality=2K图片节点 15 - 副本

把 的中国神话“应龙”放到 中

27/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=4K图片节点 16

高清重绘,构图动作保持原图不变

28/ 36视频生成star-video2mixed2videoratio=21:9 resolution=720p duration=14 enableSound=on视频节点 13
捉妖师|视频节点 13视频截图视频节点 13

继续生成14秒,场景参考 {{Portrait 3}} , {{Portrait 1}} 的男性角色趁机飞到 {{Portrait 2}} 龙的身后,从尾巴开始跑到龙的颈部,踩住颈部高高跃起,拔出腰间长剑(子弹时间),重重的插住龙的颈部,龙受伤,摔倒地上,大场面,快节奏,电影感,可以生成音效,不要生成音乐

29/ 36视频生成star-video2text2videoratio=21:9 resolution=720p duration=14 enableSound=on视频节点 13 - 副本

继续生成14秒,场景参考 , 的男性角色趁机飞到 龙的身后,从尾巴开始跑到龙的颈部,踩住颈部高高跃起,拔出腰间长剑(子弹时间),重重的插住龙的颈部,龙受伤,摔倒地上,大场面,快节奏,电影感,可以生成音效,不要生成音乐

30/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 14
捉妖师|图片节点 14图片节点 14

高清重绘,电影感,紧张氛围

31/ 36图片生成nebula-ultraimage2imageratio=16:9 quality=2K图片节点 21

Generate a single 3x3 grid image, numbered 1 to 9 left to right, top to bottom. This grid expands the story BEFORE the provided reference image (male protagonist and Yinglong). For each cell, use only the following camera/composition instructions – DO NOT describe specific content, let AI imagine: 1. Extreme wide / master establishing shot (aerial view, vast environment) 2. Wide shot (protagonist and dragon very small, spatial relationship before meeting) 3. Medium wide shot (protagonist walking alone, mysterious architectural silhouettes in bg) 4. Medium shot (Yinglong flying toward camera, low angle) 5. Close-up (protagonist noticing the dragon, facial reaction) 6. Extreme close-up (key object or dragon claw/scales, hinting at bond) 7. Medium two-shot (protagonist and dragon first eye contact, both in frame) 8. Overhead high angle (both standing on a high place or before a grand building) 9. Wide vista (protagonist on dragon back or side by side, ready to depart) All cells: cinematic lighting, 16:9 aspect ratio, character designs consistent with reference (but not copying the pose). AI is free to invent all settings, architecture styles, and prequel events.

32/ 36视频生成star-video2text2videoratio=16:9 resolution=720p duration=15 enableSound=on视频节点 19 - 副本
捉妖师|视频节点 19 - 副本视频截图视频节点 19

是场景参考, 是人物面部细节参考, 是角色设定图, 是 龙的参考图。 根据 的站位,镜头1:两人飞到空中缠斗, 镜头2: 男性角色被龙攻击到了一下, 镜头3:两人对攻造成攻击波,使得两方被震开一段距离 可以生成音效,不要生成音乐

33/ 36图片生成nebula-ultraimage2imageratio=21:9 quality=2K scene=multi_camera_grid_9多机位九宫格

/

34/ 36图片生成nebula-ultraimage2imageratio=21:9 quality=2K图片节点 24

去掉场景中的龙,增加氛围感

35/ 36图片生成nebula-ultraimage2imageratio=21:9 quality=2K图片节点 23
捉妖师|图片节点 23图片节点 23

去掉场景中的人物,镜头拉远,高质量场景渲染

36/ 36视频生成star-video2mixed2videoratio=16:9 resolution=720p duration=15 enableSound=on视频节点 25 - 副本
捉妖师|视频节点 25 - 副本视频截图视频节点 25

镜头1: {{Portrait 1}} 俯瞰大场景镜头推进。 镜头2: {{Portrait 2}} 远处云层中出现应龙的身影。 镜头3: {{Portrait 3}} 应龙特写,眨眼。 镜头4: {{Portrait 4}} 羽毛细节特写。 镜头5: {{Portrait 5}} 振翅而起,带一点风的效果。 可以生成音效,不要生成音乐。

🔥热搜词

评论0

登录后参与评论

还没有评论,来抢沙发吧