# Camera Motion LoRA — User GuideNew custom dataset collected specifically to train this camera-motion LoRA, designed to compensate for missing camera movement capabilities within the MINIMAX H3 base model.## 1. Mandatory Trigger Word<code>camera motion</code>Must be placed at the very start of the prompt.It acts as the activation switch to enable camera motion control for this LoRA, followed by specific camera movement + scene description.## 2. File Versions1000-step file | 3000-step filecamera_motion_h3_lora_v1_1000_pruned.safetensors | camera_motion_h3_lora_v1_3000_pruned.safetensors## 3. Installation &amp; LoadingPlace the <code>_pruned</code> files into <code>ComfyUI/models/loras/</code>Load via LoraLoader. Recommended <code>strength_model</code>: 0.8 ~ 1.0- 0.8: Mild camera motion- 1.0: Distinct camera motion- &gt;1.2: Risk of unstable framesBase model: MiniMax-H3 video workflow (pruned + int8 + convrot version)## 4. Prompt Usage (Core Section)### Prompt Structure for Training SamplesAnnotation format used for training data (the syntax best matched for this LoRA):<code>camera motion, [Subject], [Scene], [Camera motion description], [Lighting / Style], [Quality]</code>### Use Raw Training Prompts (Recommended)<code>camera_motion_prompt_library.md</code> contains 55 complete prompts extracted directly from training samples, sorted into 9 camera motion categories (Push-in / Pull-back / Combined Push-Pull / Orbit / Tracking / Aerial / Crane / Pan / Macro). Copy and paste directly for results closest to the training dataset.Example:<code>camera motion, a woman and a man in athletic wear, engaged in a dynamic rooftop martial arts fight with punches and defensive blocks, nighttime urban cityscape background, slow push-in to medium shot transitioning to slight pull-back as the woman falls, cinematic teal-orange grading, water droplet effects, sharp 4k, high production quality</code>### Swap Scene, Keep Camera MotionReplace <code>[Subject]</code> and <code>[Scene]</code> while retaining the camera motion phrase after the comma:<code>camera motion, [Your Subject], [Your Scene], slow push-in with subtle lateral tracking, cinematic color grading, 4k, high quality</code>### Swap Camera Motion, Keep SceneOnly modify the camera motion phrase (e.g. <code>slow push-in</code> → <code>slow orbit</code>), leave the rest unchanged.### Full Ready-to-Use Examples① Push-in (Martial Arts / Night Scene)<code>camera motion, a woman and a man in athletic wear, engaged in a dynamic rooftop martial arts fight with punches and defensive blocks, nighttime urban cityscape background, slow push-in to medium shot transitioning to slight pull-back as the woman falls, cinematic teal-orange grading, water droplet effects, sharp 4k, high production quality</code>② Pull-back (Industrial / Volumetric Light)<code>camera motion, two men in a dynamic martial arts fight, exchanging kicks and punches with defensive blocks and dodges, inside a vast abandoned industrial warehouse with crumbling pillars and dramatic shafts of sunlight, slow pull-back combined with gentle orbit around the fighters revealing the full scale of the environment, cinematic gritty texture, natural volumetric lighting, high production quality</code>③ Orbit (Wuxia / Snow Scenery)<code>camera motion, two martial artists in white and dark robes, dueling with sword and fan atop a snow-covered pavilion surrounded by swirling vortex energy, misty mountain landscape with weeping willow, dynamic pull-back from close-up vortex to wide establishing shot revealing the full rooftop confrontation, cinematic wuxia fantasy style, ethereal atmospheric lighting, 4k, high quality</code>④ Handheld Tracking (Action / Neon Rainy Night)<code>camera motion, a man and woman in athletic wear, engaged in intense martial arts combat with punches kicks and takedowns, nighttime rooftop with wet pavement reflecting neon city skyline, dynamic handheld camera with push-in and slight tracking following action, cinematic teal-orange color grading, sharp 4k, high quality</code>⑤ Aerial (Nature / Mountain Landscape)<code>camera motion, majestic green mountain peaks and misty valleys, aerial drone descending over cloud-wrapped ridges, low-angle water-level shot of splashing stream, two hikers crossing stone bridge, mountain forest with flowing creek, slow aerial push-in, water-level tracking, static wide, cinematic natural lighting, 4k, high quality</code>⑥ Miniature Diorama (Macro / Desktop Diorama)<code>camera motion, a miniature seasonal village diorama on a desk, transforming through autumn, winter, and spring with changing foliage and weather effects, office desk environment with computer monitor and stationery, slow push-in dolly glide across the diorama surface revealing each season transition, whimsical miniature aesthetic, soft natural lighting, 4k, high quality</code>More complete examples can be found in <code>camera_motion_prompt_library.md</code> (55 prompts covering all 9 camera motion categories).## Test Videos### Camera Motion Strength &amp; Direction Control| Requirement | Prompt Writing Method || ---- | ---- || Smoother motion | Add <code>slow</code>: slow push-in || Intense / Aggressive motion | Add <code>aggressive</code>: dynamic handheld / aggressive push-in || Specify direction | push-in from left / orbit right / pan left |### Supported Camera Motion Types (Validated on Training Dataset)| Type | Stability | Recommended Prompt Phrase || ---- | ---- | ---- || Push-in | ★★★ Most stable | slow push-in / dolly in / push in || Pull-back | ★★★ Most stable | pull-back / dolly out || Handheld Tracking | ★★★ Most stable | handheld tracking shot / dynamic handheld || Orbit | ★★☆ Stable | slow orbit / orbit around || Aerial | ★★☆ Stable | aerial drone shot / flyover || Crane Shot | ★★☆ Stable | crane reveal / tilt up/down || Combined Push &amp; Pull | ★★☆ Stable | push-in then pull-back || Pan | ★☆☆ Weak (limited samples) | pan left / pan right (low sample count; recommended to combine with other terms) || Macro / Close-up | ★★☆ Stable | close-up / macro / extreme close-up |

Files with SHA256: 9d7d98d7377f56efed3aa7d507f767936112d208906f49908ec9a8ae912be88b