CivArchive
    Motion-Speed-MMH3-SLIDER - v1.1
    NSFW

    Frustrated by AI videos always looking sluggish? Or are you looking for that perfect cinematic slow-motion? The Motion Speed Slider is a Concept Slider LoRA custom-built for MiniMax H3. It grants you absolute control over character action speed and spatial movement in your generations!

    No need for complex prompt engineering. By simply adjusting the LoRA weight, you can seamlessly switch between "hyper-speed dashes" and "bullet time," making it perfectly suited for high-dynamic scenes like combat, dancing, and dynamic camera tracking.

    🆕 What's New in V2

    Reduced Composition Interference: Minimized the LoRA's impact on the overall framing and composition of the generated video.

    Reference Video Generation Support: Added examples demonstrating how to use this LoRA with reference-based video generation workflows.

    ⚙️ How to Use (V2 Version) V2 has been optimized with a new parameter scaling system.

    Recommended Settings: Min Strength -8, Max Strength 8, and Default Strength 4.

    Behavior: Note that keeping the strength below 4 will largely maintain the original video content (as if no LoRA is applied), while a strength above 4 may introduce significant changes to the output.

    ⚙️ How to Use (V1 Version) V1 operates purely as a standard positive/negative slider. It requires NO trigger words. ⏩ To Accelerate (Positive Weight):

    Recommended Weight: +0.5 to +4.0

    Effect: Significantly speeds up character actions and spatial movement.

    Best For: Intense martial arts combat, high-speed running, fast camera panning, and explosive action sequences.

    ⏪ To Decelerate (Negative Weight):

    Recommended Weight: -0.5 to -4.0

    Effect: Slows down the action, delivering a smooth, slow-motion texture.

    Best For: Cinematic slow-motion, bullet time, elegant dance delays, and emotional close-up shots.

    苦恼于 AI 视频的动作总是慢吞吞的?或者想要完美的电影级慢动作? Motion Speed Slider 是专为 MiniMax H3 打造的概念滑块(Concept Slider)LoRA。它赋予你对画面中人物动作快慢和空间位移速度的绝对控制权!

    无需复杂的提示词玄学,只需简单地调整 LoRA 权重,即可实现从“极速瞬移”到“子弹时间”的无缝切换,完美适配动作打斗、舞蹈、运镜等高动态场景。

    🆕 V2 版本更新说明

    降低构图干扰: 最小化了该LoRA对生成视频整体画面取景和构图的影响。

    参考视频生成支持: 添加了相关示例,演示如何在基于参考视频的生成工作流中使用此LoRA。

    ⚙️ 使用方法 (V2 版本) V2 版本采用了全新的权重范围机制:

    推荐参数: 最小权重 -8,最大权重 8,默认权重 4。

    效果说明: 注意,当权重保持在 4 以下时,会基本保持无LoRA时的视频原内容;而大于 4 之后可能会对视频内容产生较大的改变。(适用于需要极高强度控制的场景)

    ⚙️ 使用方法 (V1 版本) V1 版本为标准滑块机制,无需触发词,仅通过正负权重(Weight)来控制画面速度:

    ⏩ 想要加速 (正权重 / Positive Weight):

    推荐权重: +0.5 到 +4.0

    效果: 显著加快人物动作与画面移动速度。

    适用场景: 激烈的武术格斗、高速奔跑、极速运镜、充满爆发力的动作片段。

    ⏪ 想要减速 (负权重 / Negative Weight):

    推荐权重: -0.5 到 -4.0

    效果: 减缓动作速度,带来平滑的慢放质感。

    适用场景: 电影级慢动作(Slow-motion)、子弹时间、优雅的舞蹈延时、抒情的特写镜头。

    Description

    Version 1.1 Update Notes:

    Reduced Composition Interference & Recommended Settings: Minimized the LoRA's impact on the overall framing and composition of the generated video. Recommended settings are Min Strength -8, Max Strength 8, and Default Strength 4. Note: keeping the strength below 4 will largely maintain the original video content (as if no LoRA is applied), while a strength above 4 may introduce significant changes.

    Reference Video Generation Support: Added examples demonstrating how to use this LoRA with reference-based video generation workflows.

    1.1版本更新说明:

    降低构图干扰与推荐设置: 最小化了该LoRA对生成视频整体画面取景和构图的影响。推荐设置:最小权重(Min Strength)-8,最大权重(Max Strength)8,默认权重(Default Strength)4。注意:当权重保持在4以下时,会基本保持无LoRA时的视频原内容;而大于4之后可能会对视频内容产生较大的改变。

    参考视频生成支持: 添加了相关示例,演示如何在基于参考视频的生成工作流中使用此LoRA。

    FAQ

    Comments (1)

    HugMeIntoFaceOct 7, 2026
    CivitAI

    Whatever you did, you did a great work!

    LORA
    MiniMax H3

    Details

    Downloads
    280
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/7/2026
    Updated
    10/7/2026
    Deleted
    -