<h1 id="kxsr-staircase-ascend-lora-for-wan-2.1-14b-ty6ia9sn7">KXSR Staircase Ascend LorA for WAN 2.1 14B</h1>In collaboration with @machinedelusions [<a target="_blank" rel="ugc" href="https://civarchive.com/user/machinedelusions">https://civarchive.com/user/machinedelusions</a>] KXSR Labs presents: This LoRA enables the creation of cinematic sequences featuring subjects ascending staircases in imaginative environments. The typical view will be from behind the subject as they ascend a spiral staircase featured on the right side of the vertical aspect ratio generation. Can still create fun results at other aspect ratios/resolutions. Use the trigger word "kxsr" to activate the model's specialized training.<h2 id="prompt-format-p6mguwprc">Prompt Format</h2><pre><code>kxsr, [person/thing] ascending a [adjective] staircase [describe the world around them]
</code></pre><h3 id="example-prompt-mryyj2lg4">Example Prompt</h3><pre><code>kxsr, high quality nature video featuring a red panda ascending a bamboo staircase while a bird lands on it's head, on the background there is a waterfall
</code></pre><h2 id="recommended-settings-8z13f1pl2">Recommended Settings</h2><ul><li>CFG: 5.5</li><li>Shift value: 4.5</li><li>LoRA strength: 1.0</li><li>~73 frames</li><li>720x1280 vertical aspect ratio</li></ul><h2 id="technical-details-6gr7rles7">Technical Details</h2><ul><li>Base model: WAN 2.1 14B Text2Video</li><li>Training dataset: 80 clips</li><li>Resolution: 720x1280 (vertical format)</li><li>Frame count: 73 frames per clip</li><li>For optimal results, maintain these specifications during inference </li></ul>This LoRA works best when you provide detailed descriptions of both the subject and the surrounding environment while following the prescribed format. Screenshot shows my typical inference testing setup for lora evals: <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5cbecd08-aef9-40c9-a77b-5cc32d08e97e/width=525/5cbecd08-aef9-40c9-a77b-5cc32d08e97e.jpeg" />

KXSR Staircase Ascend Concept WAN 14B T2V

WAN 14B T2V - Epoch 6 - trained 3/18/2025

kxsr_WAN14B_staircase_e6.safetensors

v1.0

<h1 id="kxsr-staircase-ascend-lora-for-wan-2.1-14b-ty6ia9sn7">KXSR Staircase Ascend LorA for WAN 2.1 14B</h1>In collaboration with @machinedelusions [<a target="_blank" rel="ugc" href="https://civitai.com/user/machinedelusions">https://civitai.com/user/machinedelusions</a>] KXSR Labs presents: This LoRA enables the creation of cinematic sequences featuring subjects ascending staircases in imaginative environments. The typical view will be from behind the subject as they ascend a spiral staircase featured on the right side of the vertical aspect ratio generation. Can still create fun results at other aspect ratios/resolutions. Use the trigger word "kxsr" to activate the model's specialized training.<h2 id="prompt-format-p6mguwprc">Prompt Format</h2><pre><code>kxsr, [person/thing] ascending a [adjective] staircase [describe the world around them]
</code></pre><h3 id="example-prompt-mryyj2lg4">Example Prompt</h3><pre><code>kxsr, high quality nature video featuring a red panda ascending a bamboo staircase while a bird lands on it's head, on the background there is a waterfall
</code></pre><h2 id="recommended-settings-8z13f1pl2">Recommended Settings</h2><ul><li>CFG: 5.5</li><li>Shift value: 4.5</li><li>LoRA strength: 1.0</li><li>~73 frames</li><li>720x1280 vertical aspect ratio</li></ul><h2 id="technical-details-6gr7rles7">Technical Details</h2><ul><li>Base model: WAN 2.1 14B Text2Video</li><li>Training dataset: 80 clips</li><li>Resolution: 720x1280 (vertical format)</li><li>Frame count: 73 frames per clip</li><li>For optimal results, maintain these specifications during inference </li></ul>This LoRA works best when you provide detailed descriptions of both the subject and the surrounding environment while following the prescribed format. Screenshot shows my typical inference testing setup for lora evals: <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5cbecd08-aef9-40c9-a77b-5cc32d08e97e/width=525/5cbecd08-aef9-40c9-a77b-5cc32d08e97e.jpeg" />

Files with SHA256: fcb7985381be2780ecdadf314eb227e530c6fb4c9c3b60ad7fc08ddd43af4fcb