Join my Discord for help, workflows and news of new LoRAs!

⚡ Aether Crash Zoom – Telephoto Crash-Zoom LoRA for Wan 2.2 5B (i2v)

Aether Crash Zoom is a custom LoRA trained for cinematic crash zooms from a telephoto distance — sudden, high-speed zoom-ins toward a distant subject, with compressed depth and dramatic energy.

It was trained specifically for image-to-video (i2v) workflows using Wan 2.2 5B, meaning it expects a starting image that defines the composition and subject. The LoRA then guides the model to zoom rapidly into that subject across time. It might work on t2v as well, but that's not something I've tested out yet. Focus is human subjects - that's what it was trained on, so might work best for that!

🧠 How to Use (i2v)

1. Start with a single image
   – Your image should clearly feature the subject you want to zoom into
   – Use long-distance framing to support the telephoto feel

2. Prompt using the format:

   rapid zoom in on [subject] 

   Example:

   rapid zoom in on a lone warrior standing still on a distant hilltop

   – The subject must be present in the image
   – The LoRA tells the model to crash zoom into that specific figure

⚙️ Inference Settings

• Mode: image-to-video (i2v)

• Resolution: 1280×720 (landscape)

• FPS: 24 (recommended)

• Clip length: around 1-3 seconds per generation is recommended (more might work fine - lmk if you get good results!)

✅ Best Practices

• Use an initial image with the subject clearly visible and placed at a (far) distance

• Prompt the subject directly, keeping text minimal and cinematic

• Avoid busy compositions — the LoRA is optimized for strong, focused subject motion

Aether Crash Zoom gives your i2v generations a signature kinetic identity — punchy, precise, and perfect for music videos, trailers, or poetic intros with maximum subject impact.

https://tensor.art/u/635897593804625782