Aether IN-D - Wan 2.2 14b t2i LoRA

🎬 Aether IN-D – Cinematic 3D LoRA for Stylized Character Stills (t2i / single-frame use)

Aether IN-D is a LoRA that brings cinematic 3D visuals to your single-frame generations — inspired by film lighting and framing, with a touch of indie animation style.

Though built as a t2v-LoRA, Aether IN-D is used as a still-image LoRA, generating one frame at a time with Wan 2.2 14B Low Noise (but surely you can do some video with it as well). It’s trained primarily on human characters and responds best to prompts that guide lighting, mood and visual storytelling.

💬 Default Trigger Phrase
Trigger the style using this phrase, then follow with your scene description:
cinematic 3D

🧠 How to Use (single-frame t2i)
• Recommended resolution: 1024+ (1024-1536 approx)
• Works best for people and character prompts
• Add notes about lighting, framing, or cinematic atmosphere
• Optional for speed: Stack the LightX2V 4steps LoRA (wan2.2_t2v_lightx2v_4steps_lora_v1.1_low_noise.safetensors, Org) alongside Aether IN-D

⚙️ Inference Settings

Standard usage:
• Steps: 20
• CFG: 3.5
• Mode: text-to-image (t2i), single frame output
• Model: Wan 2.2 14B – Low Noise
• Sampler: Euler
• Resolution: 1024+
• Output: 1 frame

Lightning/LightX2V mode:
• Steps: 8
• CFG: 1
• Stack LightX2V-LoRA before Aether IN-D

✅ Tip:
Guide the prompt with shot type, lighting, or mood. Think of it like describing a movie still.

📦 Trained using Ostris’s AI Toolkit

🙌 Thanks to @bblink787 and @masslevel for their beautiful showcase images!

https://tensor.art/u/635897593804625782