WAN2.2 ULTIMA 14B T2V & I2V — Low VRAM FP8 Edition

📝 Overview

WAN2.2 ULTIMA is a low-VRAM optimized hybrid of Wan2.2 T2V High & Low Noise with carefully tuned LoRAs, built to run efficiently in FP8 while preserving cinematic visuals, motion stability, and Lightning-fast 4-step convergence.

This model has been tested across six different builds before release, resulting in a stable, refined version that delivers smooth motion without sacrificing quality.

⚡ Core Setup

• Base: Wan2.2 T2V High & Low Noise (14B, FP8/FP16 friendly)

• LoRAs Applied:

  • Wan2.2-Lightning_T2V-v1.1-A14B-4steps-lora_HIGH_fp16 → (Strength: 0.45)

  • Wan2.2-Lightning_T2V-v1.1-A14B-4steps-lora_LOW_fp16 → (Strength: 0.45)

    • Enables true 4-step speed with crisp detail.

  • Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1.1-HIGH_NOISE → (Strength: 0.20)

  • Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1.1-LOW_NOISE → (Strength: 0.20)

    • Adds motion energy and prevents the “slow-motion look.”

  • Wan2.2-Fun-A14B-InP-high-noise-MPS → (Strength: 0.15)

  • Wan2.2-Fun-A14B-InP-low-noise-MPS → (Strength: 0.15)

    • Enhances flow & expressiveness in frame-to-frame motion.

• Base: Wan2.2 I2V High & Low Noise (14B, FP8/FP16 friendly)

• LoRAs Applied:

  • Wan2.2-Lightning_I2V-A14B-4steps-lora_HIGH_fp16 → (Strength: 0.45)

  • Wan2.2-Lightning_I2V-A14B-4steps-lora_LOW_fp16 → (Strength: 0.45)

    • Enables true 4-step speed with crisp detail.

  • Wan2.2-I2V-A14B-4steps-lora-rank64-Seko-V1_HIGH → (Strength: 0.35)

  • Wan2.2-I2V-A14B-4steps-lora-rank64-Seko-V1_LOW → (Strength: 0.35)

  • Adds motion energy and prevents the “slow-motion look.”

🔧 Key Improvements

✅ Low VRAM optimized — works under 16 GB VRAM with fp8_e4m3fn precision.
✅ Fast rendering — 4-step Lightning efficiency maintained.
✅ Fluid motion — balanced LoRA blend avoids the usual Lightning “pose lock” and slow-motion artifacts.
✅ Cinematic visuals — detailed, consistent frames without over-cooked sharpness or contrast.

🛠 Workflow Notes

• WAN2.2 Native workflow → Runs well, fast, and efficient.

• WAN2.2 Video Wrapper workflow → Produces better prompt alignment and more realistic visuals, especially when combined with motion-oriented models aka High and Low noise Vace (without using the wanvace2video)

• Recommended if you’re pushing for cinematic realism and stronger adherence to storytelling prompts

🎚 Recommended KSampler Setup for T2V

Dual KSampler chain tested for WAN2.2 ULTIMA:

KSampler 1 (Noise Init)

• steps: 8 ( step: 10 is better ofcouse but takes lil bit longer)

• cfg: 1.5

• sampler_name: euler

• scheduler: simple

• add_noise: enable

• control before generate: randomize

• start_at_step: 0

• end_at_step: 4

• return_with_leftover_noise: enable

KSampler 2 (Refinement)

• steps: 8 ( step: 10 is better ofcouse but takes lil bit longer)

• cfg: 1.5

• sampler_name: uni_pc_bh2

• scheduler: normal

• add_noise: disable

• control before generate: fixed

• start_at_step: 4

• end_at_step: 10000

• return_with_leftover_noise: enable (You can disable this but i prefer keeping it enabled)

🎚 Recommended KSampler Setup for I2V

See Image which contains the I2V workflow.

🚀 Wan2.2 ULTIMA Experiments for Wanvideo2Image

I’ve been testing the Wan2.2 ULTIMA 14B F8 Low & High Noise models (T2V + I2V) inside Wanvideo2Image, and the results are 🔥.

Important:
👉 Ksampler 1 should always be Wan2.2 ULTIMA I2V 14B F8 CLEAN HIGH (uploaded here).

• This model has no baked-in LoRAs, making it perfect for the “foundation pass.”

• Keep CFG ~3–4 here for smooth motion + strong prompt response.

After that, you can stack/mix I2V High/Low and T2V High/Low models. Each combo changes how motion, texture, and coherence play out.

⚡ Example 1 — Stable Cinematic

• Ksampler 1: ULTIMA I2V CLEAN HIGH

• Ksampler 2: ULTIMA I2V HIGH

• Ksampler 3: ULTIMA I2V HIGH

• Ksampler 4: ULTIMA I2V LOW

💡 Result: Strong structure + cinematic motion. “High + High” builds detail and energy, while the final Low smooths transitions so it feels stable and film-like.

⚡ Example 2 — Dreamlike Flow

• Ksampler 1: ULTIMA I2V CLEAN HIGH

• Ksampler 2: ULTIMA I2V HIGH

• Ksampler 3: ULTIMA T2V LOW

• Ksampler 4: ULTIMA T2V LOW

💡 Result: Adds a flowing, dreamlike quality. The T2V LOW samplers inject organic fluidity, making results feel painterly and surreal. Perfect for fantasy/artistic prompts.

⚡ Example 3 — Balanced Hybrid

• Ksampler 1: ULTIMA I2V CLEAN HIGH

• Ksampler 2: ULTIMA I2V HIGH

• Ksampler 3: ULTIMA I2V LOW

• Ksampler 4: ULTIMA T2V LOW

💡 Result: Structure up front, looseness later. Great if you want characters to stay recognizable while backgrounds or effects evolve with more freedom.

⚡ Example 4 — Controlled Chaos

• Ksampler 1: ULTIMA I2V CLEAN HIGH

• Ksampler 2: ULTIMA T2V HIGH

• Ksampler 3: ULTIMA T2V LOW

• Ksampler 4: ULTIMA I2V LOW

💡 Result: Adds punchy, chaotic motion (T2V HIGH), softens it with T2V LOW, then stabilizes everything with I2V LOW. Feels like a dynamic camera shot that still keeps coherence.

🧪 Final Notes

• Always anchor with Ksampler 1 = CLEAN HIGH.

• Mix/match after that depending on whether you want stability, dreaminess, balance, or chaos.

• Tweak CFG between 3–6 on later samplers for subtle motion/style shifts.

⚠️ Disclaimer

• I do not claim authorship of Wan2.2 or the LoRAs used in this merge.

• Full credit to the original model and LoRA creators — please support them on their pages.

• This release is a merge + tuning effort to help the community run Wan2.2 with lower VRAM requirements.

💡 Notes

• Results may vary depending on workflow, sampler, fps settings, and hardware.

• This version is stable and production-ready — tested across multiple setups and iterations.

• Feedback and comparisons are welcome to further improve future merges.