---
title: AnimeGen-I2V-A14B-Lightning-bf16
canonical_url: "https://www.modelscope.cn/models/mlx-community/AnimeGen-I2V-A14B-Lightning-bf16"
md_url: "https://www.modelscope.cn/models/mlx-community/AnimeGen-I2V-A14B-Lightning-bf16.md"
repository: mlx-community/AnimeGen-I2V-A14B-Lightning-bf16
last_updated: 2026-07-14
license: apache-2.0
model_type:
  - i2v
base_model:
  - aidealab/AnimeGen-I2V
  - Wan-AI/Wan2.2-I2V-A14B
base_model_relation: finetune
parameters: 34.4B
tensor_type:
  - BF16
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 22
stars: 0
tags:
  - image-to-video
  - anime
  - mlx
  - wan2.2
---

# AnimeGen-I2V-A14B-Lightning-bf16

> AnimeGen-I2V-A14B-Lightning-bf16 - mlx-community 在 ModelScope 开源的模型。AnimeGen-I2V-A14B-Lightning-bf16 (MLX)

mlx-community/AnimeGen-I2V-A14B-Lightning-bf16 是 ModelScope 魔搭社区上的 34.4B 参数机器学习模型，采用 apache-2.0 许可，基于 aidealab/AnimeGen-I2V、Wan-AI/Wan2.2-I2V-A14B 构建。

- **Repository**: mlx-community/AnimeGen-I2V-A14B-Lightning-bf16
- **License**: apache-2.0
- **Parameters**: 34.4B
- **Base model**: aidealab/AnimeGen-I2V, Wan-AI/Wan2.2-I2V-A14B
- **Tags**: image-to-video, anime, mlx, wan2.2
- **Downloads**: 22
- **Stars**: 0
- **Last updated**: 2026-07-14

Source: https://www.modelscope.cn/models/mlx-community/AnimeGen-I2V-A14B-Lightning-bf16

---

# AnimeGen-I2V-A14B-Lightning-bf16 (MLX)

An **MLX (Apple Silicon)** conversion of [aidealab/AnimeGen-I2V](https://huggingface.co/aidealab/AnimeGen-I2V) — an
anime-specialized **image-to-video** fine-tune of Wan2.2-I2V-A14B (dual-expert MoE) — in **bf16**, shipped as a
pre-merged **Lightning** checkpoint for few-step, CFG-free inference.

Built for the Swift-MLX `wan-core` / Bernini-R runtime. For lighter footprint see the
[int4 variant](https://huggingface.co/mlx-community/AnimeGen-I2V-A14B-Lightning-int4).

## What's in here

| File | What |
|---|---|
| `high_noise_model.safetensors` | fine-tuned high-noise expert (bf16), Lightning-merged (**in_dim=36**) |
| `low_noise_model.safetensors` | fine-tuned low-noise expert (bf16), Lightning-merged |
| `vae.safetensors` | 16-channel Wan VAE |
| `t5_encoder.safetensors` | umT5-XXL text encoder |
| `config.json` | Wan2.2-I2V-A14B config (`model_type: i2v`, `in_dim: 36`) |

## Image conditioning (channel-concat, no CLIP)

The DiT input is `[noisy_latent(16), mask(4), image_latent(16)] = 36` channels. Encode a video whose first frame is
the conditioning image and the rest zeros through the 16-ch WanVAE → `z_video(16)`; build a first-frame temporal
`mask(4)`; `y = [mask, z_video](20)` is constant across steps and concatenated onto the 16-ch noisy latent each
denoise step. No separate image encoder (`image_dim` is null).

## Recommended inference recipe (from the AnimeGen-I2V model card)

- Sampler: **euler**, **CFG-free** (guidance 1.0)
- **4 steps**, **shift 3.0** (FlowMatchEuler)
- 832×480, frames `16·seconds + 1` (must be 4n+1), 16 fps
- Prompt prefix: `"Japanese anime style, "`

## How it was made

Experts converted 1:1 from `aidealab/AnimeGen-I2V` (original-Wan keys → `sanitize`; key set == the A14B 1095-key
contract). The [lightx2v](https://huggingface.co/lightx2v/Wan2.2-Lightning) `Wan2.2-I2V-A14B-4steps-lora-rank64-Seko-V1`
LoRA merged offline per-expert (`set_adapters([high,low],[1.0,1.0])`, kohya).

## License & attribution

Apache-2.0. Derived from `aidealab/AnimeGen-I2V` (AIdeaLab / GENIAC) + the lightx2v Lightning LoRA. Unofficial
community MLX conversion; not affiliated with AIdeaLab.