Ray Qwen2.1 编辑强化

turbo v1.3：

Remastered based on the official "Turbo" version released on 09/10.

基于09/10官方发布的turbo版本重新制作。

Now unlocked, though complex two-person poses in T2I still result in limb errors.

以解锁，T2I复杂双人姿势还是会有肢体错误。

v1：

Based on Qwen-Image-2.1, this model features enhanced editing capabilities; it integrates a custom editing-oriented LoKR (trained on image sets) and Realtime fine-tuning, specifically optimized for editing tasks.

基于Qwen image 2.1，编辑功能强化，融合了我自己使用组图训练的编辑型lokr+Realtime微调，主要是针对编辑功能做出调整。

Set up : Euler_A+ simple, 25-40 step, cfg1-2.5,

文生图-T2I：Euler_A + simple, 25-40步，cfg1-4，编辑-E2I：Euler_A + beta57, 15-40步，cfg1-4，

Resolution : 832x1216 for 1024x1536, upscaling : 1280x1920 or 1536x2240。

Female physical characteristics have been corrected; male characteristics may require a future version. This is not a "Turbo" version, as existing Turbo LoRAs suffer from significant quality loss. Excellent results can be achieved when paired with PE models and official Skill modules.

女性身体特征以修复，男性可能还需要等后续的版本。因现有的turbo lora质量损失较大，所以并不是turbo版本。配合PE模型和官方 Skill，能实现很不错的效果。

I reposted the VAE content; please let me know if this constitutes copyright infringement so I can remove it.

VAE是我转载的，如有侵权请留言，我会删除。