JettHu
JettHu
> You are right, it works after randomly stopping half of the plugins. > > I will update the conclusion here after figuring out what the culprit is. Thanks!
ELLA is for comprehending dense prompts, encompassing multiple objects, detailed attributes, complex relationships, long-text alignment, etc. However, there is no guarantee that ELLA will bring better results in aesthetic scores....
@Akira13641 I've add a `ELLA Text Encode` node to automatically concat ella and clip condition.
可以,目前的 ELLA 和 IPAdapter 作用的方式相互不影响
I don't quite understand what you mean. Can you describe it more specifically?
I will evaluate the workload, but it is not a high priority. And welcome to contribute together.
Thanks for the feedback. The temporary solution is to use `TGate Apply(Deprecated)`. I'll try to fix it today.
换其他base model也这样吗,黑图有可能是精度问题
把 start_at 拉到 1 相当于 T-GATE 不起作用,可以从高往低降看看,我用的模型是 awpainting12,可以稳定出这个结果图