Type | |
Stats | 127 74 |
Reviews | (8) |
Published | Jan 4, 2024 |
Base Model | |
Training | Steps: 2,000 Epochs: 2 |
Usage Tips | Clip Skip: 2 Strength: 0.8 |
Trigger Words | HTA |
Hash | AutoV2 0FD431B8F7 |
这是为做动物变装过程动画做的LORA, SD1.5用的。
有ConfyUI工作流可供直接使用,
https://openart.ai/workflows/jTDvmeFCE6v1aV5aZq6T
工作流未用到animateDiff,故,初次生成的连续图片组合成动画的效果并没有animateDiff的好,我还没有找到可以直接使用animateDiff控制lora权重做动画的方法,所以只能用现有的方法,先合成单一图片,再组合。可以考虑用animateDiff重绘以增加连贯性,但不可避免的会损失一些细节。
如果您知道用animateDiff控制单个或多个LORA权重的方法,希望能给我留言,我会非常感谢您!
This is LORA designed for animation of animal cross dressing process, used in SD1.5.
There is a ConfyUI workflow available for direct use,
https://openart.ai/workflows/jTDvmeFCE6v1aV5aZq6T
which does not use animateDiff. Therefore, the animation effect of combining consecutive images generated for the first time is not as good as animateDiff. I have not found a method to directly use animateDiff to control the weight of Lora for animation, so I can only use the existing methods to first synthesize a single image and then combine it. You can consider using animateDiff to redraw to increase coherence, but inevitably losing some details.
If you know how to use animateDiff to control single or multiple LORA weights, please leave me a message. I would greatly appreciate it!