IPAdapter在复杂流程中的表现有些让人失望，试试ComfyUI原生的区域条件采样节点吧

IPAdapter's performance in complex processes is somewhat disappointing. Try the ComfyUI native region conditional sampling node

添加了一个提前确定构图区域的方法，

Added a method to determine the composition area in advance,

v1.0 regional conditioning_2

1个主体+背景

1 subject + background

v2.0 regional conditioning_3

2个主体+背景

2 subject + background

v3.0 regional conditioning_5

4个主体+背景

4 subject + background

如果你需要，区域还能被分割成更多的部分，前提是你的电脑能支持直接生成更大的图片，否则，太多小区域只会造成混乱，

The area can be divided into more parts if you want, provided that your computer can directly generate a larger picture, otherwise, too many small areas will just cause confusion.

v4.0 LatenComposite_4

3个主体+背景

3 subject + background

有别于使用区域条件采样节点的方式，LatenComposite可以更加灵活的在前期初步绘制分割的区域的内容然后合并成一个整体，缺点是容易因为提示词污染而让画面出现不可控的东西

Unlike the use of regional conditional sampling nodes, LatenComposite can be more flexible in the early stage of the initial drawing of the content of the divided area and then merged into a whole. The disadvantage is that it is easy to make the screen uncontrollable due to the pollution of the prompt word

v5.0 DenseDiffusion

使用DenseDiffusioon节点的分区绘图功能，目前来看，效果还不错，区域间的污染比较少，融合也比较自然，缺点是，有时候尤其是写实类的风格看上去像是劣质的PS，不过在较小尺寸的图片中可以通过多次放大重采样来解决这个问题，总体来说还是很不错的。

Using the partition drawing function of the DenseDiffusioon node, the effect is not bad so far, the pollution between regions is relatively small, and the fusion is more natural. The disadvantage is that sometimes, especially the realistic style looks like inferior PS, but in smaller size images, this problem can be solved by multiple enlargement resampling, which is generally very good.

另外需要注意的是，当你想要绘制某种风格的图像时，需要在所有的文本编码框里填入同样的风格提示词，而不是仅仅在整体描述框内填入这些风格词

In addition, it should be noted that when you want to draw a certain style of image, you need to fill in the same style prompt words in all the text coding boxes, rather than just filling in the overall description box