CheckPoint

Realistic Vision V6.0 B1

Animatediff_lightning_4step

1. Update to the latest version of IPAdapter.

2. Prepare at least one reference image. The workflow example uses 4 images, and the image size should be greater than 224x224.

3. The workflow uses Animatediff 4Step and Realistic Vision as checkpoints.

4. The RGB composition is based primarily on the first image, though RGB can also be omitted.

5. Adjust the BatchPrompt according to the content of your images. If the elements in your images do not match the prompt, the content will have minimal impact.

Additional Notes:

1. ClipVision Enhancer Batch provides finer image quality and sufficient relative detail even after upscaling.

2. The workflow includes upscaling and takes approximately 300 seconds to run on an RTX 4070.

3. This workflow is not a I2V (Image to Video). In reality, AnimateDiff + IPA cannot produce really I2V, but it can capture elements of the image to create a more detailed video.

4. The character video in the example was created using Kolors to produce 4 images, which were then input into this workflow for completion.

==

1、 更新最新版的IPAdapter。

2、 需要准备1张以上的图片做为参考图片，工作流范例为4张图片，图片长宽需大于224*224。

3、 工作流使用Animatediff 4Step与Realistic Vision为Checkpoint。

4、 RGB以第一张图片为主要参考色系构图，也可以不使用RGB。

5、 请依据你的图片内容调整BatchPrompt，如果你的图片中没有Prompt的要素，则内容不太会发生作用

补充说明：

1、ClipVision Enhancer Batch可提供较精细的画质，Upscale以后亦能提供足够的相对足够的细节。

2、工作流含Upscale，在RTX 4070上运行约需300秒。

3、此工作流并非真正的I2V，实际上AnimateDiff+IPA也无法做出真正的I2V，但可以抓到画面的要素产生细节较多的影片。

4、范例中的人物影片是使用Kolors制作4张图片后，再输入此工作流完成的。