[I2V] implement image-to-video for open-sora-plan v1.2

[oahzxl]

As requested in issue #217 and #179 , we want to implement image-to-video feature for open-sora-plan v1.2

Thanks @devtony10 for willing to contribute to this feature.

Open-Sora-Plan v1.2.0 provides a image-to-video feature here https://github.com/PKU-YuanGroup/Open-Sora-Plan/blob/main/opensora/sample/sample_inpaint.py .

The main difference between image-to-video and text-to-video is that the image-to-video introduce extra args for images here:
https://github.com/PKU-YuanGroup/Open-Sora-Plan/blob/294993ca78bf65dec1c3b6fb25541432c545eda9/opensora/sample/pipeline_inpaint.py#L621

I think we can add these new args and use if to determine whether to process images.

Can you help to add this feature to our Open-Sora-Plan pipeline? If so, can you first draft a simple plan (no need to write actual code) about code change?

[devtony10]

I'll get started on it pronto

[devtony10]

What of CogVideoX image to video?

[oahzxl]

i think that should be same. maybe we can do open-sora-plan video first

[oahzxl]

Hi, is there any update?