We reconstruct the data structure in the fast blending algorithm. Now you can use a sliding window in the fastest mode.
We improve the efficiency of text-guided video stylization. Now you only need half the time before.
New feature
We develop an extension for stable-diffusion-webui. This extension is an implementation of the fast blending algorithm in DiffSynth. We notice that this algorithm is very effective. Thus we develop this extension independently, making it easy to use.
Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
zjduan seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it.