Open-Sora: Democratizing Efficient Video Production for All
Python
Merge branch 'main' into dev/v1.2
Zangwei Zheng4f27cb51fc7b
diff --cc README.mdindex f5d3f0f,cf749a1..76b4e18--- a/README.md+++ b/README.md@@@ -110,11 -112,12 +110,11 @@@ see [here](/assets/texts/t2v_samples.tx<details><summary>View more</summary>+* [x] Training Video-VAE and adapt our model to new VAE.+* [x] Scaling model parameters and dataset size.+* [x] Incoporate a better scheduler (rectified flow).* [x] Evaluation pipeline.- * [x] Complete the data processing pipeline (including dense optical flow, aesthetics scores, text-image similarity, etc.).+ * [x] Complete the data processing pipeline (including dense optical flow, aesthetics scores, text-image similarity, etc.). See [the dataset](/docs/datasets.md) for more information* [x] Support image and video conditioning.* [x] Support variable aspect ratios, resolutions, durations.diff --cc opensora/datasets/datasets.pyindex a3fb169,0fb94c4..ddb28bb--- a/opensora/datasets/datasets.py+++ b/opensora/datasets/datasets.py@@@ -133,8 -129,9 +133,8 @@@ class VariableVideoTextDataset(VideoTexsample = self.data.iloc[index]path = sample["path"]- text = sample["text"]file_type = self.get_type(path)- ar = width / height+ ar = height / widthvideo_fps = 24 # default fpsif file_type == "video":