Alpha-channel video
According to the model card, Wan-Alpha generates transparent video by jointly learning RGB and alpha channels.
Open Source Model Profile · htdong
Wan-Alpha
Wan-Alpha is a text-to-video model from htdong for RGBA generation with an alpha channel. According to the model card, it jointly learns RGB and alpha and adapts Wan2.1-T2V-14B weights.
- Publisher
- htdong
- Task
- text-to-video
- Model type
- Unknown
- License
- apache-2.0
- Library
- Unknown
- Publication status
- Accepted · not indexed
Model overview
Wan-Alpha is published by htdong as a text-to-video model. Tags identify rgba, transparency, and base_model Wan-AI/Wan2.1-T2V-14B with finetune lineage. According to the model card, version 1.0 open-sources Wan2.1-14B-T2V-adapted weights and inference code for transparent video.
Recorded capabilities
Wan2.1 adaptation
Tags identify base_model Wan-AI/Wan2.1-T2V-14B; according to the model card, v1.0 adapts those weights with a VAE encoding alpha into RGB latent space.
Bilingual prompt note
According to the model card, prompts support both Chinese and English input with explicit transparency and shot-type wording.
Apache-2.0 record
Captured metadata records an apache-2.0 license with text-to-video pipeline tagging.
Use cases in the source record
- Text-to-video generation with transparency where the publisher says prompts should name transparent background, style, shot type, and subject.
- RGBA compositing tests involving semi-transparent objects, glowing effects, and fine details such as hair.
Limitations and unknowns
- No parameter count was extracted from this record.
- No architecture or library values were extracted from this record.
- No evaluation results were extracted from this record.
- Publisher superiority and rendering-quality claims come from the model card and were not independently verified.
- Provider state is historical snapshot data, not independently refreshed current availability.
Source and provenance
Source: htdong/Wan-Alpha
Captured: Unknown. Processed: 2026-09-07T19:35:56.950040+00:00.
Wan-Alpha Wan-Alpha: High-Quality Text-to-Video Generation with Alpha Channel Qualitative results of video generation using Wan-Alpha . Our model successfully generates various scenes with accurate and clearly rendered transparency. Notably, it can synthesize diverse semi-transparent objects, glowing effects, and fine-grained details such as hair. Abstract RGBA video generation, which includes an alpha channel to represent transparency, is gaining increasing attention across a wide range of applications. However, existing methods often neglect visual quality, limiting their practical usability. In this paper, we propose Wan-Alpha, a…
F001F002F003F004F005F006F007F008F010F011F013