I hope that the native and complete functions of SkyReels-V2-DF-14B will be updated and added. I feel that this is the most comprehensive model. It natively supports start and end frames, video extension, text-to-video, image-to-video, long videos, and is also compatible with VACE.
Several functions updated later on https://github.com/SkyworkAI/SkyReels-V2 have not been added and implemented yet! I hope that the native and complete functions of SkyReels-V2-DF-14B will be updated and added. I feel that this is the most comprehensive model. It natively supports start and end frames, video extension, text-to-video, image-to-video, long videos, and is also compatible with VACE! I feel that this one model can replace all others......... Although currently, when it is used in conjunction with VACE, the effect of start and end frames is the best. Since the first frame image enters the latent space simultaneously, the colors can be made almost exactly the same. However, due to the additional loading and merging of the VACE model, the generation speed becomes slower and the VRAM usage is also higher. Since the original model already supports these functions without the need to rely on VACE, I hope you can add these native functions!
I hope that the native and complete functions of SkyReels-V2-DF-14B will be updated and added. I feel that this is the most comprehensive model. It natively supports start and end frames, video extension, text-to-video, image-to-video, long videos, and is also compatible with VACE.
Several functions updated later on https://github.com/SkyworkAI/SkyReels-V2 have not been added and implemented yet! I hope that the native and complete functions of SkyReels-V2-DF-14B will be updated and added. I feel that this is the most comprehensive model. It natively supports start and end frames, video extension, text-to-video, image-to-video, long videos, and is also compatible with VACE! I feel that this one model can replace all others......... Although currently, when it is used in conjunction with VACE, the effect of start and end frames is the best. Since the first frame image enters the latent space simultaneously, the colors can be made almost exactly the same. However, due to the additional loading and merging of the VACE model, the generation speed becomes slower and the VRAM usage is also higher. Since the original model already supports these functions without the need to rely on VACE, I hope you can add these native functions!