Skip to content

sat版本不支持i2v吗 #801

Description

@nxjiang-jnx

如题,sat/sample_video.py 中:
model_config.network_config.params.in_channels 被写死成 16(configs/cogvideox_2b.yaml 第 29-34 行),没有预留额外通道;DynamicCFG 里确实支持 cond["concat"],但一旦我们把图像 latent 填进去,就会出现:expected input[...] to have 16 channels, but got 32,因为模型权重没有对应的额外卷积核。

因此,SAT 版本目前根本没有真正的 i2v 分支?代码虽然把图片编码成 latent(concat_images),但一直没有和噪声 latent 合并,变相相当于纯 t2v,是这样吗

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions