text or image

[1910183821]

Hello, may I ask if your test code uses both images and text as conditions simultaneously? Why does my changing the text conditions not affect the generation effect at all?

[XinYu-Andy]

Hi, the current code is for image conditioning. Although we still use text during training, but the result might be biased to only rely on the image information. 李铭积 ***@***.***>于2025年5月5日 周一13:24写道：

…

*1910183821* created an issue (CVMI-Lab/TEXGen#19) <#19> Hello, may I ask if your test code uses both images and text as conditions simultaneously? Why does my changing the text conditions not affect the generation effect at all? — Reply to this email directly, view it on GitHub <#19>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOPNNMYOCW7HAIVDNDF6GX3243YZXAVCNFSM6AAAAAB4NYG4TWVHI2DSMVQWIX3LMV43ASLTON2WKOZTGAZTQNRTGE3TCMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>