[enhancement]: Don't require T5-XXL text encoder model to generate images in Canvas mode

[eeditor1055]

Is there an existing issue for this?

• I have searched the existing issues

Contact Details

No response

What should this feature add?

I have tried to use Anima model. Added it and Qwen text encoder. That's all that is needed to launch it and for it to work correctly. Then I went to Canvas menu. The requirement of T5-XXL makes it impossible to launch it at all. Downloading and loading this T5 model makes the great use of VRAM and RAM and is not suitable for work on medium machines (12GB VRAM + 32GB RAM). Thanks for understanding.

Alternatives

No response

Additional Content

No response

[Pfannkuchensack]

On Which version do you tried this?
There is the Problem that we need a little thing from the T5 Repo, but that was fixed in the new version.

[eeditor1055]

6.13.0 - just updated to it today. Launcher (v1.8.1) says that I can update to 6.13.5.rc1, but I will wait for stable release. Glad to hear that it is in the process of being fixed. Thank you.

[Pfannkuchensack]

#9244
the RC should fix that