Now you can improve text results with just swapping one file.
Video guide here: https://youtu.be/gCzJI4sKr1I
Download text encoder: https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/resolve/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors?download=true
Comfy: Place in /models/clip/
Forge: Place in /models/text-encoder/
Load text encoder by replacing clip_l in DualClip Loader
Or select it in VAE/Text-encoder for Forge
That's it! Generate away and get better text results (mostly).

Adriana
2024-11-08 14:29:30 +0000 UTCSebastian Kamph
2024-09-18 09:42:45 +0000 UTCSam Gómez Visual
2024-09-12 18:08:51 +0000 UTC