Yes, I found that the XL LoRAs were very difficult to get words from .... Even standard XL without LoRAs is still much more difficult to get accurate wording than the newer models available now. But then adding LoRAs... Unless the dataset also had creations with words in it (very few did, but I think someone made a LoRA for that, but it didn't work for the words I tried to get). Anyway yeah adding LoRAs only diminished the ability for 1.0 to produce text even more. Lol
<lora:Creepy cool XL:0.5> <lora:Red Rising:0.5> <lora:BigBeautifulBodies:0.5> SILHOUETTE of a plus size singer with a microphone. "50s Tunes with June" written in red on the album sleeve. Hyperdetailed hyper realistic 16k resolution
Yes, I found that the XL LoRAs were very difficult to get words from .... Even standard XL without LoRAs is still much more difficult to get accurate wording than the newer models available now. But then adding LoRAs... Unless the dataset also had creations with words in it (very few did, but I think someone made a LoRA for that, but it didn't work for the words I tried to get). Anyway yeah adding LoRAs only diminished the ability for 1.0 to produce text even more. Lol