>>12925
It's not F2LF, it's normal i2v process. You need to test lora and model compatibility to reach the best consistent output. Prompting is also important, I find abliterated LLMs could write better explicit prompt than GPTs, and it's cost effective than grok.