Based on the results of ChatGPT's version of Studio Ghibli.
Description
FAQ
Comments (4)
Hello, I’d like to ask for your advice. I created an image inversion workflow in the Studio Ghibli style, which works wonderfully with your LoRA models. However, when I use these two LoRA models from Hugging Face, the results are terrible. I don’t understand why, despite all being Flux Dev LoRAs, the results are so contradictory. Are there differences between LoRAs? How can I determine if one is suitable for my needs?
https://openart.ai/workflows/bFS0ghvv1UkL5qTT2QI2
I'm not too sure.. I have stuck to FluxDEV checkpoint for all my trainings making it highly compatible with all associated checkpoints.. Have you tried experimenting with different samplers and schedulers ?
@RojerDodger In the Flow Edit workflow, there are only schedulers available, no samplers to choose from. I tested using the most common one, 'Simple,' and I adjusted parameters such as trigger words and CFG guidance during my testing. However, these two LoRA models still performed very poorly. I've raised an issue in the Easy Control GitHub repo, but I haven't received a response yet.
i tried it and its very nice an accurate , but its too anime like
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.