What I would be interested in is how do I filter out certain features? If I want to have a specific face, or just the hairstyle, how do I prevent, for example, that the face is not taken over when the focus should be on the hairstyle?
Let's imagine we're going to create different hairstyles for a specific woman. First, gather some sample photos that showcase various hairstyles. Then, swap the faces in all the photos to match the same person. Afterward, you can train a LoRA model. This way, all the photos you generate using the LoRA will have the same face while featuring different hairstyles.
Thanks, so let’s say you want to train a Lora with images of a guy’s beard. How do you use the Lora afterwards if you just want the beard, but not the guy?
Train the LoRA on images of different beards without including other features tied too closely to the specific person. Aim for diversity in the training dataset (e.g., various beard shapes, styles, and colors) to make it versatile.
What I would be interested in is how do I filter out certain features?
If I want to have a specific face, or just the hairstyle, how do I prevent, for example, that the face is not taken over when the focus should be on the hairstyle?
Let's imagine we're going to create different hairstyles for a specific woman. First, gather some sample photos that showcase various hairstyles. Then, swap the faces in all the photos to match the same person. Afterward, you can train a LoRA model. This way, all the photos you generate using the LoRA will have the same face while featuring different hairstyles.
You are welcome to join our new forum to discuss this matter. Visit us at: cyberscap.com.
@ Thank you for your answer
Can you show how to use this in comfyui?
Sure ♥️ I previously posted a video about comfyui.
how to generate consistent character using tensor art plz plz help
Hi join TensorArt discord ♥️
Is there a model that is specifically categorized as "private" or for personal use only?
If you have a paid subscription you can keep your models private ♥️
@@AkalankaEkanayake If you train a model, can you download and delete it after that?
bokka
@AkalankaEkanayake If you train a model, can you download and delete it after that?
Yes, you can.
So we just need to load images that we created already??? Thats all???
Yes including the trigger word
Do i have to upload 1000 image?
No, But many as you can. ❤️
Are those LORAs public by default? Can one make them private (with a free account)?
No, you can't. You need to pay so that you can be able to host private loras.
😮👍
Thanks, so let’s say you want to train a Lora with images of a guy’s beard. How do you use the Lora afterwards if you just want the beard, but not the guy?
Train the LoRA on images of different beards without including other features tied too closely to the specific person. Aim for diversity in the training dataset (e.g., various beard shapes, styles, and colors) to make it versatile.
I don't think that FLUX.1 Dev model in TensorArt
👀♥️
This is not free yall