repeat model train

[oztrkoguz]

I want to use this method for more than one image of the same person. Do I need to do lora training every time? When I try it, success decreases in images without lora training. Does training lora with multiple images solve this problem? If so, how can I do this?
Thanks

[Kevin-thu]

Please check out the multi_image folder. We have added basic support for multiple image inputs. Next we will also test training a single LoRA for multiple images. Stay tuned.

[oztrkoguz]

For multiple images, I ran main.py from the multi_image folder in my local.
images : []
Trace back (last call):
File "main.py", line 108, in
images[0].save(f"{args.output_path}/output.gif", save_all=True,
IndexError: list index out of range
I get an error, the image list is empty.
How can I solve the problem?
Thank you for your help

[YeeZ93]

The problem is that your image urls are not in the "img_paths". This should work:

‘’‘

def __call__(
        self,
        img_0=None,
        img_1=None,
        img_path_0=None,
        img_path_1=None,
        prompt_0="",
        prompt_1="",
        imgs=[],
        img_paths=None,
        prompts=None,
        save_lora_dir="./lora",
        load_lora_path_0=None,
        load_lora_path_1=None,
        load_lora_paths=None,
        lora_steps=200,
        lora_lr=2e-4,
        lora_rank=16,
        batch_size=1,
        height=512,
        width=512,
        num_inference_steps=50,
        num_actual_inference_steps=None,
        guidance_scale=1,
        attn_beta=0,
        lamd=0.6,
        use_lora=True,
        use_adain=True,
        use_reschedule=True,
        output_path = "./results",
        num_frames=50,
        fix_lora=None,
        progress=tqdm,
        unconditioning=None,
        neg_prompt=None,
        save_intermediates=False,
        **kwds):


    self.scheduler.set_timesteps(num_inference_steps)
    self.use_lora = use_lora
    self.use_adain = use_adain
    self.use_reschedule = use_reschedule
    self.output_path = output_path


    imgs = [Image.open(img_path).convert("RGB") for img_path in img_paths]
    assert len(prompts) == len(imgs)

’‘’