How to delete image margin from Custom Diffusion generated images?

[daeunni]

Hi, I generated my personalized images via Custom Diffusion.
However, how can I delete these border margins from them?

This was my training shell.

  CUDA_VISIBLE_DEVICES=0,1,2,3,4 accelerate launch --config_file /d1/daeun/diffusers/configs/gpu1.yaml train_custom_diffusion.py\
    --pretrained_model_name_or_path=$MODEL_NAME  \
    --instance_data_dir=$INSTANCE_DIR \
    --output_dir=$OUTPUT_DIR \
    --class_data_dir="/d1/daeun/diffusers/cyw_data/prior_class/$class" \
    --class_prompt="$class" \
    --with_prior_preservation \
    --prior_loss_weight=1.0 \
    --num_class_images=200 \
    --instance_prompt="photo of a ${tokens[$i-1]} $class"  \
    --modifier_token "${tokens[$i-1]}" \
    --max_train_steps=$step \
    --resume_from_checkpoint="latest" \
    --max_train_steps=$(($step * $i)) \
    --checkpointing_steps=$step \
    --resolution=512  \
    --train_batch_size=1  \
    --learning_rate=1e-5  \
    --hflip  \
    --enable_xformers_memory_efficient_attention \
    --validation_prompt="${tokens[$i-1]} $class sitting in a bucket" \
    --no_safe_serialization \
    --num_validation_images=5 \

[sayakpaul]

Maybe these are stemming from the input images? Cc: @nupurkmr9