Thank you :)

Yes, I can describe you my workflow and settings, although I'm sure and want to do better in the next version (for example I need to add another concept for the car brand logo, which is currently not never visible.. and probably for this reason it's a better approach LoHa instead of a LoRA)

Initially I prepared a dataset starting from high resolution images, 38 in total, cropped and resized with Photoshop to 768x768, where the car was always completely in the frame and with no other objects in front of the car, the background and the illumination instead more different are better is for flexibility
Then with Automatic1111, using v2-1_768-nonema-pruned.ckpt and vae-ft-mse-840000-ema-pruned.ckpt I had 3800 images generated at 768x768 resolution with the following prompt: "city car" (no negative prompt )
These were my Kohya settings:
Repeats: 100
{ "pretrained_model_name_or_path": "stabilityai/stable-diffusion-2-1", "v2": true, "v_parameterization": true, "logging_dir": "LoRA_training_data/drc city car v0.3\\log", "train_data_dir": "LoRA_training_data/drc city car v0.3\\img", "reg_data_dir": "LoRA_training_data/drc city car v0.3\\reg", "output_dir": "LoRA_training_data/drc city car v0.3\\model", "max_resolution": "768,768", "learning_rate": "0.0001", "lr_scheduler": "cosine", "lr_warmup": "10", "train_batch_size": 1, "epoch": 10, "save_every_n_epochs": 1, "mixed_precision": "fp16", "save_precision": "fp16", "seed": "1234", "num_cpu_threads_per_process": 2, "cache_latents": true, "caption_extension": ".txt", "enable_bucket": true, "gradient_checkpointing": false, "full_fp16": false, "no_token_padding": false, "stop_text_encoder_training": 0, "xformers": true, "save_model_as": "safetensors", "shuffle_caption": false, "save_state": false, "resume": "", "prior_loss_weight": 1.0, "text_encoder_lr": "5e-5", "unet_lr": "0.0001", "network_dim": 128, "lora_network_weights": "", "color_aug": false, "flip_aug": false, "clip_skip": "1", "gradient_accumulation_steps": 1.0, "mem_eff_attn": false, "output_name": "drccitycar_v0.3", "model_list": "stabilityai/stable-diffusion-2-1", "max_token_length": "75", "max_train_epochs": "", "max_data_loader_n_workers": "0", "network_alpha": 1, "training_comment": "drc city car", "keep_tokens": "0", "lr_scheduler_num_cycles": "", "lr_scheduler_power": "", "persistent_data_loader_workers": false, "bucket_no_upscale": true, "random_crop": false, "bucket_reso_steps": 64.0, "caption_dropout_every_n_epochs": 0.0, "caption_dropout_rate": 0, "optimizer": "AdamW8bit", "optimizer_args": "", "noise_offset": "", "LoRA_type": "Standard", "conv_dim": 1, "conv_alpha": 1, "sample_every_n_steps": 0, "sample_every_n_epochs": 0, "sample_sampler": "euler_a", "sample_prompts": "", "additional_parameters": "", "vae_batch_size": 0, "min_snr_gamma": 0, "down_lr_weight": "", "mid_lr_weight": "", "up_lr_weight": "", "block_lr_zero_threshold": "", "block_dims": "", "block_alphas": "", "conv_dims": "", "conv_alphas": "", "weighted_captions": false, "unit": 1, "save_every_n_steps": 0, "save_last_n_steps": 0, "save_last_n_steps_state": 0 }

@fuleshu personally I haven't tried it yet but one of the best guides I've found on the internet (https://rentry.org/59xed3#selecting-images) reports: "You may also want closeups of certain details. If you notice the AI is not giving a specific part enough definition, then a full-resolution image of that part might help. This could be useful for intricate decorations, accessories and such."