雰囲気だけのLoRAです。後ろ姿がうまく学習出来ていません。
後ろから見た際に胸部の装飾が背面に現れてしまう場合は、プロンプトに「スーツの背面は無地の黒色で、肩当てはなく、胸部アーマーは胸の範囲のみを覆っている」「ウィングマンの背面は装飾が一切なく、完全な無地の黒です。」といった内容を入れたら改善するかもです。翼もプロンプトで出せますが形状はランダムだと思います。形状を固定できるのか確認してないので分かりません。バイクのウイナアII は適当なプロンプトで似たような物が出るかもです。
This LoRA is just for capturing the vibe.
The model has not learned the rear view well.
If the chest ornamentation shows up on the back when viewed from behind, you might be able to fix this by including details in your prompt such as: "The back of the suit is plain black with no shoulder pads, and the chest armor covers only the chest area," or "The back of the Wingman has no ornamentation whatsoever and is completely plain black."
You can generate wings using prompts, but I believe their shape is random. I haven't checked if it's possible to lock in a specific shape, so I'm not sure.You might be able to generate something similar for the "Winaldo II" motorcycle by using the right prompt.
Does it generate the same face every time, or a random one? Ultimately, this is a LoRA created by an amateur, so there is an element of chance involved, and results can vary depending on the environment and node settings. It may also vary depending on the model used.Please understand that I am doing this purely as a hobby, so there are limits to my technical expertise.
Model Used : minimax_h3_fl2va_pruned_int8_convrot.safetensors
Text encoder : qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
vae : minimax_h3_audio_vae_fp32.safetensors minimax_h3_video_vae_fp16.safetensors
Minimax H3 Turbo Loras : minimax_h3_turbo_v4_step600.safetensors
Here is the workflow.
https://note.com/api/v2/attachments/download/6a07a5c09fdf615456cde2410024a8db
