Browse categories
Explore
Fiverr Pro
English
$
USD
I will train a custom ai character lora for consistent faces in any scene
Meric Kartal
About this gig
Tired of your AI character looking different in every image? I train custom LoRA models so the same face, hair and style show up in any scene, outfit or lighting.
The cover image is a real example: one fictional character, trained once, generated in 8 completely different scenes.
HOW IT WORKS
Step 1: You send 15-30 clear images of your character or subject
Step 2: I prepare and caption the dataset
Step 3: I train your LoRA on a dedicated RTX 5090
Step 4: You get the .safetensors file, test images and prompt tips
WHAT I NEED
Images you own the rights to: fictional characters, original mascots, your own product or art style. No real people without written consent, no celebrities, no copyrighted characters.
The LoRA is trained for Z-Image Turbo (works in ComfyUI). I will tell you honestly if your dataset needs improvement before training. All work is AI-generated.
Delivery style preference
Please inform the freelancer of any preferences or concerns regarding the use of AI tools in the completion and/or delivery of your order.
AI Avatar Design disclosure
I understand that by acquiring AI Avatar Design services, I must abide by any and all disclaimer rules required by applicable laws and the platform(s) where this content will be posted. Read more about Fiverr’s AI Guidelines.
Get to know Meric Kartal
Meric Kartal
AI Character LoRA Training Creator
- FromTurkey
- Member sinceOct 2026
Languages
Turkish, English
I create consistent AI characters and train custom LoRA models, so the same face and style show up in any scene, outfit or lighting. I work on a high-end local workstation (RTX 5090), which lets me test many variations quickly and keep your files private. I'm new on Fiverr, so I put extra care into communication and clear previews before delivery. Message me before ordering and I'll tell you honestly if your idea and images will work.
FAQ
Which model is the LoRA made for?
Z-Image Turbo, which works in ComfyUI. Other base models only on request, after I confirm I can train them well.
How many images do I need to send?
15 to 30 clear images with varied angles, lighting and expressions. The better the dataset, the better the result. I will tell you if anything needs replacing before training.
Can I use the LoRA commercially?
You own the LoRA trained on your images. Commercial use of generated images depends on the base model license, so message me before ordering and I will share the license terms.
