Tata is an original, fully synthetic character LoRA for Krea 2.

🛠️ Made with LoRA Dataset Studio — the open-source tool I use to build and train all my character LoRAs (dataset curation, auto-captioning, face-similarity scoring, training, checkpoint testing, all in one UI). Read the write-up / tutorial.

💬 Join my Discord: discord.gg/wrAWFTPJe

This LoRA is face-centered: it locks Tata's facial identity with strong consistency from shot to shot, while leaving body, silhouette, hair style and outfit completely free to your prompt. Change the hair color, the setting, the framing - the face stays Tata.

Trigger word: tata_cv (put it at the start of your prompt).

Training: 3200 steps / 68 epochs on Krea 2 Raw (final checkpoint). Works great on the Krea 2 Turbo base too - all example images were generated that way, then finished with my Improve Detail edit workflow (also on my profile).

Recommended settings: strength 1.0 when used alone.

Pro tip - blend characters to design new faces: it is really interesting to combine several of my character LoRAs together at low strength, around 0.4 to 0.45 each. The identities merge into brand new, gorgeous faces that keep the photorealistic consistency of each LoRA. A great way to create your own original character from existing ones.

Tata dataset pack on Patreon

Get 32 images, 32 matching .txt prompts/captions and the available saved training parameters.

Current dataset edition; images and captions may differ from this older LoRA training run. The only saved training parameter supplied is rank 48; a complete training configuration is not included.

Get the Tata dataset pack - €9 as a one-time purchase. See the pack page for full contents and membership access.