Enzino Dataset Studio V1
- Type: ComfyUI workflow
- Base model: Other
- Price: $29
One photo in. A whole character LoRA dataset out. Dataset Studio builds a complete, captioned, likeness-checked training dataset inside a single ComfyUI node. Drop a reference photo, load the 25-shot preset, press Run. No loader nodes, no wires, no manual sorting. 🎯 TWO ENGINES - Krea 2 Identity Edit (default): the most faithful identity, 6 steps at 2 MP with IntoRealism Krea 2 Flash - MiniMax H3: up to 9 references, true single-frame stills, optional Krea 2 texture pass ⚡ WHAT'S INSIDE - 25 pro prompts for character LoRAs: front, profile, three-quarter, full body, expressions, light, motion. Captions included - Face isolation: only the face of your reference is used, outfits and places follow each prompt - Likeness score on every image (0–100). Low scores are re-rolled automatically - Face detailer with an identity guard that reverts any drift - LoRA stack, upscale + refine - Review grid: reject, restore, edit captions in place - One-click ZIP export (1024 / 1536 px, PNG or JPG, 001.png + 001.txt) ready for kohya, ai-toolkit, OneTrainer - Resume: only new or edited cards are rendered again - EN / FR interface ➕ BONUS: IMAGE STUDIO A second node for single images of your character from reference photos. Same engines, LoRAs, face detailer, upscale and likeness score. 📦 INCLUDED - Both nodes + 2 example workflows - Full HTML documentation (install, models, settings, troubleshooting) - Free 1.x updates 🖥️ REQUIREMENTS - A recent ComfyUI (tested on 0.41) - Krea 2 and/or MiniMax H3 models (not included, download links in the docs) - A GPU that already runs Krea 2 or H3 in ComfyUI ⚖️ LICENSES Models are not included and keep their own licenses. The recommended krea2_identity_edit_v2 LoRA (by conradlocke / lbouaraba) is free for personal and non-commercial use.
More tools
Enzino Dataset Studio V1
About this creation
Everything you need to know.
One photo in. A whole character LoRA dataset out. Dataset Studio builds a complete, captioned, likeness-checked training dataset inside a single ComfyUI node. Drop a reference photo, load the 25-shot preset, press Run. No loader nodes, no wires, no manual sorting. 🎯 TWO ENGINES - Krea 2 Identity Edit (default): the most faithful identity, 6 steps at 2 MP with IntoRealism Krea 2 Flash - MiniMax H3: up to 9 references, true single-frame stills, optional Krea 2 texture pass ⚡ WHAT'S INSIDE - 25 pro prompts for character LoRAs: front, profile, three-quarter, full body, expressions, light, motion. Captions included - Face isolation: only the face of your reference is used, outfits and places follow each prompt - Likeness score on every image (0–100). Low scores are re-rolled automatically - Face detailer with an identity guard that reverts any drift - LoRA stack, upscale + refine - Review grid: reject, restore, edit captions in place - One-click ZIP export (1024 / 1536 px, PNG or JPG, 001.png + 001.txt) ready for kohya, ai-toolkit, OneTrainer - Resume: only new or edited cards are rendered again - EN / FR interface ➕ BONUS: IMAGE STUDIO A second node for single images of your character from reference photos. Same engines, LoRAs, face detailer, upscale and likeness score. 📦 INCLUDED - Both nodes + 2 example workflows - Full HTML documentation (install, models, settings, troubleshooting) - Free 1.x updates 🖥️ REQUIREMENTS - A recent ComfyUI (tested on 0.41) - Krea 2 and/or MiniMax H3 models (not included, download links in the docs) - A GPU that already runs Krea 2 or H3 in ComfyUI ⚖️ LICENSES Models are not included and keep their own licenses. The recommended krea2_identity_edit_v2 LoRA (by conradlocke / lbouaraba) is free for personal and non-commercial use.
Product not found
This creation may have been moved or removed.
Return to the library