Drop an image and the vision model dissects it β appearance, outfit, personality, location, five boxes of Danbooru tags. The original Kyn Smith flow, now in liquid glass. Key stays in your browser.
These models are on your NanoGPT subscription plan β they use your weekly token allowance, not your cash balance.
Five boxes matching Soulkyn's tag editor. Every chip traces to the line it came from (hover = the reason). Suggested chips start unchecked β click to approve. Identity tags are yours β suggestion-only, never auto-filled. Edit any box, then use "Copy tags" to grab one box or save the character below.
Each box holds up to 18 tags (35-char limit). Tags copy as comma-separated values β paste straight into the matching SK box.
The sheet below is built from your approved tags and the vision analysis. Export it however you need.
Send to imagegen saves it there as a custom character β and its outfit & location land as custom presets, preselected and ready to gen.