A workflow that takes hand-drawn sketchbook art and converts it through a LoRA-trained ML model into stylized 3D characters — delivered straight to an interactive web viewer.
2026
Artist
Developer
UX/UI
ML Pipeline
3D Generation
ComfyUI / Meshy
Web Viewer
The sketchbook has always been where the ideas start. This pipeline connects that physical starting point to a fully realized 3D character — using a custom LoRA trained on my illustration style, ComfyUI for image generation, Meshy for 3D mesh creation, and a GitHub-hosted GLB viewer as the delivery target.
A character prompt is sent via Telegram. The LoRA + Juggernaut-XL model generates a style-consistent reference image. That image is passed to Meshy's image-to-3D API, producing a textured GLB. The model is committed to GitHub and the viewer link is returned — all from a single message.
The goal was to make the gap between a sketch and a portfolio-ready 3D piece as small as possible. Everything from prompt to published viewer link runs without leaving the phone.
Generate a 3D model from my sketchbook character 🎨
LIVE DEMO
A workflow that takes hand-drawn sketchbook art and converts it through a LoRA-trained ML model into stylized 3D characters — delivered straight to an interactive web viewer.
2026
Artist
Developer
UX/UI
ML Pipeline
3D Generation
ComfyUI / Meshy
Web Viewer
The sketchbook has always been where the ideas start. This pipeline connects that physical starting point to a fully realized 3D character — using a custom LoRA trained on my illustration style, ComfyUI for image generation, Meshy for 3D mesh creation, and a GitHub-hosted GLB viewer as the delivery target.
A character prompt is sent via Telegram. The LoRA + Juggernaut-XL model generates a style-consistent reference image. That image is passed to Meshy's image-to-3D API, producing a textured GLB. The model is committed to GitHub and the viewer link is returned — all from a single message.
The goal was to make the gap between a sketch and a portfolio-ready 3D piece as small as possible. Everything from prompt to published viewer link runs without leaving the phone.
Generate a 3D model from my sketchbook character 🎨
LIVE DEMO
A workflow that takes hand-drawn sketchbook art and converts it through a LoRA-trained ML model into stylized 3D characters — delivered straight to an interactive web viewer.
2026
Artist
Developer
UX/UI
ML Pipeline
3D Generation
ComfyUI / Meshy
Web Viewer
The sketchbook has always been where the ideas start. This pipeline connects that physical starting point to a fully realized 3D character — using a custom LoRA trained on my illustration style, ComfyUI for image generation, Meshy for 3D mesh creation, and a GitHub-hosted GLB viewer as the delivery target.
A character prompt is sent via Telegram. The LoRA + Juggernaut-XL model generates a style-consistent reference image. That image is passed to Meshy's image-to-3D API, producing a textured GLB. The model is committed to GitHub and the viewer link is returned — all from a single message.
The goal was to make the gap between a sketch and a portfolio-ready 3D piece as small as possible. Everything from prompt to published viewer link runs without leaving the phone.
Generate a 3D model from my sketchbook character 🎨
LIVE DEMO