LoRA Pilot ep.3: character dataset prep with browser upload, image review and captions
no3us · reddit · 2026-10-12
The creator of LoRA Pilot (free and open source) published episode 3 of a tutorial series on preparing a character dataset: uploading an image archive via the browser, reviewing which views add useful information, and editing captions in TagPilot. All eight example images are kept — eight is just the example size, not a claimed training minimum. The first two videos cover a first session and camera prompting in ComfyUI, where the tracking take also changes the robot's action, discussed as a directing exercise. Recorded on a prepared RunPod workspace; setup from an empty pod is not covered.
More from Multimodal
- StreamRender-H3 renders interactive game worlds in real time with a 2-step diffusion model at 24 FPS — songhan_mit · 2026-10-12
- Dev builds a free AI filmmaking platform with support for H3 — RADIO02118 · 2026-10-12
- AI-Generated Short Film 'Made For Each Other' Tells a Robot Love Story — BotLordBotflix · 2026-10-12
- Two prompts in Claude turn FACEIT clips into a CS2 fragmovie with Serial Experiments Lain vibes — TAbrodi · 2026-10-12
- Open-Source Qalign Node Fixes Pixel Drift in Qwen Image 2.1 Edits — listopalafoto · 2026-10-12
- Open-Source Studio for H3 and Krea 2 Runs Fully Serverless on Modal — RADIO02118 · 2026-10-12