Skip to content
note · kiryu_mizuki · 图像视频创作分享·· 2026-10-14AI score8

ComfyUI beginner's log: building a workflow from scratch to generate images, with models and prompts the biggest hurdles

出てきた画像はなんとも……

AI briefing

An AI image generation beginner shares the experience of building a workflow from scratch in ComfyUI, using a "cooking in the kitchen" analogy for the relationship between workflows, nodes, models and LoRA. The author picked a model from the SD1.5 model list on Civitai; the first generation produced only a blurry image due to insufficient steps, and after adjustments it took about an hour to produce a fairly decent male image.

Open original note.com