https://peer.madiator.cloud/videos/watch/e0af28a1-bf05-4885-8789-4ef73f0301fc Training your own LoRA usually means dealing with dozens of confusing settings, switching between separate apps, and losing GPU hours to mistakes. In this Runpod community session, community helper notrius walks through LoRA Pilot, his open-source workspace that puts dataset prep, LoRA training, and image/video generation into a single Runpod template. Community helper notrius spent 25 years in digital marketing before turning to Stable Diffusion to make consistent characters and products for ads without a production budget. He talks about why he built LoRA Pilot, what it takes to keep several AI apps working together, and the dataset mistakes that ruin most LoRAs, then gives a live demo. Stay for the Q&A on the future of character LoRAs, Runpod global volumes, and synthetic training data. What's in LoRA Pilot: Three training backends: Kohya SS, AI Toolkit, and diffusion-pipe, covering around 30 models, ComfyUI and InvokeAI for image and video generation, TrainPilot: pick a model, a dataset, and a quality level, and it sets the training parameters for you, TagPilot and dataset tools for captioning, tagging, finding duplicates, resizing, and cropping, ControlPilot dashboard: GPU and disk monitoring, service controls, logs, password protection, and scheduled shutdown so idle pods don't run all night, One shared model folder for every app, plus a built-in model catalog and downloader, LINKS LoRA Pilot: https://www.lorapilot.com/ GitHub: https://github.com/vavo/lora-pilot Docs: https://www.lorapilot.com/docs/ Deploy the template on Runpod: https://console.runpod.io/deploy?template=gg1utaykxa&ref=o3idfm0n Community Spotlight blog post: https://runpod.io/blog/community-spotlight-dataset-prep-training-and-inference-in-a-single-workspace-with-lora-pilot Join the Runpod Discord: https://discord.com/invite/runpod Reddit: https://www.reddit.com/r/RunPod/ CHAPTERS 0:00 Intro & welcome 1:16 Meet notrius: from marketing to Stable Diffusion 3:03 Why train LoRAs? Consistent characters across scenes 6:57 Why Kohya SS is hard for beginners 10:39 From personal toolkit to open-source project 11:56 What's inside LoRA Pilot 13:15 Checkpoints, base models & LoRAs explained 15:15 Training, inference & dataset tools 17:41 Building it with AI coding agents 20:29 Keeping a multi-app stack working 23:38 Storage, network volumes & global volumes 25:30 Common mistakes: datasets & captions 26:52 Publishing a Runpod template 29:19 Learn the basics first: SDXL & Flux 30:27 Live demo: ControlPilot dashboard 32:50 Scheduled shutdown for idle pods 34:25 Settings: passwords, Hugging Face token & Copilot 36:35 Bundled services & logs 38:11 Preparing a dataset 38:55 TrainPilot: starting a training run 40:56 Model catalog & downloads 43:32 ComfyUI video workflows 45:50 Q&A: Will character LoRAs stay standard? 47:51 Q&A: Who is LoRA Pilot for? 49:48 Q&A: What are global volumes? 51:47 Q&A: Sourcing training data & synthetic data 52:31 Wrap-up Tue, 20 Oct 2026 10:50:03 GMT https://validator.w3.org/feed/docs/rss2.html PeerTube - https://peer.madiator.cloud https://peer.madiator.cloud/lazy-static/avatars/e2927aea-9e0a-454c-9327-68ec87b4da11.png https://peer.madiator.cloud/videos/watch/e0af28a1-bf05-4885-8789-4ef73f0301fc All rights reserved, unless otherwise specified in the terms specified at https://peer.madiator.cloud/about and potential licenses granted by each content's rightholder.