CivArchive
    ← All articles
    Published February 15, 2025by nova_sterling

    The Tools I Use

    476 views10 reactions3 comments on CivitAI19 collected
    tool guidetools

    The Tools I Use

     

    OS: Windows with Docker/WSL

    • Main Tool: Pinokio (https://pinokio.computer/) – This is a dead simple tool for installing, maintaining and using the various tools needed for LoRa creation and Image/Video creation. Finding this tool changed me from having to maintain my environment(s) to using ComfyUI, Fluxgym and Forge as appliances.

     

    LoRa Generation

    • Fluxgym running in Pinokio (https://pinokio.computer/)

      • Desktop computer with two RTX 3060s (12 GB VRAM each)

      • Creates a LoRa in about 5 hours

    Image Generation

    • ComfyUI running in Pinokio (https://pinokio.computer/)

      • Laptop with mobile RTX4090 (16 GB VRAM)

      • GGUF Flux1.D workflow

        • Breaks out 18 different prompts in batches of 2

        • 36 images created in 15 minutes

      • GGUF Flux1.D LoRa tester workflow (NEW)

        • Workflow tests LoRa strength from 0.6 to 1.2 and outputs an image of the same prompt and seed at each strength labeled

    • Forge running in Pinokio (Brand NEW)

      • Learned some LoRas don’t work well in Forge, so I installed it and will test going forward

     

    Other Tools  

    • Diffusion Toolkit – Image Library and Cataloguer (https://github.com/RupertAvery/DiffusionToolkit)

      • Keeps track of all the generated images. If the metadata about generation is present, that is displayed. Allows for ratings and setting of NSFW content.

    • MagicQuill – Inpainting and image cleanup (https://magicquill.art/demo)

      • Runs in Docker container

      • Runs on Huggingface, but downloadable to PC w/ Docker

      • Does clean-up of bad anatomy and will add clothes to NSFW images

    • Wekan – Local, Open-source Kanban board (https://wekan.github.io/)

      • Keeps track of Loras that I want to build, built, generated images and uploaded \

    • Blinko – Local, Open-source AI powered notebook (https://github.com/blinko-space/blinko)

      • Keep track of ComfyUI workflows, prompts, sample images other notes. It’s multimodel (images, text, video, audio)

      • Can generate new content based on the content in the notebook

    • Files – File Manager replacement (https://files.community/)

      • Multipane file manager with advanced rename and preview

      • Got to keep those files organized. I have quite the file structure, kind of overdone but I'm obsessed that way.