This repository serves as a resource pool for you Game Design practicum.
Given the extremely scattered, multi-flavored, usecase-oriented and undocumented nature of free GenAI models, as well as poor track of obsolete resources, the Chaotic World of Open Source greets a newbie in 2026 with an unnecessarily steep learning curve.
As someone who tries to get into image generation without voluntarily selling your soul to token economy, you might watch a couple of youtube vids that suggest trying out, for example, the SotA model Qwen Image Edit, but once you search for it on the Internet, you may become accidentally intimidated by such resource pages:
image
"What's 2511? Wtf is Q5_0, what in tarnation is a .GGUF?! I don't have time for this shit"
Indeed, one has to swim around for a while in the cesspit of GenAI not to get lost in the whimsical naming and format conventions. To make your super busy student lives just a little bit easier, I gathered here all the models necessary to successfully run the workflows created for our Game Design lectures, using our modest 8GB VRAM rigs.
I'll try to keep the amount of text as little as possible, hence you'll need to use your logic extensively. Still, the first workflow might include a more or less detailed explanation of what does what. I think that for starters I'll embed the guides into workflows to make stuff more portable, but that might change in the future.
Now I'll provide instructions on how to set up our work environment for the lectures.
We'll need to additionally install ComfyUI Manager to ease our work with models and custom nodes. While you can find an extensive tutorial here: https://docs.comfy.org/manager/install, let me just provide you with tl;dr steps for this:
open the command line inside your ComfyUI_windows_portable_nvidia folder and paste the following command: .\python_embeded\python.exe -m pip install -r ComfyUI\manager_requirements.txt. This will install the requirements for our manager
open the run_nvidia_gpu.bat file with the notepad and paste --enable-manager --enable-manager-legacy-ui after --windows-standalone-build. Save and close the file.
Next thing we'll do is pack up ComfyUI with models necessary to run our workflows. Currently all the resources are located here: https://huggingface.co/braindanxxx/utm_dj_resources/tree/main/models. It contains three different, SotA models that are just simply amazing at creating high quality media for free. You can find the nutshell descriptions of these models further in this document, but for now you need to copy the items just like I tell you:
on your PC, go to path/to_your/ComfyUI/models
assuming you have already downloaded our resource repository, open it and navigate to the model directories. These are located in:
models/flux models/qwen models/sdxl
Inside each of these folders you will find several subfolders containing the files required to run each model.
Next, go to your ComfyUI installation folder and open its models directory. Inside it you will find folders with the same names (like for example checkpoints, diffusion_models, vae). Copy the contents from the corresponding folders in the repository you downloaded from Hugging Face and paste them into the matching folders inside your ComfyUI models directory. Make sure the folder names stay the same so ComfyUI can correctly detect the models. If for some reason a folder doesn't exist in your local installation, don't worry. Simply paste the entire file-containing folder to ComfyUI/models then.
Now everything that's left is to launch ComfyUI. For that, simply double-click run_nvidia_gpu.bat. After some time, a new browser tab should open displaying your workspace. Either press Ctrl+O to open a new workflow, or drag and drop one of the workflows provided in this repo onto your workspace - it will be loaded automatically
If you're getting the "Missing custom node" error, don't worry! Assuming you've done correctly all the previous points, you only need to accept the installation of all the missing nodes, which will in turn require a restart. Afterwards, test out the workflow by clicking "Run" or pressing Ctrl+Enter.
