Get Pro$25 / month

Getting started

From download to first image

How to install an enzinoAI model, LoRA or workflow in ComfyUI and get the results you see on the product pages. About ten minutes the first time.

1. Install ComfyUI

All enzinoAI models and workflows are made for ComfyUI, the free node-based interface for image generation. If you do not have it yet, install the desktop version from comfy.org (Windows and macOS) or the portable version from GitHub.

Also install ComfyUI-Manager if your version does not include it: it adds missing nodes in one click, which most workflows need.

2. Put each file in its folder

ComfyUI looks for each kind of file in its own folder, inside ComfyUI/models/. A file in the wrong folder simply does not appear in the list.

You downloadedPut it in
A full checkpoint (one .safetensors with everything inside, as for SDXL models)models/checkpoints
A diffusion model that loads its text encoder and VAE separately (common on newer architectures)models/diffusion_models
A LoRA or a character LoRAmodels/loras
A text encoder or a VAE required by a modelmodels/text_encoders · models/vae
A workflow (.json)Anywhere: you open it from ComfyUI

Not sure which kind a model is? Open the workflow that comes with it: the loader node shows which folder it reads. If a loader is red after loading the workflow, the file is missing or in another folder.

After adding files, refresh ComfyUI (press R or restart it) so the new files appear in the lists.

3. Load the workflow

Drag the workflow's .json file onto the ComfyUI window. The whole graph appears, already wired.

Workflows are listed under Workflows, and many model pages mention the workflow they were tested with.

4. Use the recommended settings

Each model is tuned for its own sampler, number of steps and CFG. When they are documented, you find them on the product page, under Recommended settings and Technical details. Start with those values: they are what the example images were made with.

Different settings are not wrong, but if your images look very different from the examples, the settings are the first thing to check.

5. BF16 or FP8?

Some models come in several formats. They are the same model saved at different precisions:

BF16Full precisionThe reference quality. Larger file, needs more video memory (VRAM).
FP8LighterAbout half the size, runs on graphics cards with less VRAM, usually with a very small difference in detail.
?Which one?Enough VRAM: take BF16. Out-of-memory errors or slow generations: take FP8.

With enzinoAI Pro, every format of every model is included, so you can try both.

6. Add a character LoRA

  1. Put the LoRA file in models/loras.
  2. In the workflow, add a LoraLoader node (or use the one already there) between the model loader and the sampler, and pick your LoRA.
  3. Set the strength to the value given on the LoRA's page (often between 0.6 and 1.0), and write its trigger word in your prompt when it has one.

Want a character of your own? Create a character turns one reference image into a reusable character LoRA, delivered with a compatible workflow and a setup guide.

7. Common problems

The model does not appear in the loader's list

It is in another folder than the one this loader reads (see step 2), or ComfyUI was not refreshed. Press R or restart ComfyUI.

Some nodes are red when I open the workflow

They are custom nodes you do not have yet. Open Manager → Install Missing Custom Nodes, install them, then restart ComfyUI.

"Out of memory" or very slow generations

Use the FP8 version when one exists, lower the image resolution, or close other programs using the graphics card.

My images look very different from the examples

Check the sampler, steps and CFG against the recommended settings on the product page, and that the right model (and version) is selected in the loader.

My character does not look like my character

Check that the LoRA is connected, that its strength is not too low, and that its trigger word is in the prompt.

Where do I download what I bought?

Products bought on this site are in My purchases. Products and the Pro library bought on Ko-fi stay in your Ko-fi account. Character LoRAs you ordered are in My LoRAs.