Guides
Notes from running ComfyUI in production
What we learned setting up other people's workflows on rented GPUs. Useful whether you run them yourself or hand them to us.
How to run a ComfyUI workflow from Civitai
Why downloaded ComfyUI workflows fail: missing custom nodes, model files, paths and versions. A checklist to get a Civitai or OpenArt workflow running.
Read →
Can you use Civitai models commercially?
How to check whether a Civitai checkpoint or LoRA, Flux or WAN model can be used commercially, and what the base model license adds on top.
Read →
How to scale a ComfyUI workflow to thousands of outputs
Batch size, queue, API scripts, several GPUs, rented GPUs on Vast.ai: how to take a ComfyUI workflow from 10 test images to a production run.
Read →
How many images do you need to train a character LoRA?
How many photos a character, mascot or product LoRA needs, which angles and lighting to include, captions, and how to check the result.
Read →
WAN 2.2 image-to-video in production: how we make clips in batches
Our real WAN 2.2 image-to-video ComfyUI workflow: Lightning LoRA settings, two-clip chaining, RIFE, RTX upscale, start frames and batch runs.
Read →Contact
Discuss a project
Tell us about the template and the volume. We will get back to you via the contact you leave.