What is a “NOA” workflow?

NOA (Native OpenArt) is a term I made up to denote that the workflow was built natively with OpenArt‘s ComfyUI web interface.  So it should run with no need to hunt down and install checkpoints, LoRAs, VAEs etc. that won’t be spun up automatically when choosing the “Run on Cloud” button.

There are so many great cloud workflows on OpenArt, but some take a bit of time and effort to actually get running---which is great!  The more worflows shared the better.  

I just want mine to be a ready-to-go environment to learn with.  

I consider these to be beginner/learning workflows.  They may tend to get a little bit messy or noodly, but this is so that the workflow can be easily inspected, to make it easier to follow along with a workflow as it is running.  I find it easiest to learn when it is easy to ‘look under the hood’ of a workflow to see how everything interacts.  
These workflows don’t do anything that can’t be done better than other workflows already out there– but hopefully make things easier to follow and understand as folks like myself learn.

I am just beginning to develop this idea for workflows, so if you have any questions, suggestions, or advice for me, please let me know!          

                               
Features I’m striving for in a NOA workflow:

- Runs “Out of the Box”

- Easy to follow node connections

- Often contains notes that I have made for myself as I learn

- Contains links to any workflows that I have borrowed heavily from, in whole or in partContain links to videos or other resources that I found particularly helpful

- ???

What does this  workflow do? (Turbo SDXL- with Prompt Stylers for inspiration and learning)

Nothing special that can’t be done better using another workflow.

This workflow spins up with the Turbo SDXL checkpoint so that it will run out of the box.  Switch the checkpoint to anything that suits you and adjust the CFG, steps and sampler in the KSampler to suit your needs.

I’ve been using it with a DreamShaper XL Turbo checkpoint (DreamShaper XL - v2.1 Turbo DPM++ SDE | Stable Diffusion Checkpoint | Civitai) for better results and eliminating the need for an SDXL refiner.

I’ve been using it to explore the effect that prompts have on the generated images. The use of prompt stylers helps me quickly come up with ideas I probably would have never thought of, and since this runs relatively quickly, it encourages me to experiment and see what happens.

This workflow contains a prompt switch to make it easier to modify the positive prompt generated by the various stylers.  I have discovered that altering the styler-generated prompts to reduce “junk” tokens often helps produce another result. (longer prompts often don’t lead to better results).

For example, changing the positive prompt from “X” to something like “Y” to reduce junk tokens can do a lot.
The seed in the KSampler is set to fixed to better interrogate changes to the workflow and see their effect.

Cheers

TIP: If things aren’t working, try updating the custom nodes via the ComfyUI manager, restart ComfyUI, refresh your browser and try again.
tuall