This workflow experiments with image detailing using a mix of techniques like latent space manipulation and interpolation. I've been playing around with multiple passes to enhance detail while trying to avoid an overprocessed look. The process involves combining methods like manipulating latents with unsampler, LUT application, and grain addition.
Flux Latent Detailer
Description
This was an experiment that just seems to work, I really don't know how or why. It seems that interpolation of latents with Flux yields more fine details in images. To vary an image more substantially you can try adding a node to set another seed for the 2nd pass, this allows you to change the image details while retaining quality and most of the composition. I haven't explored other types of styles with this workflow besides photos.
I CANNOT PROVIDE SUPPORT FOR THIS, I'M JUST SHARING!
Resources
This workflow uses araminta_k_flux_koda.safetensors which can be found at CivitAI.
Koda Diffusion Flux -- Amazing lora!
The optional Kodachrome LUT file can be downloaded from here:
LUT goes into ComfyUI\custom_nodes\ComfyUI_essentials\luts, if you are using cubiq's comfy_essential nodes like in this workflow.
Setup
The Flux.1 checkpoint used in this workflow is the dev version. If you're missing any custom nodes or get errors/red nodes:
1. Click manager
2. Click on "Install Missing Custom Nodes"
3. When the installation finishes, click "Restart" as instructed
4. Reload the workflow
## Performance
I'm using an RTX 4090 with 24GB of RAM. Each image takes approximately 98 seconds.
If you modify and share this workflow, all I ask is that you credit me.