Milovana.com

Here are a few I tried:

it's not a exactly picture but an Ai generated video.but i really like the result hope it's ok to share it here
https://grok.com/imagine/post/3936bb6f- ... atform=web

alba<3 wrote: Fri Jan 16, 2026 8:45 am it's not a exactly picture but an Ai generated video.but i really like the result hope it's ok to share it here
https://grok.com/imagine/post/3936bb6f- ... atform=web

Very cool. I haven't used Grok yet.

Thanks for sharing these! It’d be awesome if you could mention which AI model you’re using

e-lustic wrote: Sun May 17, 2026 12:58 pm

stick_x wrote: Sun May 17, 2026 11:33 am Thanks for sharing these! It’d be awesome if you could mention which AI model you’re using

Sure, no pb. It doesn't get any simpler than this: ComfyUI, z-image-turbo, with this workflow.

Completely offline, completely private, zero cost per token - except your electricity

z-image-turbo.comfyui.workflow.png

Sounds like a beast to run! You must have a top-tier GPU to handle that workflow smoothly

I built my new system over Christmas and should have made the jump to local generation awhile ago.

Still learning ComfyUI but starting to get the hang of it. Definitely a learning curve.

Had a decent photoshoot with Mia last night using Z-Image_Turbo_BF16 with a mix of Bastard and NSFW_ZIT_master LORAs A few more from 116 total generations.

Spoiler: show

Mia005.jpg (214.52 KiB) Viewed 3297 times

Mia015.jpg (200.35 KiB) Viewed 3297 times

Mia018.jpg (215.86 KiB) Viewed 3297 times

Mia023.jpg (199.4 KiB) Viewed 3297 times

Mia035.jpg (208.28 KiB) Viewed 3297 times

Mia036.jpg (166.23 KiB) Viewed 3297 times

Mia041.jpg (186.08 KiB) Viewed 3297 times

Mia056.jpg (198.96 KiB) Viewed 3297 times

47dahc wrote: Thu May 21, 2026 1:19 pm I built my new system over Christmas and should have made the jump to local generation awhile ago.

Still learning ComfyUI but starting to get the hang of it. Definitely a learning curve.

Had a decent photoshoot with Mia last night using Z-Image_Turbo_BF16 with a mix of Bastard and NSFW_ZIT_master LORAs
Mia003.jpg
A few more from 116 total generations.

Spoiler: show

Mia005.jpg
Mia015.jpg
Mia018.jpg
Mia023.jpg
Mia035.jpg
Mia036.jpg
Mia041.jpg
Mia056.jpg

Looks like a beast to learn but the results are insane. I’m trying to generate a consistent character set (nothing NSFW, just poses/outfits with different fashion clothes) and was wondering if this setup you mentioned would work for that?

stick_x wrote: Sun May 24, 2026 10:20 am
Looks like a beast to learn but the results are insane. I’m trying to generate a consistent character set (nothing NSFW, just poses/outfits with different fashion clothes) and was wondering if this setup you mentioned would work for that?

I think consistency comes down to the LORAs and obviously character description. Not sure how true it is but I've read a few times that Z-Image was based on images of the same model so facial features will look alike. I found one LORA to combat that but haven't had much luck with it. With Mia's pics above, I set my "control after generate" on the KSampler to "increment". Not sure if that helps with consistency or not. I would think it would as long as character/clothing description in the prompt doesn't change that much. I'll do some tests with my workflow when I get a chance and try to upload a few with the metadata intact. You should be able to drag and drop an image into Comfy and it will load the workflow if the metadata is there. Unfortunately, I had to resize these images and in doing so, it stripped the metadata.