Super High Resolution Images with UltraPixel | Local Comfy UI Installation

[@krakenunbound]

I created a workflow leveraging Ollama to generate a prompt and then share that prompt with the Process Node and a SDXL image generation which sends the image to the "controlnet_image" input. Setting the Controlnet_weight = 1.0 it is pretty close to the original image with amazing details and upscaling. The initial image is 1344x768 and is upscaled to 3840x2304 which allows me to make the images a wallpaper. Works very well. Falling in love with this.

[@AI_Made_Simple.]

Nice work 👏

[@runebinder]

Added the nodes for this on Thursday and have been very impressed. I'm finding the backgrounds have extra detail and pop to them and it's nice not to have to upscale.

[@AI_Made_Simple.]

💯 and so much faster 😃

[@laylasmart]

So where is this Ultrapixel and how can I download a portable version of it? his Github is a mess.

[@OriBengal]

Does it work with custom trained Loras though? I often make large images, but they are portraits of people, so the resemblance is hyper-important.

[@RickySupriyadi]

is there any upscaler works with potato and result still great?

[@loquacious1956]

Great tool ... it takes a long time to render with my 3090 ti though

[@louiewashere3]

now if we can get 3d models from the created images for use in Unreal Engine.:)

[@trsd8640]

How this compares to Supir?

[@flyingdogsoft]

both are non-commercial only. So I am more interested to in best alternatives to these. Anyway for research purpose it would be interesting to compare both.

[@MichaelWinterberg]

on a 4090 i can generate up to 5120x5120 and needs 23,5 GB VRAM -> amazing image quality

[@AI_Made_Simple.]

That's really impressive

[@bboyfxxu]

The examples shown for Midjourney are pretty bad. Midjourney is still way better than anything I've seen done with Stable Diffusion.