SDXL TURBO - Local Install Guide - 1 Step Render in 0.2 Seconds

SDXL Turbo local install Guide! SDXL Turbo can render a Image in only 1 Steps. Render images in 0.2 Seconds and get realtime Image generation while you are typing the prompt

#### Links from Video ####

#### Join and Support me ####

[intangur]

Very impressive. Realtime video seems like it is not too far off at this point.

[zynga]

All these videos about comfyUI made me install finally install it. It's actually easier to use than i expected. The UI looks complicated but it's not. I didn't know the workflow is saved in the image meta data. So we can drop an image in the comfyUI canvas and it created the workflow for that image. So easy.

[sundaygloom]

Being able to prompt in real time can be helpful to learn what prompts and combinations can provide the best results fast. I dont really care for comfyui much but i have it installed so maybe ill give this a try.

[westingtyler]

when Jensen Huang, founder and CEO of Nvidia said "in the future pixels will not be rendered, they will be generated" this is what he means, eventually we'll generate 60 coherent frames per second. NUTS. actual 3d modeling will be able to be low poly or even boxed out levels, and the gpu will turn it to whatever scene is desired.

[nDman]

There are some fine tuned turbo models on Civitai which they generate in higher quality and they use normal Comfyui nodes.

[manleycreations]

0:20 That might be the most tabs I have ever seen open on a browser. Olivio, always hard at work!

[Im_that_guy_man]

I haven’t looked at the video, but since it’s not in the title, i’ma go ahead and think this is for comfyui, in which case i won’t watch the video further

[havemoney]

If you give the model fragments of a book, you will get something similar to how we perceive a book when reading.

[jeqyeyy]

Yes it's amazing how much is fast what I like in your workflow you can change even the lipstick of a person brilliant man keep it up I like your work what you do and thank you to help us a lot

[psionicxxx]

I think it made its way into the mainstream, there are quite a few checkpoints based on Turbo on CivitAI. Also, I think they kind of integrated it within the ComfyUI nowadays, so you can just use it with a regular Ksampler... And the trained models look AWESOME!

[LetondAtreides]

I just did a basic install of ComfyUI and added the model, and it seems to work fine. It's fairly fast. Although I haven't tried making the suggested tweaks here.

[wesleybilly]

okay not the video i thought this would be but very cool and informative. I could see this being useful to some people. i am not that person but somebody here is.

[fernandomasotto]

Amazing! I`d like to see a turbo SDXL workflow with AnimateDiff

[chloexu]

thank you Olivio for sharing such great video and resources, you are the best!

[muuuuuud]

Thanks Olivio for the great info ^-^ <3

[jcvijr]

Very useful if you want a preview for your prompt

[mkaleborn]

Saw this news (on Google’s news page of all things, AI art is big time!). Anyway this seems like it could be a great time saver for testing prompts and general workflow quickly.

I assume there would be a way to have Comfy show you a ‘preview slide show’ of generated images every few seconds? I guess an X/Y grid would work just as well.

But the idea of generating huge X/Y grids in seconds to test out parameters is, in your words: Really Cool. Hehe thanks for the vid!

[thanasiszazoum]

Thank you Olivio! I'm a NSFW artist on X and I'm learning how to build my own workflow for the contest, and since you uploaded your SVD guide I experimented and read A LOT. Now I'm trying to build my custom logic with custom nodes and ChatGPT for the innovative category. Do you have any detailed guides/manuals/tutorials for custom nodes? It would be very helpful. I have 15 days ahead and won't even sleep. I think you are the best AI ART Evangelist out there. Everything straight to the point!

[eimantasbutkus]

wow, this thing looks quite interesting

[FrejthKing]

you sound like how I imagine a magpie to sound if it could talk