Best AI Photorealism yet? NEW Model!

preview_player
Показать описание


Рекомендации по теме
Комментарии
Автор

2010: Get rid of your pores in Photoshop!
2023: How do we get pores in our AI pics???

jonmichaelgalindo
Автор

Sebastian, your art and tuts are very helpful,

wernerblahota
Автор

We are testing these tips to create the images used on our channel and the results are surprising. Thank you very much!!

ai-madeart
Автор

The SDXL model is absolutely the most photorealistic yet. These fine tuned checkpoints look great too!

officialhush
Автор

Hey bro, nice video, could you tell me what checkpoint and lora did you use in the image of your video?

AlexFloyd-wlrz
Автор

These generated images always look like they were taken with the best camera and edited by a Photoshop professional. How can I get medium-quality photos taken with a smartphone? Ones that aren't perfect in quality but are genuinely realistic.

sebastianszwarc
Автор

What a great checkpoint! I really loved it. It saved me from that shining effect on AI images that make theme look fake.
Thank You!

AbdulKhamis
Автор

That was a great dad joke, I actually laughed this time 😄

pfalstad
Автор

Can't wait for people to merge these models into something really amazing.

NineSeptims
Автор

My issue is the upscaling. I can get a nice 1920 x 1080 image from txt2img (960x540 + 2x high res fix), but if I try to go to 3840x2160 (4k) the different upscalers I've tried just don't seem to really help. In many cases it actually seems to hurt. I've tried numerous upscalers and just not happy with one yet.

FantasyArtworkAI
Автор

will detailed eye work fine if i am not using sdxl and rather using sd 1.5 with automatic1111 locally?? what should i do then??

rryann
Автор

Have you considered doing a video on merging/mixing checkpoint models?

Seems like a really interesting topic, but also quite complex

ferparis
Автор

What program is Sebastian using to generate the images?
I can only us comfy UI for SD XL and would like an alternative that works good with 8GB VRAM

Levitas
Автор

What extension or interface are you using, which can mix several LoRA's with strength slider? Is this avaiiable in automatic 1111? Thanks

benz.
Автор

I think the problem is still not the style of image, but subtle relations between the subject and background that looks fake. Tho, it’s easy to fix with by hand with a control net tools.

Vlow
Автор

What is the difference between Fooocus-MRE and RuinedFooocus?

MarcoEscobedo
Автор

What is that model_meta_data file in your models folder at 04:30 ?

magnusberge
Автор

sucks that you got the exact same female face 4 times in a row with the viking warrior... is there a way to mix that up?

AtheistExpert
Автор

I'm confused, is realistic stock photo a standalone model, or does it have to be used in conjunction with the sdxl base model? Standalone it runs very slowly on basic settings, and corrupts the image in the last steps.

jonescrusher
Автор

Honestly, as interesting as this is, the major lack I feel is in the lack of detailed compositions more than the lack of detailed skin or eyes. If we could just get stick figures to do precisely what we want them to do in images with complex compositions, I'd give up a lot in terms of photorealistic perfection.

michaelleue