Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

[Local Wordmason]

Made some more Dall-E 3 images this time the news hamster crew. I tried my hand at some light image editing and combining other AI generated assets. I suck at paint.net so the quality may vary.

Here's our normal news hamster, all dressed up for the spookiest month

The Lolcow hamster is doing some live reporting from the (corn) field

Ni Hao China hamster! He's here to give us some party approved propaganda news.


SHIEEET its the Inner-City hamster. He would give us news about black affairs, but he can't read.

Stunning and brave as always, Troon hamster.

And last but not least, the Russian-Ukraine war hamster. Be careful out there, buddy!

 

[888Flux]

Reddit got super pissy over that and tried to cancel him. Between that and a lack of a license early on there was a lot of screeching, the stable diffusion discord even banned him. The big flaw with the plan was no one was able to step up and replace his webui until recently. Now SD.Next, comfyUI, and even InvokeAI rival or even surpass the automatic version. SDXL isn't a wet fart like SD2 and a1111 doesn't handle SDXL well.

Any screencaps of that? would be funny to read. Tools for open source models seem to be dominated by /g/ 4channner types (no surprise) from oobabooga to the guy with Uncle Ho as his PFP. Which is probably why it makes corporations seethe even more.

 

[inception_state]

Thanks to @Lidl Drip's collection of Josh photos, I made a LoRA as promised. It's kind of close, but it was difficult since the images had a variety of different styles/hair/ages. Sorry @Null, this was too funny to pass up. LoRA safetensors below.

 

[Sulla]

View attachment 5401858View attachment 5401859

Josh with a beard is just sargon kek

 

[Epoch]

KF won't let me upload a safetensors file and catbox is throwing 504s, where should I upload this?

KF should let you upload it if you stick it in a zip file first

 

[inception_state]

I managed to get it to upload by putting the safetensors file in a 7zip archive. So just extract it and then put the safetensors file in your LoRA directory, like usual. The trigger is "null_josh".

 

[The Last Stand]

Thanks to @Lidl Drip's collection of Josh photos, I made a LoRA as promised.

Probably a high order, but could you make one of Hedy Lamarr for me?

 

[inception_state]

Probably a high order, but could you make one of Hedy Lamarr for me?

There are already a couple of textual inversions on CivitAI:

https://civitai.com/models/150849/hedy-lamarr-1940s-textual-inversion

https://civitai.com/models/61900/hedy-lamarr

 

[Lidl Drip]

Amazing! Thank you!



so handsome

 

[seri0us]

Thanks for the Nullie LORA, you guys rock!

 

[The Last Stand]

Okay, why is Dall-e 3 resetting my images AND credits?

Ava Gardner enjoying a Wendy's milkshake and burger outside. Sunny day with some clouds

 

[GenociderSyo]

Isn't think thing supposed to give me more credits weekly? or is that only if you hit 0?

 

[The Last Stand]

Isn't think thing supposed to give me more credits weekly? or is that only if you hit 0?

When you hit zero, evidently, you have the option to get more with MS Rewards. Otherwise, from my experience, it generates weekly.

I truly think MS is overcompensating with their "content policy" with the AI tool. The launch from a week ago to now, the AI generation is neutered beyond reasonable doubt. I guarantee you many people using it are just fooling around.

 

[GenociderSyo]

When you hit zero, evidently, you have the option to get more with MS Rewards. Otherwise, from my experience, it generates weekly.

I truly think MS is overcompensating with their "content policy" with the AI tool. The launch from a week ago to now, the AI generation is neutered beyond reasonable doubt. I guarantee you many people using it are just fooling around.

Only time it blocked me was when i tried to make an AI of my avatar because of her name. But I've seen on reddit people getting banned for stupid things.

 

[The Last Stand]

Only time it blocked me was when i tried to make an AI of my avatar because of her name. But I've seen on reddit people getting banned for stupid things.

It hasn't banned me (yet) but it's consecutively giving me blocked prompts. Why does everybody hate Old Hollywood?

Who wants some music?



Now there's the Bing I love.

 

[Potatoherder]

Any screencaps of that? would be funny to read. Tools for open source models seem to be dominated by /g/ 4channner types (no surprise) from oobabooga to the guy with Uncle Ho as his PFP. Which is probably why it makes corporations seethe even more.

Seems the discord i was in with automatic got nuked. A lot of it can be found on the reddit.

https://old.reddit.com/r/StableDiffusion/comments/y34h2a/emad_posts_a_public_apology_to_automatic1111_on/

https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/2509

https://archive.ph/fv3BY

https://old.reddit.com/r/StableDiffusion/comments/y1uuvj/automatic1111_did_nothing_wrong_some_people_are/

A lot of it had to do with the NovelAI leak and accusations of stolen code. It was a year ago so i am super hazy on the exact details.


Here is a link to the bitching about it not being open source. https://old.reddit.com/r/StableDiff...utomatic1111_there_is_no_requirement_to_make/

 

[anustart76]

Reddit got super pissy over that and tried to cancel him. Between that and a lack of a license early on there was a lot of screeching, the stable diffusion discord even banned him. The big flaw with the plan was no one was able to step up and replace his webui until recently. Now SD.Next, comfyUI, and even InvokeAI rival or even surpass the automatic version. SDXL isn't a wet fart like SD2 and a1111 doesn't handle SDXL well.

So which of the UIs is the best choice right now?

 

[The Last Stand]

Who wants some music?

If you're wondering the prompt:

Marlene Dietrich on an album cover for opera music. paternal advisory sticker. album is named good heavens in German

I was just expecting Dietrich posing with a microphone with that sticker on the corner. This exceeded my expectations. Look, the cover even looks worn through years of ownership and time.

 

[Potatoherder]

So which of the UIs is the best choice right now?

I personally use SD.Next for sd1.5 stuff and comfyUI for SDXL. That is mostly a case of SDXL running like shit on automatic based stuff when i tried like 3 months ago. InvokeAI is looking more interesting so i may give that a go.

tldr
It all comes down to personal preference really. If a1111 works for you keep using it. I jumped to SD.Next because automatic was taking a break earlier this year and there were like 2 months with no updates. If you can get your head around comfyUI then you can do crazy things. A lot of the video stuff seems to be happening there. Though you need a lot of hardware for video.

 

[DavidS877]

I guess with my new video card I should figure out how to train SDXL, because the world needs null_XL.

I started with SD1.5 on A1111 but my old card didn't like SDXL on A1111, ComfyUI has been fine with it. I'm enjoying playing with randomized prompts and checkpoints and just batching up a bunch and letting it go. But ComfyUI really can be painful. Next up I'm trying to get my HiRes Fix style workflow to have fewer extra horrible artifacts.