2023-06-23 - AI Images - Chewing the Fat 📈

[sethdaily]

Hey artists!

Let's talk about this rule

[bradystroud]

I had fun making this. I wanted the Llama for a thumbnail for a video about editing websites using TinaCMS.
Some artisits are worried about AI using their art, but I don’t see how its different than GitHub Copilot using my code to get smarter

I crafted this masterpiece with Midjourney 🎨:

Prompt: An orange llama proudly admiring his changes to a website. include red black and grey on the website

[Geordie88]

Brady and I had an interesting discussion and wanted to open up the discussion to others :).

AI art scrapes a lot of images for it's data set, in many cases copyrighted images and at this time there isn't really a way for artists to opt out of having there work used in this manner. While there are some analogies to artists using the work of others as reference, the fact that it lacks another persons interpretation is a sticking point for many. Personally I feel for the creatives but I am not sure what practical steps can or are being taken. What's your take??

Brady if I missed anything from our discussion please pop in a reply :D

[matt-goldman]

Pretty difficult to argue with that.

[evecogan]

I created this image with DALLE 2:
{{

}}
The prompt I used: {{ Create a visual metaphor of artists original work being ripped off by AI image generators }}
My experience was {{ quick and easy, but the image was more artistic and less literal than I would have liked}}

[tiagov8]

I created this with Bing Chat:

The prompt I used: create an image of computer ordering system background
My experience was OK. I would spend more time to get a better prompt and image if it was for real (e.g. For SSW blue is
first-degree)

[brydeno]

I created this with Dall-E using the prompt "a koala eating a bus"

[sethdaily]

@brydeno I tried your exact prompt in Midjourney:

[leahy268]

I created this with BingChat:

The prompt I used: Create me an image representing clean energy
My experience was fantastic. I generated that image in under a minute. I could never design something that could in that amount of time.

[tkapa]

I created this with Midjourney: -
The prompt I used: -
My experience was: -

I think that the rule has some theoretically cool uses for AI art, and there are certainly times when it can be applied. I personally might apply what I was thinking of - Generating Blog hero art based on the title, if I can be bothered.

I'm not interested in using AI generated art in a professional sense at SSW. Clients shouldn't be receiving this kind of stuff unless they want it.

At the moment, I take serious issue with how these models were created (by taking art from unwilling creators, for the most part).

Keep me in the loop for a project that is sourced entirely on art that people contributed to. Not only would it be more interesting, it can be used in clear conscience.

[bradystroud]

I convinced @tkapa to proompt engineer this image with Leonardo.ai

[sethdaily]

@tkapa @bradystroud put proompt engineer on your resume

[ozdemir-mehmet]

I created this with Stable Diffusion:

The prompt I used: GT86 white with TE37 wheels
My experience was fine

[jakebayliss]

I created this with Midjourney:

The prompt I used was: progression

Rule seems like re-written documentation

[Marxoz]

that guy is going to fall

[Geordie88]

I created this on Gencraft.

prompt: Boat on a sparkling ocean, small islands scattered throughout Style, mythical

Feedback on gencraft: limited number of prompts per day but has an art style selector which I find very useful.

[TobyChurches]

I created this with dream studio:

The prompt I used: A software developer using a computer with the angular logo
My experience was: Average.. you have to be much more specific then something like chat GPT (my other prompts produces bad results)

[calumjs]

Created with Bing Chat
Prompt: A powerful software developer crafting azure/cloud resources with his pure will. Branding should be shades of grey and highlights of red. Make it an epic sci-fi fantasy scene. Make sure there is no text on the image. Make it a high def splash screen. Make it more obvious that the guy is provisioning azure resources

[apforsyth]

very cool

[sethdaily]

@calumjs your exact prompt in midjourney:

[LiamHarveyssw]

Created in DreamStudio
Prompt: Roronoa Zoro wearing a flower crown, surrounded by corgis

My experience was good, I had a lot of fun creating an image with a lot of random prompts

[landonmaxwell]

What, no swords?

[lukecookssw]

I created this with Dream Studio:

The prompt I used: website logo depicting a developer violently facepalming in frustration
My experience was mixed. So many fingers...

[vladislav-kir]

I created this with DreamStudio:

The prompt I used: Mining trucks of orange color moving coal
My experience was: Not as good as I expected - none of the trucks that I got looked like the once I expected

[AntPolkanov]

I created this with DreamStudio:

The prompt I used: insurance company, professional indemnity, trustworthy, good deal, established and reputable company.
My experience was mixed. The image is not what I expected, but I just need to be better at prompting. The people hasn't been drawn correctly though.

[Freego1783]

I created this with Stable Diffusion Online :

The prompt I used : IT company meeting, people getting mad at each other

It was quick to generate the image but I'm not really glad of the result. People's faces are coming out of a bad nightmare 😄

[sethdaily]

@Freego1783 your prompt in midjourney:

[ozairashfaqueSSW]

Generated in BlueWillow
Prompt: A corn in suit working in a retail space, cute cartoon style, landing page style

AI-generated images are fun but they now come with price

[hveraus]

I created this with BingChat:

The prompt I used: draw me a image of group of people attending a tech talk. high res. professional look.
It's alright, but I can quickly find something similar and free on Google Image

[Aibono1225]

I created this with dreamstudio:

The prompt: Shiba Inu on a vehicle
Although the dog's paws look a bit strange, the image generated by AI is much more interesting and creative than what I thought, and I find it difficult to find similar images on Google Images.

[tino-liu]

I created this with Leonardo.ai

The prompt: A beautiful girl in champagne color trendy urban clothes in a bar. Full body, hyperdetailed painting, luminism, art by {{Carne Griffiths}} and {{Wadim Kashin}} concept art, 4k resolution, fractal isometrics details bioluminescence, 3d render, octane render, intricately detailed, cinematic, full color, hand drawn, gritty, realistic, hit definition, on paper, ethereal background, abstract beauty, stand, approaching perfection, pure form, golden ratio, minimalistic, unfinished, concept art, intricate details, 8k post production, high resolution, sharp focus, studio photo.

Negative Prompt: (Leonardo.ai has this feature, not sure about Midjourney) double body, double face, double features, incorrect posture, close up, two heads, two faces, plastic, Deformed, bad anatomy, bad eyes, crossed eyes, disfigured, poorly drawn face, mutation, mutated, {{extra limb}}, ugly, poorly drawn hands, missing limb, blurry, floating limbs, disconnected limbs, malformed hands, blur, out of focus, long neck, long body, {{{mutated hands and fingers}}}, {{{out of frame}}}, long fingers, blender, doll, cropped, low-res, close-up, poorly-drawn face, out of frame double, two heads, blurred, ugly, disfigured, too many fingers, deformed, repetitive, black and white, grainy, extra limbs, bad anatomy, zoomed, soft light, smooth skin, closeup, deformed, extra limbs, extra fingers, mutated hands, bad anatomy, bad proportions , blind, bad eyes, ugly eyes, dead eyes, blur, vignette, out of shot, out of focus.

My experience with AI generated artwork have been a blast. I love playing around with Procreate and when I draw... I usually avoid doing full-body artwork just because the anatomy of the human body (ESPECIALLY THE HANDS) are extremely hard to draw. I guess AI is having the same problem with hands as well.

[UlyssesMaclaren]

Bloody hell, Tino! Next level!

[AnkyChe]

I created this with DreamStudio:

The prompt I used: Cartoon： A man and a dog, the man is mad while the dog is happy. No photos.
My experience was: it's quick to generate but the picture looks scary.

[jimmidier]

I created this with BingChat:

The prompt I used: abandoned wooden villa with vines on the outer wall, damaged car afar in the corner, one or two broken windows, overgrown weeds, inclined road sign, chilly daylighting, heavy fog, forest landscape, viewed from across the tarred street with puddles and discarded luggage infront of the camera, ground level view, cinematic, high res
My experience was: It was quite hard for me to come up with the prompts that match the exact expectation. I spent nearly an hour playing around different combinations and phrases but none worked perfectly well. I guess AI plays more of an assistant role in the workflow of art creation, where human interfere of fine tuning is still required at this stage.

[johnxu16]

I created this with Midjourney

The prompt I used: A white American shorthair wearing a black suit plays the violin on top of the Griffith observatory --q 2 --v 5 --s 200
My experience was: It is much more difficult to tweak the engine to meet specific requirements. He's mad at me and refused to improvise there 🥲

[sylhuang]

I created this with DreamStudio:

The prompt I used: an online ordering system with brand name Northwind
My experience was very hard to get the letter script (brand) clear and visible. But the overall image quality is fine.

[rubycogan]

I created this with Midjourney The prompt I used: 8 year old girl eating fresh fig in Ikaria picked off wild tree

My experience was very positive. I am keen to use this software in the future. I did find it very odd posting my prompt on a shared chat with strangers...

[BrookJeynes]

I created this with -
The prompt I used: software developers desk with plants and old books
My experience was not good. I couldn't generate anything due to most websites either requiring a subscription or requiring me to create an account.

[2fernandez]

Prompt I used: Modern samurai with gladiator clothing

[samsmithssw]

i created this with dream studio:

prompt i used: robot wizard koala

my feedback was it's not very robotic, nor magical

[chrisschultzssw]

I created this with Bing:

The prompt I used: a well dressed person plugging an ethernet cable into a data rack

My experience was OK. Big only does 1:1 ratio, and I wanted some neater cables - this was the best of the 4 options.

[SofieHong]

I created this with DALLE-3:

The prompt I used: 16:9, image for PowerPoint cover page, for a company called Northwind, it’s an ordering system.
My experience was amazing 🤩