**Created**:: [[2022-12-22 Thursday]]
**Link**:: https://youtu.be/4pxDotgnyls
**Duration**:: 14:45
# Summary
In this video, the presenter explores the role of AI art in enhancing visual thinking and personal knowledge management. They discuss various AI art generation tools, compare their features, and showcase practical workflows for leveraging these tools in creative contexts. The video concludes by addressing the limitations of AI-generated art, emphasizing the importance of understanding these challenges when using such technology.
# Key Takeaways
- AI art generation involves inputting descriptive texts to create images, with various platforms offering both free and paid services.
- Practical use cases include using AI art for illustrations, lateral thinking, and problem-solving by transforming abstract ideas into visual representations.
- While AI art can enhance creativity, limitations such as inconsistent image quality and difficulty in maintaining character continuity should be considered when utilizing these tools.
# Content
Hey everyone! In today's video, we are going to talk about AI art to support visual thinking and personal knowledge management. For those unfamiliar with AI generation, it's a process where you input words to describe the desired image, and the machine creates a unique picture based on your words. There are a number of sites that offer this service, and we'll be looking at four of them in this video: two paid and two free options. We'll also be discussing four visual thinking use cases and workflows that I've been experimenting with over the past few weeks. In closing, I will share some of the limitations I've encountered while using these tools. [* ](https://youtu.be/4pxDotgnyls?t=1)
Let's start by looking at the two paid options: DALL-E from OpenAI and Midjourney. Both of these will require you to register for their services. To compare the different options, I will use the same prompt with all of the art generators. Here's the prompt that I will be using: "A photorealistic image of an anthropomorphic robot sitting at a desk in the living quarters of a spaceship, holding a stylus in hand and staring at an empty screen." With DALL-E, you get 50 credits when registering for the trial. This means you can try the engine with 50 different prompts, and each time you will receive four image variants, so in total, you will get 200 images. After the trial, you'll receive 15 free credits each month, plus you can purchase additional credits at 15 credits for $15, which is approximately 13 cents per generation prompt or roughly three cents per image. [* ](https://youtu.be/4pxDotgnyls?t=4)
Midjourney is another paid service that is accessed through a Discord bot. It can be used on the Midjourney Discord server or installed on any other Discord server if you are an administrator. Midjourney offers many options for generating images and has a helpful page for learning more—see the link in the video description. In this demonstration, using "--v4" instructs the engine to use the version 4 stable diffusion algorithm, and "--ar 3:2" creates an image with a three-to-two aspect ratio, which is ideal for use as a video thumbnail, for example. When you first register for the service, you get 25 minutes of GPU time, which is roughly enough for 25 images. Each time you provide a prompt, Midjourney generates four variants of the image. After the trial period, you can either register for a new trial with a different Discord account or purchase a subscription. The basic subscription costs eight dollars per month if you pay a year in advance, or ten dollars per month if you pay monthly. This includes 200 GPU minutes per month, or approximately 4 cents per image prompt. [* ](https://youtu.be/4pxDotgnyls?t=58)
The free option that we'll be discussing is Stable Diffusion Online. While this service may generate less impressive images, it can still be useful for supporting visual thinking. I highly recommend exploring the prompt database on the site for inspiration, learning, and access to ready-made images. I had planned to also include DALL-E Flow on Google Colab in this video, but it was not functioning when I tried to use it again after an initially successful attempt. I've included the link to it in the video description for those who want to give it a try, but even if you don't want to fiddle with the Colab workbook, you may want to read the guide to writing prompts for text-to-image AI and the clip templates; both are available as links from the workbook, but I've also included these links in the video description. Finally, there's Diffusion B. This software can be downloaded and run on a local PC, but currently it is only available for Mac. Since I don't have access to a Mac, I was not able to try it. However, they are reportedly working on a Windows version, and you can sign up to get beta access when it becomes available. If you have a Mac and have used Diffusion B, please let us know your experience in the comments below. [* ](https://youtu.be/4pxDotgnyls?t=230)
Now let's explore how you can use these art generators to help you think better by looking at four use cases. First up, let's look at using AI art for illustration. One way to use image generators is to describe the image you want in detail. For example, if you want a drawing of a wizard, you can describe this style, such as a cyberpunk wizard, or the feeling you want to convey, such as "the will to endure," or the artistic style such as "in the style of Salvador Dali," or the medium, such as "a pencil drawing." You can also describe the background, such as "in a forest" or "on a white background." Here's what a pencil drawing in the style of Salvador Dali of a cyberpunk wizard with the will to endure on a white background looks like. If you want to blend the image into a mind map or some other drawing, I recommend instructing the AI engine to use a white background. Then you can use a tool like LunaPic to replace the white background with a transparent color. To do this, upload the image to LunaPic, click edit, and select "Transparent Color." Choose either "Transparent Area" or "Transparent Color" and select the color you want to replace. You may need to experiment with the settings to find the best transparency for your image and the background color of your Excalidraw drawing. When you're happy with the transparency, save the image as a transparent PNG and drop it into Excalidraw to use in your mind map. [* ](https://youtu.be/4pxDotgnyls?t=333)
If you want something more abstract or if you struggle to think of a good illustration, then I recommend first feeding the description of the idea to chat GPT-3, asking it to write a short poem based on your text. I recommend short poetic forms such as a haiku, a tanka, a sequin, or a couplet. Use this poem to prompt the image generator. Take for example this quote from *Emergence* by Stephen Johnson: "What image would you use to illustrate it? Like ant colonies or the cells of a developing embryo, neighborhoods are patterns in time; no one wields them into existence single-handedly. They emerge by a kind of tacit consensus: the artists go here, the investment bankers here, the great preponderance of city dwellers live by those laws without any legal authority mandating compliance. It is the sidewalk, the public space, where interactions between neighbors are the most expressive and the most frequent that helps us create those laws." Let's first convert this paragraph into a haiku. If you don't like GPT-3's first response or would like to see additional responses, you can generate the response as many times as you like. Now, take this haiku and let's see what Midjourney is able to do with it. Notice that I added "an avant-garde painting" to the haiku just to throw in a bit of additional fun. Whether you like the result or not will depend on your taste; however, I find these images thought-provoking as I reflect on the idea of neighborhoods as patterns in time. [* ](https://youtu.be/4pxDotgnyls?t=450)
You can also use AI-generated art as a tool for lateral thinking. The term "lateral thinking" was coined by Edward de Bono. It is a way of solving problems by looking at them from unconventional angles. It involves breaking away from your existing thought patterns, the linear ways of thinking, and approaching problems with a more open and flexible mindset. Lateral thinking involves making connections between seemingly unrelated ideas and using imagination and intuition to come up with novel solutions. It is often used to find creative solutions to complex problems or to break out of a mental rut. The goal of lateral thinking is to come up with fresh, original ideas that can lead to new, innovative solutions. One approach for generating lateral ideas is writing down your problem and asking GPT-3 to recommend visual metaphors, then feeding these metaphors to the art generator and reflecting on the resulting image. This is what I did to illustrate lateral thinking. When writing the prompt for the art generator, I combined multiple suggestions offered by GPT-3 into a single metaphor. Another approach for generating unconventional solutions to a problem, as recommended by de Bono, is visiting a shopping mall, picking up random items in the stores, and finding a logical connection between the item in your hand and the problem on your mind. Doing this will help you see the issue in unexpected ways, and you can use AI art in the same way. You may ask DALL-E to surprise you with an image or request a random word or random scene from GPT-3 by typing, for example, "surprise me with a random scene," then feeding this into Midjourney or DALL-E and finally creating a mind map with this image and finding as many connections between the image and the problem you want to solve as possible. [* ](https://youtu.be/4pxDotgnyls?t=583)
While these tools can be really useful for generating unique and interesting images, there are a few limitations that I've encountered. First, the quality of the images can vary quite a bit. Some of the images I've generated have been really impressive, but others have been less successful. Even images that look perfect at first glance will have tiny errors, like malformed hands or misplaced eyes or similar, which can be frustrating if you're trying to create a specific image and the AI doesn't quite get it right. Second, AI can sometimes generate images that are a little bit difficult to understand. This can be especially true when it comes to abstract concepts. On the upside, these more difficult images are ideal for sparking lateral ideas; ambiguous images might actually be exactly what you're looking for when using AI art for visual thinking. Third, I wanted to create a comic strip using AI art, but I found that it was hard to get the AI to generate the same character in different scenes. This is easier to accept when working with a less-defined character like a cartoon worm, but I found the same problem even using well-known characters like Mickey Mouse. Notice how all the Mickeys look different in these images. This makes it super difficult to create visually cohesive stories. Fourth, I found that AI struggled with abstract concepts. To work around this, I tried using the GPT-3 chatbot to convert abstract ideas into more concrete descriptions that I could use for generating art. Finally, I tried using these tools to upscale low-quality old photos, but I wasn't very successful in my first few attempts. Overall, while AI art generators can be really useful, they do have their limitations, and it is important to keep these limitations in mind when using these tools. [* ](https://youtu.be/4pxDotgnyls?t=735)
Thanks for watching, and I hope you found this video helpful! [* ](https://youtu.be/4pxDotgnyls?t=879)
# YouTube Details
## YouTube Description
In this video, we'll be exploring AI art generation and how it can be used for visual thinking and personal knowledge management. We'll be looking at four different sites that offer this service, and discussing four visual thinking use cases and workflows that I've been experimenting with. At the end, we'll talk about some of the limitations I've encountered.
Don't forget to check out the links below, I've included a number of useful resources.
------
Timestamps
[00:00](https://youtu.be/4pxDotgnyls?t=0) Intro
[00:50](https://youtu.be/4pxDotgnyls?t=50) PART 1 - ART GENERATORS
[00:55](https://youtu.be/4pxDotgnyls?t=55) DALL-E
[02:12](https://youtu.be/4pxDotgnyls?t=132) Midjourney
[03:47](https://youtu.be/4pxDotgnyls?t=227) Stable diffusion online
[04:13](https://youtu.be/4pxDotgnyls?t=253) DALL-E Flow on Google Colab
[04:51](https://youtu.be/4pxDotgnyls?t=291) Diffusion Bee
[05:24](https://youtu.be/4pxDotgnyls?t=324) PART 2 - VISUAL THINKING
ILLUSTRATION
[05:46](https://youtu.be/4pxDotgnyls?t=346) Use Case 1: Illustrating Concrete Ideas
[06:34](https://youtu.be/4pxDotgnyls?t=394) Setting a transparent background
[07:30](https://youtu.be/4pxDotgnyls?t=450) Use Case 2: Illustrating Abstract Concepts with Chat GPT3
[09:36](https://youtu.be/4pxDotgnyls?t=576) LATERAL THINKING
[10:36](https://youtu.be/4pxDotgnyls?t=636) Use Case 3: Lateral Thinking with Visual Metaphors
[11:08](https://youtu.be/4pxDotgnyls?t=668) Use Case 4: Lateral Thinking with Random Images
[12:08](https://youtu.be/4pxDotgnyls?t=728) PART 3 - LIMITATIONS
[12:22](https://youtu.be/4pxDotgnyls?t=742) Limitation 1
[12:53](https://youtu.be/4pxDotgnyls?t=773) Limitation 2
[13:22](https://youtu.be/4pxDotgnyls?t=802) Limitation 3
[13:58](https://youtu.be/4pxDotgnyls?t=838) Limitation 4
[14:15](https://youtu.be/4pxDotgnyls?t=855) Limitation 5
[14:25](https://youtu.be/4pxDotgnyls?t=865) Closing Remarks
------
Check out my cohort-based visual thinking course: https://www.visual-thinking-workshop.com
If you find my videos helpful, please say thanks by buying me a coffee: https://ko-fi.com/zsolt
If you want to connect, you can reach me (@zsviczian) in the Obsidian Members Group (OMG) on Discord or on Twitter.
------
ART GENERATORS:
Chat GPT3: https://chat.openai.com/chat
DALL-E 2: https://openai.com/dall-e-2/
Midjourney
Stable Diffusion Online: https://stablediffusionweb.com/
DALL-E Flow on Google Colab: https://colab.research.google.com/github/jina-ai/dalle-flow/blob/main/client.ipynb
DiffusionBee: https://diffusionbee.com/
USEFUL LINKS:
Prompt database: https://stablediffusionweb.com/prompts
Midjourney image parameters help: https://midjourney.gitbook.io/docs/imagine-parameters
A Guide to writing prompts for Text-to-Image-AI: https://docs.google.com/document/d/17VPu3U2qXthOpt2zWczFvf-AH6z37hxUbvEe1rJTsEc/edit?usp=sharing
Clip templates: https://docs.google.com/document/d/1j2IAumYz4iZopOTAAOcCUKbFXP0jHK8mRgD4NLFKkaw/edit
OTHER TOOLS AND LINKS:
LUNAPIC: https://www4.lunapic.com/editor/
Obsidian.md: https://obsidian.md/
Exclidraw plugin: https://github.com/zsviczian/obsidian-excalidraw-plugin
My blog: https://zsolt.blog
## YouTube Transcript
[hey everyone in today's video we are](https://youtu.be/4pxDotgnyls?t=1) [going to talk about AI art to support](https://youtu.be/4pxDotgnyls?t=4) [visual thinking and personal Knowledge](https://youtu.be/4pxDotgnyls?t=6) [Management](https://youtu.be/4pxDotgnyls?t=9) [for those unfamiliar with Air Generation](https://youtu.be/4pxDotgnyls?t=10) [it's a process where you input words to](https://youtu.be/4pxDotgnyls?t=13) [describe the desired image and the](https://youtu.be/4pxDotgnyls?t=17) [Machine creates a unique picture based](https://youtu.be/4pxDotgnyls?t=20) [on your words](https://youtu.be/4pxDotgnyls?t=22) [there are a number of sites that offer](https://youtu.be/4pxDotgnyls?t=24) [this service and we'll be looking at](https://youtu.be/4pxDotgnyls?t=26) [four of them in this video two paid and](https://youtu.be/4pxDotgnyls?t=29) [two free options we'll also be](https://youtu.be/4pxDotgnyls?t=33) [discussing four visual thinking use](https://youtu.be/4pxDotgnyls?t=36) [cases and workflows that I've been](https://youtu.be/4pxDotgnyls?t=38) [experimenting with over the past few](https://youtu.be/4pxDotgnyls?t=40) [weeks](https://youtu.be/4pxDotgnyls?t=42) [in closing I will share some of the](https://youtu.be/4pxDotgnyls?t=43) [limitations I've encountered while using](https://youtu.be/4pxDotgnyls?t=46) [these tools](https://youtu.be/4pxDotgnyls?t=48) [let's start by looking at the two paid](https://youtu.be/4pxDotgnyls?t=55) [options dollar E from openai and](https://youtu.be/4pxDotgnyls?t=58) [mid-journey both of these will require](https://youtu.be/4pxDotgnyls?t=61) [you to register to their services to](https://youtu.be/4pxDotgnyls?t=64) [compare the different options I will use](https://youtu.be/4pxDotgnyls?t=67) [the same prompt with all of the art](https://youtu.be/4pxDotgnyls?t=70) [generators](https://youtu.be/4pxDotgnyls?t=73) [here's the prompt that I will be using](https://youtu.be/4pxDotgnyls?t=74) [a photorealistic image of an](https://youtu.be/4pxDotgnyls?t=77) [anthropomorphic robot sitting at a desk](https://youtu.be/4pxDotgnyls?t=81) [in the living quarters of a spaceship](https://youtu.be/4pxDotgnyls?t=84) [holding a stylus in hand and staring at](https://youtu.be/4pxDotgnyls?t=86) [an empty screen](https://youtu.be/4pxDotgnyls?t=91) [with Dolly you get 50 credits when](https://youtu.be/4pxDotgnyls?t=93) [registering for the trial this means you](https://youtu.be/4pxDotgnyls?t=96) [can try the engine with 50 different](https://youtu.be/4pxDotgnyls?t=100) [prompts and each time you will receive](https://youtu.be/4pxDotgnyls?t=102) [four image variants so in total you will](https://youtu.be/4pxDotgnyls?t=106) [get 200 images](https://youtu.be/4pxDotgnyls?t=109) [after the trial you'll receive 15 free](https://youtu.be/4pxDotgnyls?t=112) [credits each month plus you can purchase](https://youtu.be/4pxDotgnyls?t=116) [additional credits 15](https://youtu.be/4pxDotgnyls?t=119) [415 credits which is 13 cents per](https://youtu.be/4pxDotgnyls?t=122) [generation prompt or roughly three cents](https://youtu.be/4pxDotgnyls?t=127) [per image](https://youtu.be/4pxDotgnyls?t=131) [mid-journey is another paid service that](https://youtu.be/4pxDotgnyls?t=133) [is accessed through a Discord bot it can](https://youtu.be/4pxDotgnyls?t=136) [be used on the mid Journey Discord](https://youtu.be/4pxDotgnyls?t=140) [server or can be installed on any other](https://youtu.be/4pxDotgnyls?t=142) [Discord server if you are an](https://youtu.be/4pxDotgnyls?t=146) [administrator](https://youtu.be/4pxDotgnyls?t=148) [mid Journey offers many options for](https://youtu.be/4pxDotgnyls?t=150) [generating images and has a helpful page](https://youtu.be/4pxDotgnyls?t=153) [for learning more see the link in the](https://youtu.be/4pxDotgnyls?t=157) [video description](https://youtu.be/4pxDotgnyls?t=159) [in this demonstration dash dash V4](https://youtu.be/4pxDotgnyls?t=161) [instructs the engine to use the version](https://youtu.be/4pxDotgnyls?t=165) [4 stable diffusion algorithm and dash](https://youtu.be/4pxDotgnyls?t=168) [dash AR3 colon 2 to create an image with](https://youtu.be/4pxDotgnyls?t=172) [a three to two aspect ratio which is](https://youtu.be/4pxDotgnyls?t=177) [ideal for use as a video thumbnail for](https://youtu.be/4pxDotgnyls?t=181) [example](https://youtu.be/4pxDotgnyls?t=184) [then you first register for the service](https://youtu.be/4pxDotgnyls?t=185) [you get 25 minutes of GPU time which is](https://youtu.be/4pxDotgnyls?t=187) [roughly enough for 25 images each time](https://youtu.be/4pxDotgnyls?t=191) [you provide a prompt mid-journey](https://youtu.be/4pxDotgnyls?t=194) [generates four variants of the image](https://youtu.be/4pxDotgnyls?t=197) [after the trial period you can either](https://youtu.be/4pxDotgnyls?t=200) [register for a new trial with a](https://youtu.be/4pxDotgnyls?t=203) [different Discord account or purchase a](https://youtu.be/4pxDotgnyls?t=206) [subscription the basic subscription](https://youtu.be/4pxDotgnyls?t=208) [costs eight dollars per month if you pay](https://youtu.be/4pxDotgnyls?t=211) [a year in advance or ten dollars per](https://youtu.be/4pxDotgnyls?t=215) [month if you pay monthly this includes](https://youtu.be/4pxDotgnyls?t=217) [200 GPU minutes per month or](https://youtu.be/4pxDotgnyls?t=220) [approximately 4 cents per image prompt](https://youtu.be/4pxDotgnyls?t=223) [the free option that we'll be discussing](https://youtu.be/4pxDotgnyls?t=227) [is stable diffusion online while this](https://youtu.be/4pxDotgnyls?t=230) [service May generate less impressive](https://youtu.be/4pxDotgnyls?t=234) [images it can still be useful for](https://youtu.be/4pxDotgnyls?t=236) [supporting visual thinking](https://youtu.be/4pxDotgnyls?t=239) [I highly recommend exploring the prompt](https://youtu.be/4pxDotgnyls?t=241) [database on the site for inspiration](https://youtu.be/4pxDotgnyls?t=245) [learning and access to ready-made images](https://youtu.be/4pxDotgnyls?t=248) [I had planned to also include Dolly flow](https://youtu.be/4pxDotgnyls?t=253) [on Google collab in this video but it](https://youtu.be/4pxDotgnyls?t=257) [was not functioning when I tried to use](https://youtu.be/4pxDotgnyls?t=260) [it again after an initially successful](https://youtu.be/4pxDotgnyls?t=262) [attempt I've included the link to it in](https://youtu.be/4pxDotgnyls?t=265) [the video description for those who want](https://youtu.be/4pxDotgnyls?t=268) [to give it a try but even if you don't](https://youtu.be/4pxDotgnyls?t=270) [want to fiddle with the collab workbook](https://youtu.be/4pxDotgnyls?t=273) [you may want to read the guide to](https://youtu.be/4pxDotgnyls?t=276) [writing prompts for text to image Ai and](https://youtu.be/4pxDotgnyls?t=278) [the clip templates both are available as](https://youtu.be/4pxDotgnyls?t=282) [links from the workbook but I've also](https://youtu.be/4pxDotgnyls?t=285) [included these links in the video](https://youtu.be/4pxDotgnyls?t=288) [description and finally there's](https://youtu.be/4pxDotgnyls?t=290) [diffusion B this software can be](https://youtu.be/4pxDotgnyls?t=292) [downloaded and run on a local PC but](https://youtu.be/4pxDotgnyls?t=295) [currently it is only available for Mac](https://youtu.be/4pxDotgnyls?t=298) [since I don't have access to a Mac I was](https://youtu.be/4pxDotgnyls?t=301) [not able to try it however they are](https://youtu.be/4pxDotgnyls?t=305) [reportedly working on a Windows version](https://youtu.be/4pxDotgnyls?t=307) [and you can sign up to get beta access](https://youtu.be/4pxDotgnyls?t=310) [when it becomes available](https://youtu.be/4pxDotgnyls?t=312) [if you have a Mac and have used the](https://youtu.be/4pxDotgnyls?t=315) [Fusion B please let us know your](https://youtu.be/4pxDotgnyls?t=318) [experience in the comments below](https://youtu.be/4pxDotgnyls?t=321) [now let's explore how you can use these](https://youtu.be/4pxDotgnyls?t=330) [art generators to help you think better](https://youtu.be/4pxDotgnyls?t=333) [we'll look at four use cases first up](https://youtu.be/4pxDotgnyls?t=337) [let's look at using AI art for](https://youtu.be/4pxDotgnyls?t=341) [illustration](https://youtu.be/4pxDotgnyls?t=344) [one way to use image generators is to](https://youtu.be/4pxDotgnyls?t=346) [describe the image you want in detail](https://youtu.be/4pxDotgnyls?t=349) [for example if you want a drawing of a](https://youtu.be/4pxDotgnyls?t=352) [wizard you can describe this style such](https://youtu.be/4pxDotgnyls?t=355) [as a cyberpunk wizard or the feeling you](https://youtu.be/4pxDotgnyls?t=358) [want to convey such as the Ville to](https://youtu.be/4pxDotgnyls?t=362) [endure or the artistic style such as in](https://youtu.be/4pxDotgnyls?t=365) [the style of Salvador Dali or the medium](https://youtu.be/4pxDotgnyls?t=369) [such as a pencil drawing you can also](https://youtu.be/4pxDotgnyls?t=373) [describe the background such as in a](https://youtu.be/4pxDotgnyls?t=377) [forest or on a white background](https://youtu.be/4pxDotgnyls?t=380) [here's what a pencil drawing in the](https://youtu.be/4pxDotgnyls?t=383) [style of Salvador Dolly of a cyberpunk](https://youtu.be/4pxDotgnyls?t=386) [wizard with the will to endure on a](https://youtu.be/4pxDotgnyls?t=389) [white background looks like if you want](https://youtu.be/4pxDotgnyls?t=393) [to blend the image into a mind map or](https://youtu.be/4pxDotgnyls?t=396) [some other drawing I recommend](https://youtu.be/4pxDotgnyls?t=399) [instructing the AI engine to use a white](https://youtu.be/4pxDotgnyls?t=401) [background then you can use a tool like](https://youtu.be/4pxDotgnyls?t=404) [Luna pick to replace the white](https://youtu.be/4pxDotgnyls?t=407) [background with a transparent color to](https://youtu.be/4pxDotgnyls?t=409) [do this upload the image to Luna pick](https://youtu.be/4pxDotgnyls?t=412) [click edit and select transparent color](https://youtu.be/4pxDotgnyls?t=415) [choose either transparent area or](https://youtu.be/4pxDotgnyls?t=419) [transparent color and select the color](https://youtu.be/4pxDotgnyls?t=422) [you want to replace](https://youtu.be/4pxDotgnyls?t=425) [you may need to experiment with the](https://youtu.be/4pxDotgnyls?t=427) [settings to find the best transparency](https://youtu.be/4pxDotgnyls?t=429) [for your image and the background color](https://youtu.be/4pxDotgnyls?t=432) [of your excalator drawing](https://youtu.be/4pxDotgnyls?t=435) [when you're happy with the transparency](https://youtu.be/4pxDotgnyls?t=438) [save the image as a transparent PNG and](https://youtu.be/4pxDotgnyls?t=440) [drop it into X colored row to use in](https://youtu.be/4pxDotgnyls?t=445) [your mind map](https://youtu.be/4pxDotgnyls?t=448) [if you want something more abstract or](https://youtu.be/4pxDotgnyls?t=450) [if you struggle to think of a good](https://youtu.be/4pxDotgnyls?t=453) [illustration then I recommend first](https://youtu.be/4pxDotgnyls?t=455) [feeding the description of the idea to](https://youtu.be/4pxDotgnyls?t=457) [chat gpt3 asking it to write a short](https://youtu.be/4pxDotgnyls?t=460) [poem based on your text I recommend](https://youtu.be/4pxDotgnyls?t=464) [short poetic forms such as a haiku a](https://youtu.be/4pxDotgnyls?t=467) [Tonka a sequin or a couplet](https://youtu.be/4pxDotgnyls?t=471) [use this poem to prompt the image](https://youtu.be/4pxDotgnyls?t=474) [generator](https://youtu.be/4pxDotgnyls?t=478) [take for example this quote from](https://youtu.be/4pxDotgnyls?t=479) [emergence by Stephen Johnson what image](https://youtu.be/4pxDotgnyls?t=482) [would you use to illustrate it](https://youtu.be/4pxDotgnyls?t=486) [like ant colonies or the cells of a](https://youtu.be/4pxDotgnyls?t=489) [developing embryo neighborhoods are](https://youtu.be/4pxDotgnyls?t=491) [patterns in time no one Wheels them into](https://youtu.be/4pxDotgnyls?t=494) [existence single-handedly they emerge by](https://youtu.be/4pxDotgnyls?t=497) [a kind of tacit consensus the artists go](https://youtu.be/4pxDotgnyls?t=501) [here the investment bankers here](https://youtu.be/4pxDotgnyls?t=504) [the great preponderance of City dwellers](https://youtu.be/4pxDotgnyls?t=507) [live by those laws without any legal](https://youtu.be/4pxDotgnyls?t=510) [Authority mandating compliance](https://youtu.be/4pxDotgnyls?t=513) [it is the sidewalk the public space](https://youtu.be/4pxDotgnyls?t=516) [where interactions between neighbors are](https://youtu.be/4pxDotgnyls?t=519) [the most expressive and the most](https://youtu.be/4pxDotgnyls?t=522) [frequent that helps us create those laws](https://youtu.be/4pxDotgnyls?t=524) [let's first convert this paragraph into](https://youtu.be/4pxDotgnyls?t=529) [a haiku if you don't like GPT 3's First](https://youtu.be/4pxDotgnyls?t=532) [Response or would like to see additional](https://youtu.be/4pxDotgnyls?t=537) [responses you can generate the response](https://youtu.be/4pxDotgnyls?t=540) [as many times as you like](https://youtu.be/4pxDotgnyls?t=543) [now take this Haiku and let's see what](https://youtu.be/4pxDotgnyls?t=547) [mid journey is able to do with it](https://youtu.be/4pxDotgnyls?t=550) [notice that I added an avant-garde](https://youtu.be/4pxDotgnyls?t=553) [painting to the Haiku just to throw in a](https://youtu.be/4pxDotgnyls?t=556) [bit of additional fun](https://youtu.be/4pxDotgnyls?t=559) [whether you like the result or not will](https://youtu.be/4pxDotgnyls?t=562) [depend on your taste however I find](https://youtu.be/4pxDotgnyls?t=566) [these images thought provoking as I](https://youtu.be/4pxDotgnyls?t=568) [reflect on the idea of neighborhoods as](https://youtu.be/4pxDotgnyls?t=572) [patterns in time](https://youtu.be/4pxDotgnyls?t=575) [you can also use AI generated art as a](https://youtu.be/4pxDotgnyls?t=577) [tool for lateral thinking](https://youtu.be/4pxDotgnyls?t=581) [the term lateral thinking was coined by](https://youtu.be/4pxDotgnyls?t=583) [Edward de Bono it is a way of solving](https://youtu.be/4pxDotgnyls?t=586) [problems by looking at them from](https://youtu.be/4pxDotgnyls?t=589) [unconventional angles it involves](https://youtu.be/4pxDotgnyls?t=591) [breaking away from your existing thought](https://youtu.be/4pxDotgnyls?t=594) [patterns the linear ways of thinking and](https://youtu.be/4pxDotgnyls?t=597) [approaching problems with a more open](https://youtu.be/4pxDotgnyls?t=600) [flexible mindset lateral thinking](https://youtu.be/4pxDotgnyls?t=603) [involves making connections between](https://youtu.be/4pxDotgnyls?t=606) [seemingly unrelated ideas and using](https://youtu.be/4pxDotgnyls?t=608) [imagination and intuition to come up](https://youtu.be/4pxDotgnyls?t=612) [with Noble Solutions](https://youtu.be/4pxDotgnyls?t=614) [lateral thinking is often used to find](https://youtu.be/4pxDotgnyls?t=617) [Creative Solutions to complex problems](https://youtu.be/4pxDotgnyls?t=619) [or to break out of a mental rut](https://youtu.be/4pxDotgnyls?t=622) [the goal of lateral thinking is to come](https://youtu.be/4pxDotgnyls?t=627) [up with fresh original ideas that can](https://youtu.be/4pxDotgnyls?t=629) [lead to new innovative solutions](https://youtu.be/4pxDotgnyls?t=632) [one approach for generating lateral](https://youtu.be/4pxDotgnyls?t=635) [ideas is writing down your problem and](https://youtu.be/4pxDotgnyls?t=638) [asking gpt3 to recommend visual](https://youtu.be/4pxDotgnyls?t=641) [metaphors then feeding these metaphors](https://youtu.be/4pxDotgnyls?t=645) [to the art generator and reflecting on](https://youtu.be/4pxDotgnyls?t=648) [the resulting image this is what I did](https://youtu.be/4pxDotgnyls?t=651) [to illustrate lateral thinking when](https://youtu.be/4pxDotgnyls?t=654) [writing the prompt for the art generator](https://youtu.be/4pxDotgnyls?t=657) [I combined multiple suggestions offered](https://youtu.be/4pxDotgnyls?t=660) [by gpt3 into a single metaphor](https://youtu.be/4pxDotgnyls?t=664) [another approach for generating](https://youtu.be/4pxDotgnyls?t=668) [unconventional solutions to a problem as](https://youtu.be/4pxDotgnyls?t=670) [recommended by the Bono is going to a](https://youtu.be/4pxDotgnyls?t=673) [shopping mall visiting shops picking up](https://youtu.be/4pxDotgnyls?t=676) [random items in the stores and finding a](https://youtu.be/4pxDotgnyls?t=679) [logical connection between the item in](https://youtu.be/4pxDotgnyls?t=683) [your hand and the problem on your mind](https://youtu.be/4pxDotgnyls?t=686) [doing this will help you see the issue](https://youtu.be/4pxDotgnyls?t=690) [in unexpected ways](https://youtu.be/4pxDotgnyls?t=693) [and you can use AI art in the same way](https://youtu.be/4pxDotgnyls?t=695) [you may ask Dolly too to surprise you](https://youtu.be/4pxDotgnyls?t=699) [with an image or request a random word](https://youtu.be/4pxDotgnyls?t=702) [or random scene from gpt3 by typing for](https://youtu.be/4pxDotgnyls?t=705) [example surprise me with a random scene](https://youtu.be/4pxDotgnyls?t=710) [then feeding this into mid-journey or](https://youtu.be/4pxDotgnyls?t=713) [Dolly and finally creating a mind map](https://youtu.be/4pxDotgnyls?t=717) [with this image and finding as many](https://youtu.be/4pxDotgnyls?t=720) [connections between the image and the](https://youtu.be/4pxDotgnyls?t=722) [problem you want to solve as possible](https://youtu.be/4pxDotgnyls?t=724) [while these tools can be really useful](https://youtu.be/4pxDotgnyls?t=732) [for generating unique and interesting](https://youtu.be/4pxDotgnyls?t=735) [images there are a few limitations that](https://youtu.be/4pxDotgnyls?t=737) [I've encountered](https://youtu.be/4pxDotgnyls?t=740) [first the quality of the images can vary](https://youtu.be/4pxDotgnyls?t=742) [quite a bit](https://youtu.be/4pxDotgnyls?t=746) [some of the images I've generated have](https://youtu.be/4pxDotgnyls?t=747) [been really impressive but others have](https://youtu.be/4pxDotgnyls?t=750) [been less successful even images that](https://youtu.be/4pxDotgnyls?t=752) [look perfect at the first glance will](https://youtu.be/4pxDotgnyls?t=755) [have tiny errors like malformed hand or](https://youtu.be/4pxDotgnyls?t=758) [a misplaced eye or similar](https://youtu.be/4pxDotgnyls?t=761) [this can be frustrating if you're trying](https://youtu.be/4pxDotgnyls?t=764) [to create a specific image and the AI](https://youtu.be/4pxDotgnyls?t=766) [doesn't quite get it right](https://youtu.be/4pxDotgnyls?t=770) [second AI can sometimes generate images](https://youtu.be/4pxDotgnyls?t=773) [that are a little bit difficult to](https://youtu.be/4pxDotgnyls?t=777) [understand this can be especially true](https://youtu.be/4pxDotgnyls?t=779) [when it comes to abstract Concepts on](https://youtu.be/4pxDotgnyls?t=782) [the upside these more difficult images](https://youtu.be/4pxDotgnyls?t=785) [are ideal to spark lateral ideas](https://youtu.be/4pxDotgnyls?t=788) [ambiguous images might actually be](https://youtu.be/4pxDotgnyls?t=792) [exactly what you're looking for when](https://youtu.be/4pxDotgnyls?t=794) [using AI art for visual thinking](https://youtu.be/4pxDotgnyls?t=797) [third I wanted to create a comic strip](https://youtu.be/4pxDotgnyls?t=802) [using AI art but I found that it was](https://youtu.be/4pxDotgnyls?t=804) [hard to get the AI to generate the same](https://youtu.be/4pxDotgnyls?t=808) [character in different scenes this is](https://youtu.be/4pxDotgnyls?t=810) [easier to accept when working with a](https://youtu.be/4pxDotgnyls?t=813) [lens defined character like a cartoon](https://youtu.be/4pxDotgnyls?t=816) [Worm but I found the same problem even](https://youtu.be/4pxDotgnyls?t=818) [using well-known characters like Mickey](https://youtu.be/4pxDotgnyls?t=822) [Mouse](https://youtu.be/4pxDotgnyls?t=825) [notice how all the Mickeys look](https://youtu.be/4pxDotgnyls?t=826) [different in these images](https://youtu.be/4pxDotgnyls?t=829) [this makes it super difficult to create](https://youtu.be/4pxDotgnyls?t=831) [visually cohesive stories fourth I found](https://youtu.be/4pxDotgnyls?t=835) [that AI struggled with abstract Concepts](https://youtu.be/4pxDotgnyls?t=839) [to work around this I tried using gpt3](https://youtu.be/4pxDotgnyls?t=842) [chatbot to convert abstract ideas into](https://youtu.be/4pxDotgnyls?t=846) [more concrete descriptions that I could](https://youtu.be/4pxDotgnyls?t=849) [use for generating art](https://youtu.be/4pxDotgnyls?t=852) [finally I tried using these tools to](https://youtu.be/4pxDotgnyls?t=855) [upscale low quality old photos but I](https://youtu.be/4pxDotgnyls?t=858) [wasn't very successful in my first few](https://youtu.be/4pxDotgnyls?t=862) [attempts](https://youtu.be/4pxDotgnyls?t=864) [overall while a IR the generators can be](https://youtu.be/4pxDotgnyls?t=866) [really useful they do have their](https://youtu.be/4pxDotgnyls?t=869) [limitations it is important to keep](https://youtu.be/4pxDotgnyls?t=872) [these limitations in mind when using](https://youtu.be/4pxDotgnyls?t=875) [these tools](https://youtu.be/4pxDotgnyls?t=877) [thanks for watching and I hope you found](https://youtu.be/4pxDotgnyls?t=879) [this video helpful](https://youtu.be/4pxDotgnyls?t=882)