- The Boost
- Posts
- Never Overpay for Stock Images Again: A Guide to Midjourney
Never Overpay for Stock Images Again: A Guide to Midjourney
Hi Friends š
In this weekās email:
Never Pay for Stock Images Again: A Guide to Midjourney
ChatGPT Prompts: Sales training & toolkit updates
If you struggle to write but have ideas to share, we have news for you
Everyone will have their own ChatGPT or LLM soon as costs come down
Letās get startedā¦
AI for Marketing š
š¬ ChatGPT
Turn your blog into an engaging Tweet Thread. Tweet Threads are the best way to attract new Twitter followers. [Link]
Sales training prompt. This prompt for an interactive sales training session with ChatGPT as your sales mentor. It allows the user to role-play different sales scenarios, receive immediate feedback, improve negotiation skills, and adapt to various levels of complexity based on performance. [Link]
The ChatGPT Marketing Magic Toolkit is constantly being updated. But more importantly, scanning the prompts and the types will give you ideas on how to create your own prompts [Link]
š If you donāt publish your ideas and knowledge due to lack of time, read this:
Is lack of time holding you back from sharing your valuable insights? What if you could effortlessly document your ideas without fearing sounding uninformed or spending too much time?
We can help you do exactly that by being interviewed by an AI bot. You reply, and it turns your insights into a beautifully crafted article as if written by you, in a voice you choose for just $38 a month. Sign up for the waitlist here: [Link]
Deep Dive
Never Pay for Stock Images Again: A Guide to Midjourney
Why You Should Care: The correct image in a social media ad can 10x the results. Every business, whether a start-up or an established firm, needs visuals to tell its story effectively. But procuring high-quality stock images can be costly and time-consuming. Now you can get stunning visuals, that attract attention without breaking the bank custom designed to your exact requirements.
What You Need to Know: Mid Journey is a generative AI tool that produces photorealistic images based on a text prompt. This means you can tailor-make images for your campaigns without worrying about a hefty price tag or exhausting search. All you need is a clear vision of your desired image, and Midjourney will do the rest.
On with the storyā¦
Ā£5 life insurance leads.
That was our cost to generate Life Insurance leads using Facebook ads.
When a single click on Google costs Ā£15!
Let me take you back to 2019, when I ran lead generation agency.
Our work involved generating tens of thousands of leads for insurance and financial products, at a cost per lead that was just a fraction of the cost per click on Google Search.
A crucial factor in acquiring these low-cost leads?
The images used in our ads.
Barbie AI. Come on Barbie lets go party!
The right image could outperform others tenfold.
Stopping viewers mid-scroll, engaging their interest, and enticing them to read our content.
Back then, our only option was stock photography from expensive libraries like Shutterstock, costing us both time and money.
Now with AI we have a much better optionā¦
Enter Midjourney
Dopamine alert: Playing with Midjourney is seriously addictive.
Midjourney, the AI image generation tool, is a game-changer.
It allows you to generate your own stock photography. Or any image you can imagine in fact.
Unlike traditional image-sourcing methods, Midjourney creates images based on your text prompts.
So, instead of browsing through thousands of pre-made images, you can generate custom images that perfectly suit your requirements.
Hereās how it works.
You give Midjourney an image creation prompt, and it creates the image for you.
I dictated this Midjourney image prompt whilst walking my dog:
Imagine a man who's 50 in the Norfolk countryside walking his Golden Lab and small, black, furry dog past open fields that have been freshly cut with stubble near old Norfolk-style churches. It should be a landscape shot, viewing the dogs and the man from the side. ultra-detailed photography, high detail, 8k quality
Which gave this Midjourney output a few minutes later:
Midjourney recreated my dogs and me
How to Use Midjourney
Midjourney runs in Discord, the forum system.
It works great because you can see other peopleās images and the prompts they used to create them. So we can learn all together! But it does take getting used to.
Step 1: Join the Midjourney Discord
Discord works best as an app on your PC, do make sure to download it.
If you haven't already, join the Midjourney Discord server by clicking on the āJoin the betaā link provided on the Midjourney website
Step 2: Go to the Newbies Channels to Test It Out (Free Credits Explained)
Once you are in the Midjourney Discord server, navigate to the "Newbies" or "Getting Started" channels designed for new users.
These channels are ideal for testing out Midjourney and familiarizing yourself with its features.
As a new user, you may receive free credits to use the bot and explore its capabilities.
Step 3: How to Initiate a Prompt to Create an Image
Choose a newbies channel (e.g., #newbies-122) where you want to interact with Midjourney and initiate a prompt.
To start a new prompt, type the command:
/imagine
Add your prompt to create the image For example:
/imagine a stock photo of an AI expert
Hit return to send the command, and Midjourney will start to create your image
Step 4: Finding Previous Messages via the Inbox
If you wish to review the previous messages sent by Midjourney, head to your inbox on Discord. This is an icon next to the (?) in the top right of the desktop app.
Look for the conversation with Midjourney and click on it to access the chat history.
Scroll through the messages to find the prompts and your responses with images.
Step 5: Jump to the Original Message Using Jump
Once you've found the message you want to revisit, click on it to select it.
Click "Jump" in the message to return to this image and message.
Step 6: Expanding the Prompt with U1 and V1, etc.
You can edit the created images using variables like U1, V1.
Image numbered left to right, top to bottom 1,2,3,4
Vās choose a version to expand on. Eg V1 give more versions of the top left image.
Uās upscale the image you have chosen.
Step 7: How to Upgrade Using /subscribe
You can consider upgrading your subscription if you want to access additional features or more credits.
To do this, use the
/subscribe
command in the appropriate channel, and Midjourney will guide you through the subscription process.
Video Tutorial - Generating images with Midjourney
Using Midjourney and Discord can be a bit confusing, I have wrapped it all up in a tutorial video, click below to watch
(You can speed up the video if you find it too slow using the playback speed button at the bottom right of the video).
Here are some images I created earlier
Here are the Midjourney generated, royalty-free images and the prompt that generated them.
funny stock photo -
Very beautiful 50 years old woman looking at the camera wearing a haute couture dress in a art gallery in London at night movie scene, extremely detailed, dynamic lighting, dynamic pose, high fantasy, detailed, Depth of Field, Editorial Photography, UHD, HDR, HD, Natural Lighting, High Quality, Super - Resolution, Saturated Colors, Ray tracing Reflections, Cinematic, Unreal Engine 5.2, MetaHuman, Highly Detailed, ultra detailed, 8k, hdr, cinematic lighting, depth of field, Photorealistic, Highly Intricate Details, ultra high, ultra detail, Photorealistic, Cinematic, Highly Detailed, Camera Shots, Dramatic, Realistic, high quality, hd, Chromatic Aberration, GB Displacement, Scan Lines, Ray Traced, Ray Tracing Ambient Occlusion, Anti - Aliasing, FKAA, TXAA, RTX, SSAO, Shaders, OpenGL - Shaders, GLSL - Shaders, Post Processing, Post - Production, Cell Shading, Tone Mapping, CGI, VFX, SFX, insanely detailed and intricate, elegant, super detailed, dynamic pose, volumetric, ultra - detailed, intricate details, 8K, super detailed , in 2023 --s 250
An old man, sitting in a london cafe, looking out of the window, into the rain. The photo has been taken through the window at the old man. ultra-detailed photography, high detail, 8k quality.
A group of friends close up, at a summer music festival in the UK, extremely detailed, social media style, UGC, dynamic lighting, dynamic pose, high fantasy, detailed, Depth of Field, Editorial Photography, UHD, HDR, HD, Natural Lighting, High Quality, Super - Resolution, Saturated Colors, Ray tracing Reflections, Cinematic, Unreal Engine 5.2, MetaHuman, Highly Detailed, ultra detailed, 8k, hdr, cinematic lighting, depth of field, Photorealistic, Highly Intricate Details, ultra high, ultra detail, Photorealistic, Cinematic, Highly Detailed, Camera Shots, Dramatic, Realistic, high quality, hd, Chromatic Aberration, GB Displacement, Scan Lines, Ray Traced, Ray Tracing Ambient Occlusion, Anti - Aliasing, FKAA, TXAA, RTX, SSAO, Shaders, OpenGL - Shaders, GLSL - Shaders, Post Processing, Post - Production, Cell Shading, Tone Mapping, CGI, VFX, SFX, insanely detailed and intricate, elegant, super detailed, dynamic pose, volumetric, ultra - detailed, intricate details, 8K, super detailed , in 2023 --s 250 -
Last weekās newsletter thumbnail was created by Midjourney - of course!
A photorealistic recycling image showing a long form text being recycled automatically in other forms of content like Twitter, Linkedin, Instagram. Have a man looking out in the middle looking really happy and a bit crazy -
Youtube thumbail funny stock photo
A weird stock photo
What about copyright - are the images mine?
OK, I donāt like the difficult questions!
Text and images Generative AI copy is a grey area. No one really knows. The law hasnāt caught up yet.
Hereās what I found smarter people than me saying:
āCurrently, there is no clear answer to this question. In the United States, copyright law is governed by the Copyright Act of 1976 which does not explicitly mention AI-generated images. Furthermore, case law regarding copyright and AI is scantā¦
ā¦ if you pay for your [Midjourney account], the company says āYou basically own all Assets you create using Midjourneyās image generation and chat services.ā
So according to Midjourney, the images you create are yours. Feel free to use any of the images I created above. A link back would be appreciated!
Time for you to try Midjourney
Tools like Midjourney have revolutionized the process of image creation and sourcing. They've allowed us to break free from the restrictions of traditional stock images and costly photoshoots. With AI-powered tools like Midjourney, we now have the ability to generate a vast array of custom images from simple text prompts. This has enabled us to unleash our creativity and maximize the impact of our marketing campaigns.
Finally, for those interested in LLM Development
Language learning models (LLMs), such as ChatGPT, are currently dominated by OpenAI with its flagship GPT-4, which is considered the best model available. However, the landscape is changing rapidly, which is promising for those who aspire to build their own models. This evolution means you no longer need to share your data with major LLM providers like OpenAI. Affordable APIs are set to spearhead these advancements. Below are some links that showcase these developments:
Companies are building their own LLMs. Stripe announced they have built an internal LLM and that 30% of their team use it. More companies will follow suit. [Link]
Opening source LLMs are improving. The team at Stability has announced Stable Beluga an open-source language model comparable to GPT-4. Stable Beluga, a cutting-edge language model exhibits impressive capabilities nearly on par with those of GPT-4 according to some measurements [Link]
The trend on LLM API costs is down. OpenAI has announced substantial reductions in API costs, ranging from 25% to 75% on various models, to make cutting-edge technology more accessible. The popular text-embedding-ada-002 model will see a 75% price cut to $0.0001 per 1,000 tokens, while the widely-used GPT-3.5 Turbo chat model will have its input tokens reduced by 25%. Additionally, the GPT-3.5 Turbo 16K model will be priced at $0.003 per 1,000 input tokens and $0.004 per 1,000 output tokens [Link]
LLMs models compared for output quality [Link]
Open API costs by inputs and outputs by word as calculated by Code Interpreter. [Link]
Thatās a wrap! Donāt miss the next issueā¦if this newsletter is in your spam folder, please don't forget to move it to your primary inbox.
Doing so really helps ensure you donāt miss future issues.
Finally, if you got value from this email, please share the love and forward it to your friends.
Thanks for reading,
Tim