Sometimes, understanding new concepts & their potential takes time.
Recently, I realized that AI-gen assets offer more monetization opportunities beyond just selling art, stock images, or NFTs.
While some may already be aware of this, I wanted to share it for those still exploring the possibilities.
POV
If it's not open, it hasn't reached its full potential. Quality thrives in the open-source community.
Here’s a representation of Ogbe Meji in psychedelic style, created using open-source AI tools 👀
Ogbe Meji teaches us that light and clarity lead to greatness - just as open-source fuels creativity and innovation.
#Ifa #AIArt #OgbeMeji #Yoruba
Are you looking to train your style or face LoRA model with Flux locally for FREE? Here’s a simple guide as promised:
1. Set up Python – It’s easy on Linux; Windows users may need extra software like Python.
Tip: Use Anaconda, this helps manage installation of multiple python versions on any computer.
2. Install the AI Toolkit from Ostris Github: github. com/ostris/ai-toolkit
Just follow the steps in the ReadME.
3. Sign in to Hugging Face – You’ll need this to get Flux.1 model files.
4. Create a dataset – Prepare your images, my trick is use images with different facial expressions and different backgrounds, then add descriptions.
5. Start training – Adjust a few settings and let it run.
Tip: Fine-tune the output – Set LoRA strength between 1.3-1.5 for best results.
Share your success – Help others by sharing your results!
Happy creating! #Comfyui #FluxLoRA #Flux1 #AIArtWork #XLabsAI
I finally trained a face-focused Flux LoRA LOCALLY!
🟢Dataset: 13 images, 1024px, diverse expressions, green screen background.
🟢Training Steps: 20k.
Honestly, the data wasn't the best—it was woof—but the results are impressive!
Just imagine the potential with a properly prepped dataset!
Hope this doesn’t create a new breed of Catfishers…
This is a Sign.
Workflow in coming 🙌
#AI #MachineLearning #FluxLoRAs
Week 11 Notes - #buildinpublic
75 days in | At 53% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
Week 11 🫡 - Finally! Just one more week to go! 🤩
My main goal for last week was to build a virtual exhibition website using Unreal Engine or Unity!
Showcasing AI arts with an immersive website experience is appealing, and if you are an artist, it would be cool to have your collections displayed this way. 🩷
Although this is overkill because I can quickly pay for a spatial account and get this moving, my CHEAP nature took the best of me! 😅
After researching, I discovered that Unity has WebGL, but the quality could be better than Unreal Engine. Unfortunately, UE has discontinued support for HTML5, so I had to learn how to do it manually. It was a challenging experience, but I was motivated by the free blueprint on the UE Marketplace.
You think I am cheap; I say I am resourceful! 😂
Well, it was a Huge learning curve, but I figured it out - video below! I will upload this to the domain and work hard to meet the deadlines for the 12th week.
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕Use AI to generate art depicting the Yorubas' culture from the 17th Century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕Understand how GANs and Ultra Realistic AI Arts work.
➕Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The plan 🗓:
➕Stage 1 - Research and Discovery: 25% Done
➕Stage 2 - Consume and Activate: 25% Ongoing
➕Stage 3 - Showcase Results - 3% ongoing
➕Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
🔙 From Last week:
⏳ I am considering a virtual gallery display. I may have to pivot to another option if this does not work out.
🔜 Moving Forward:
⏳ Finish the immersive gallery,
⏳ Prompt out the arts
⏳ Round up the project
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕ How to package UE4 project for HTML + ALSV4 on HTML
➕ How to install old versions of UE
➕ Creating a virtual art gallery with no $ cost but TIME and FRUSTRATION cost
________
The Challenges 😢:
➕ The usual: managing ad-hoc commitments, work and family.
I apologize, frens, for my Ghost Mode; except for my replies, I have automated my posts :D
🎴🎴🎴🎴🎴🎴
Total costs So Far ���: I was advised not to include these parts anymore. But my DM is open if you want to know the details.
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
If you have any questions or suggestions, kindly leave them in the comments or send them to DMs.
Thank you for your support and words of encouragement!
I appreciate you
Week 9 Notes - #buildinpublic
61 days in | At 45% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
Week 9 was a slow week filled with surprises. However, I made some progress in training the model using Stable Diffusion XL with Kohya LoRA. Combining these with a good workflow has proven to be highly effective in enhancing the performance and accuracy of my models.
I took time analyzing the images extracted from video clips and am happy with the promising results from the checkpoints. The quality and clarity of the visuals are truly astounding, thanks to SDXL's advanced image upscaling and Kohya LoRA's exceptional image processing capabilities!💪
Although it might not have been my most productive week, every win counts!🤍
PS: I am so grateful I had the dual GPU setup!
Recently, Stable Video Diffusion, an image-to-video model, was released on November 21, 2023. The model can generate 14 frames at a resolution of 576x1024, given a context frame of the same size. However, it requires 40GB GPU VRAM to run smoothly. Luckily, I have a combination of 48GB, so I shouldn't have any issues when I work on bigger projects.😍
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕Use AI to generate art depicting the Yorubas' culture from the 17th Century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕Understand how GANs and Ultra Realistic AI Arts work.
➕Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The plan 🗓:
➕Stage 1 - Research and Discovery: 25% Done
➕Stage 2 - Consume and Activate: 20% Ongoing
➕Stage 3 - Showcase Results
➕Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
🔙 From Last week:
⏳ Learn how to train LoRa models using Kohya_ss GU
⏳ Pre-process my datasets
⏳ Train the LoRA model using the ComfyUI interface, providing the pre-processed images and a set of prompts.
⏳ Finetune my model using the ComfyUI interface by training it on additional images or prompts.
🔜 Moving Forward:
⏳ Continue training the LoRA model using the ComfyUI interface, providing the pre-processed images and a set of prompts.
⏳ Finetune my model using the ComfyUI interface by training it on additional images or prompts.
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕ The art of prioritizing - getting better! ✅
➕ The function of denoise, cfg, sampler, VAE encode, inpainting, samplers, etc✅
➕ Building ComfyUI workflows like a PRO!✅
➕ Finally broke the prompting hack! ✅
Tip: you get better responses when you adopt a conversational style of prompting rather than the typical "create a cool image of a cat."
________
The Challenges 😢:
➕ Time and managing commitments.��
🎴🎴🎴🎴🎴🎴
Total costs So Far 💵: I was advised not to include these parts anymore. But my DM is open if you want to know the details.
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
As always, if you have any questions or suggestions, please feel free to leave them in the comments or send them to DMs.
Thank you and GN. 🙏🏻
Image: My attempt at testing prompts to create coloring books for children 🤷♂️
Week 8 Notes - #buildinpublic
54 days in | At 42% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
Week 8! Super Stoked!
It's been an incredible week 8; this week was 🔥!
I made significant progress this week and finally completed the PC build, which WORKED perfectly!
FROM
🚦🚦🚦🚦
- Cable Management: vs last week's 😅
🚦🚦🚦🚦
🚦🚦🚦🚦
- First Boot
🚦🚦🚦🚦
🚦🚦🚦🚦
Say hello to my lil friend "MetrBox."
🚦🚦🚦🚦
🚦🚦🚦🚦🚦🚦🚦
Moving forward, I explored the ComfyUI + SD 1 world. I learned how to connect nodes, use checkpoints and VAE decode, finetune images with the power of Denoise, KSampler, seed, inpainting, and create workflows for training models.
Mind-blowing! I won't lie; it felt like my brain was being filled to capacity with all this new knowledge. But hey, that's what makes it so darn fascinating!
Who knew learning could be this thrilling?
I want to continue training my model with my dataset until I reach a checkpoint where I feel the model is ready for the task.
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕Use AI to generate art depicting the Yorubas' culture from the 17th Century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕Understand how GANs and Ultra Realistic AI Arts work.
➕Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The plan 🗓:
➕Stage 1 - Research and Discovery: 25% Done
➕Stage 2 - Consume and Activate: 17% Ongoing
➕Stage 3 - Showcase Results
➕Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
🔙 From Last week:
⏳ Finish building the hardware
⏳ Install all required software and pull all repo files
⏳ Learn how to train LoRa models using Kohya_ss GU (this may take time)
🔜 Moving Forward:
⏳ Learn how to train LoRa models using Kohya_ss GU
⏳ Pre-process my datasets
⏳ Train the LoRA model using the ComfyUI interface, providing the pre-processed images and a set of prompts.
⏳ Finetune my model using the ComfyUI interface by training it on additional images or prompts.
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕ Patience ++
➕ How to prioritize commitments in a better way
➕ How to build ComfyUI workflows
➕ The function of denoise, cfg, sampler, VAE encode, inpainting, samplers, etc
________
The Challenges 😢:
➕ Had some med hiccups but
➕ As usual, my prompts, but I am getting closer each day!
🎴🎴🎴🎴🎴🎴
Total costs So Far 💵: I was advised not to include these parts anymore. But my DM is open if you want to know the details.
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
As always, if you have any questions or suggestions for me, please feel free to leave them in the comments or send them to DMs.
Thank you and GN. 🙏🏻
Week 6 Notes - #buildinpublic
40 days in | At 35% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
In week 6, I had to make a major decision regarding my PC build. My initial plan was to use dual GPUs, but after consulting with BestBuy's Geek Squad, I was convinced to use only one GPU.
They explained that having a second GPU would be unnecessary, as my PC would only utilize it partially. Additionally, housing two RTX-3090s requires some serious cooling strategies. So, I changed my plan and got all the necessary components for a single GPU build.
But the next day, Nvidia released a new feature that allows applications to share VRAM - talk about perfect timing!
It's called System Memory Fallback for Stable Diffusion. You can find more details here: https://t.co/uWm8rWA1N1.
So, once again, another pivot in my plan! As they say, if something is worth doing at all, it's worth doing well.
Let me walk you through my updated configuration:
- GPU: RTX 3090 2x
- CPU: Intel Core i7-12700K
- MOBO: GIGABYTE Z690 AORUS ELITE AX
- RAM: DDR4 48GB
- SSD: 1tb Nvme SSD
- Case: Dynamic EVO XL - Full Tower (the largest case I've ever seen!)
- Cooler: Noctua NH-L9i
- PSU: EVGA 1600w fully modular
For more detailed information on each component, check out this link on PC Part Picker: https://t.co/Kmthr2bLNx
Now, putting aside the technical aspects for a moment...
In preparing for the machine learning tasks, I started collecting and compiling training assets, and I couldn't help but be amazed by the incredible craftsmanship and advancements achieved then as well as today within this culture.
Unfortunately, though, this rich culture and its artifacts have not received the attention they deserve, with many stolen and locked away in European museums.
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕Use AI to generate art depicting the Yorubas' culture from the 17th Century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕Understand how GANs and Ultra Realistic AI Arts work.
➕Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The plan 🗓:
➕Stage 1 - Research and Discovery: 25% Done
➕Stage 2 - Consume and Activate: 10% Ongoing
➕Stage 3 - Showcase Results
➕Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
🔙 From Last week:
⏳ Get the missing PC parts ✅ and finish building the hardware ❌
⏳ Install all required software and pull all repo files ❌
⏳ Learn how to train LoRa models using Kohya_ss GU (this may take time) ❌
⏳ Collect datasets looking at 30 - 100 assets ✅
🔜 Moving Forward:
⏳ Finish building the hardware
⏳ Install all required software and pull all repo files
⏳ Learn how to train LoRa models using Kohya_ss GU (this may take Time)
⏳ Collect datasets looking at 30 - 100 assets - Ongoing
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕ Hardware setups, SLI, and airflow!
➕ You can't capture the real essence of a specific culture if the generative model hasn't been trained with information on them. I tried multiple prompts for unique artifacts peculiar to the Yorubas but failed to generate tangible outputs.
➕ The richness of Yoruba culture through legacy arts proves a high level of civilization pre-colonialization.
➕ I was shocked to discover attempts by non-yoruba's telling our stories!
________
The Challenges 😢:
➕ As usual, my prompts, but it's a WIP
➕ Time is still the same issue, but I am improving a day at a time.
🎴🎴🎴🎴🎴🎴
Total costs So Far 💵: I was advised not to include these parts anymore. But my DM is open if you want to know the details.
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
THANK YOU if you have made it this far! My summaries are always a long read! 😅
If you have any questions or suggestions for me, feel free to leave them in the comments or send them via direct message.
Thank you. 🙏🏻
My prompt: neon cubist courtesy @HBCoop_
Week 4 Notes - #buildinpublic
26 days in. | Currently, I am at 25% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
I can't believe it's already week 4. Time really flies!
I've been diving headfirst into my research this past week, and let me tell you, it's been such a delight to uncover all the amazing things about Yoruba culture.
It fills me with so much pride🔥!
Did you know Yoruba traditions, beliefs, and music have impacted the global stage?
Especially in the Americas! It's incredible how deeply rooted their spirit is.
I discovered some fascinating stuff while learning about syncretic religions like Santería and Candomblé.
These practices have incorporated elements from Yoruba gods and rituals—absolutely mind-blowing! And can we talk about vibrant ceremonies filled with dancing, music, and colorful outfits? So lively 😍!
Yes, these beautiful festivities are celebrated far beyond the homeland of the Yorubas.
During various festivals and carnivals, you'll find their influence shining through from Cuba to Brazil—and even in other Caribbean countries, blending local customs with unique Yoruba traditions for an unforgettable experience 🩷.
In addition, I discovered that Yoruba rhythms and instruments, such as the talking drum, have been incorporated into various music genres, including salsa, jazz, and reggae.
The diaspora has played a pivotal role in ensuring that Yoruba musical traditions remain alive and continue to evolve.
Indeed, these festivals and music genres serve as bridges, fostering cultural exchange and mutual appreciation between the African continent and the diaspora.
I must admit that discovering that the spirit of the Yoruba people continues to resonate and influence cultures far from their homeland has made me awestruck! 🌟⭐️
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕Use AI to generate art depicting the Yorubas' culture from the 17th century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕Understand how GANs and Ultra Realistic AI Arts work.
➕Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The Plan 🗓:
➕Stage 1 - Research and Discovery: 25% Done
➕Stage 2 - Consume and Activate
➕Stage 3 - Showcase Results
➕Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
🔙 From Last week:
⏳Keep trying more prompts and fine-tuning - Done ✅
⏳Research the Yoruba influence in the Americas and other parts of the world.- Done ✅
🔜 Moving Forward:
⏳Continue trying more prompts and fine-tuning using upscales
⏳Explore the modern-day manifestations of Yoruba culture. 20th century to date.
⏳Research the possibility of using immersive platforms to host the AI arts instead of regular web pages.
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕Santería in Cuba.
➕Candomblé in Brazil.
➕Yoruba's influence in music.
________
The Challenges 😢:
➕My prompts: I am still struggling with the below:
💠Deformity! I always add negative prompts, but AI is too heady; it won't listen 🤯. I get 10% deformed objects 😢☹️
💠Another major challenge is setting the perfect lighting and camera position, but I will try some tips shared by @OnwardsProject today.
➕Last week was quite hectic as I struggled to balance my research, journaling, family, and other commitments. I am hoping that this week will be better in terms of managing my tasks. Although I had most of my tweets scheduled, I still missed the therapeutic effect of self-reflection.
🎴🎴🎴🎴🎴🎴
Total costs So Far 💵: $160
➕$60 Udemy Courses
➕$25 Domain name Registration - 2 years
➕$15 VPS Monthly fees from DigitalOcean
➕$60 MidJourney Subscription
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
A huge thanks to you if you read this far!
You can get involved by giving feedback and suggestions to improve my experience and follow me for updates.
If you have any suggestions, please leave them in the comments or send them via DM.
Thank you.
Week #2 Notes 👇 - #buildinpublic
12 days in. | Currently, I am at 13% of my Goal.
🎴🎴🎴🎴🎴🎴
Summary 📝:
During Week 2, I found my research to be very interesting. Initially, I thought I was done with my research, but I later realized I had only skimmed the surface. Therefore, I am currently in the process of gathering more information.
My main objective is to identify the crucial events, personalities, and cultural shifts I will present during each era. With that in mind, I have collected reference images, articles, books, and documentaries.
I explored the origins and migration patterns of the Yoruba people, as well as their early settlements, kingdoms, culture, and leadership.
While researching, I developed a keen interest in the art of Benin Bronze, tribal markings, traditional dress, and adornments. I want to better showcase the flourishing of Yoruba art and culture during this period.
On the technical side, I could finish studying two of the three courses I had purchased. I aim to complete the third one in the coming week.
➕ Midjourney Pro Learn The Most Recent Midjourney V5.2 AI Art Generation - Next watch
Finally, I created monogram logotypes for the website but have yet to decide which to use.
Let me know which one you like.
🎴🎴🎴🎴🎴🎴
The Goal 🔔:
➕ Use AI to generate art depicting the Yorubas' culture from the 17th century to the present in a time series or era-based format.
🎴🎴🎴🎴🎴🎴
The Strategy 🧠:
➕ Understand how GANs and Ultra Realistic AI Arts work.
➕ Discover Yoruba artists representing various styles and contributions to Yoruba art, including the Lucumi from Cuba and Brazil.
➕ Categorize the time period into four sections, namely pre-19th Century, 19th Century, 20th Century, and 21st Century
➕ Once I have a collection of creatives, the next step is to develop a portfolio website to showcase and receive feedback.
🎴🎴🎴🎴🎴🎴
The Plan 🗓:
➕ Stage 1 - Research and Discovery: 13% Done
➕ Stage 2 - Consume and Activate
➕ Stage 3 - Showcase Results
➕ Stage 4 - Feedback and Share Lessons Learned
🎴🎴🎴🎴🎴🎴
Next Steps 👀:
➕ Complete the final course
➕ Explore @pixio_ai to see if they can help with a GAN LLM, as Ross @tsi_org mentioned on Innovators Think Tank with @JoshMHellman
➕ Try some baby prompts on MidJourney; missed this last week
➕ Explore the colonial period, the Yoruba Diaspora, and the Modern Era.
➕ Finalize the art presentation structure
➕ Finanlize website logo
➕ Configure web server for the portfolio website
➕ Configure DNS, Host, and throw a coming-soon landing page
➕ Subscribe for MidJourney +or Pixio
🎴🎴🎴🎴🎴🎴
What I've Learned 🤓:
➕ Yoruba has the longest recorded internal civil war in history, 15 years and 227 days, to be exact. WOW!
➕ Fascinating works of renowned artists, movements, and styles such as Frida Kahlo (Surrealism/Realism), Banksy (Street Art), Kehinde Wiley (Modern/Realism), and Yinka Shonibare (Contemporary)
➕ MidJourney is quite interesting, and I am feeling comfortable with it and might subscribe this week
________
The Challenges 😥:
➕ Time management: I am getting clumsy.
➕ Still struggling to find a balance between Juggling work + family & friends + entrepreneurial endeavors.
🎴🎴🎴🎴🎴🎴
Costs So Far 💵: $100
➕ $60 Udemy Courses
➕ $25 Domain name Registration - 2 years
➕ $15 VPS Monthly fees from DigitalOcean
🎴🎴🎴🎴🎴🎴
Revenue So Far 💰: $0 - Non-profit venture
🎴🎴🎴🎴🎴🎴
The plan is to report my journey every week.
You can follow me to keep up-to-date.
Brain dump completed - if you read this far, here is how you can help:
➕ Engage with the content if you like it
➕ Provide advice;
I'm building this publicly, and I feel super vulnerable, so if you have any advice, ideas, or suggestions, please throw them in the comments 👇or DMs. 😊
Thank you.
@sector00007 It’s all about perspective!
Mandela during the apartheid was seen as a terrorist while to his people, a celebrated HERO!
Life is always two sided!
Nigeria's Confirmed Cases Of Coronavirus Jumps To 5162, See States Affected
Coronavirus
Nigeria has confirmed 1...
https://t.co/aQNUKWSB2d
https://t.co/aQNUKWSB2d