- This event has passed.
Training LoRAs : LoRAs for A.I. Art Explained and Demystified
March 3, 2024 @ 7:00 pm – 9:00 pm GMT
We’ve unfortunately had to cancel the LoRA event, we didn’t quite get enough premium ticket holders to justify hosting the event. A minimum number is required to cover our costs.
We’re sorry for any inconvenience caused and please get in touch if you have any further questions about this event.
THIS EVENT HAS BEEN CANCELLED
If you have been exploring AI Art generation, then you have likely come across the term of LoRA (Low-Ranking Adaptations). What are LoRAs and why are they crucial to learn?
- Customization: They tailor AI image generation to precise concepts yourcheckpoint model cannot do on its own.
- Efficiency: LoRAs offer a specialization that uses files as small as 144MB,
as opposed to several gigabytes from full checkpoint models. - Learning from Less: They’re effective even with limited data, ideal for
simple concepts like cartoon characters. - Better Quality: LoRAs enhance the realism and detail in AI-generated
images that require a specific subject specialization. - Creative Potential: Mastery of LoRAs unlocks new creative possibilities
in image generation, like the combination of one or multiple artistic styles. - Edge in AI: Understanding LoRAs puts you at the forefront of AI
development.
Ken Kawashima provides an insightful presentation and demonstration to show you how to train Low-Ranking Adaptations (LoRAs) models and understand them in the process.
If you’re keen on enhancing your ability to understand and create images on
specialized subjects with greater precision and detail, this webinar is your gateway to
grasping LoRAs.
Whether you’re a seasoned AI aficionado or just starting out, our comprehensive walkthrough will empower you to unlock new potential in image generation.
We will provide an overview and explanation of LoRAs so that you can:
- Understand basics of how and why LoRAs work
- Elevate your ability to generate images on specialized subjects with better
clarity and detail. - Acquire the knowledge to train LoRAs with your unique datasets for specialized image creation tasks.
Ken will be showing his workflow on creating LoRAs which includes the use of:-
- Kohya_SS v22.4.1
- Automatic1111 v1.7.0 (Stable Diffusion)
- Booru Dataset Tag Manager v1.6.3.0
This is for you if …
- You’re looking to deepen your understanding of AI image generation.
- You’re aspiring to train your own LoRAs for your favorite subjects.
- You desire to push the boundaries of your creative projects using AI.
We are going to explain LoRAs to you even if you:
- Have never engaged with AI image generation beyond the basics.
- Have little experience with specialized AI model training.
- Are new to the concept of image generation through artificial intelligence.
This is for you if you want to accomplish…
- An understanding of how to curate and manage a higher-quality image dataset for training AI models.
- Learning to tag images to fine-tune the AI’s learning process.
- A foundation in setting up some key training parameters to optimize your
LoRA’s training effectiveness. - Identifying and avoiding common pitfalls in creating LoRAs
Webinar Details
Session 1
Date: Sunday Mar 3rd
Time: 19:00 GMT (London) / 14:00 EST (New York) / 11:00 PST (Los Angeles)
Duration: 1.5 – 2 hours
Session 2
Date: Sunday Mar 10th
Time: 19:00 GMT (London) / 14:00 EST (New York) / 11:00 PST (Los Angeles)
Duration: 1.5 – 2 hours
Webinar Ticket Options
Free ticket
- Entry to Part 1 for an introductory session only (30 minutes)
- Excludes recordings and any bonus content
Premium ticket
- Entry to all sessions of this series.
- HD recording of both webinars with optional captions.
- Bonus video by Ken “Viewport” : Complete Guide to Optimizing your Realtime Experience with DAZ Studio (worth $37)
VIP Ticket
- Entry to all sessions of this series.
- Searchable HD recording set of this mini course with optional captions.
- Bonus video by Ken “Viewport” : Complete Guide to Optimizing your Realtime Experience with DAZ Studio (worth $37)
- A 1-1 mentoring Zoom call (up to1 hour) with Ken to get your own LoRA training started.
About the presenter : Ken Kawashima
Ken is a long standing experienced vendor at Renderosity, who has also sold at Hivewire 3D.
He’s a programmer as well as a 3D artist, creating content for Poser and DAZ Studio. He’s delved into scripting for creating a range of Python plugins for Poser 12. He also works with CC3 and iClone “which is a lot of fun”.
Now he’s taken a deep dive into how A.I. generated art works and how LoRAs can benefit and fine tune this kind of artwork further.