Author: tkalcan

Posted on

Text to Image – a look at Midjourney’s parameters

The field of GenAI technologies is rapidly evolving, regularly introducing exciting new applications, features and updates. In this blog post, I’ll explore a particular text-to-image (T2I) tool that has seen some cool updates in the past couple of months.

So we are returning to Midjourney to see what the new features are, but also to give you an overview of the most important parameters which make all the difference when prompting for something more specific. You might want to read up on the basics which I have written about in this previous blog post called Harnessing AI – the Midjourney case.

Caravaggio’s painting depicting the cutest white mouse ever, eating cheese on a kitchen table, soft light coming in from a window

I’ll be using the latest version available which currently is Midjourney version 6 Alpha (April 5, 2024) moreover, my preferred method of prompting is through the web browser. This is now possible in version 6 to users who have generated more than 1000 images in Midjourney (MJ). The Discord interface is too overwhelming with all the servers, emojis and you needed to remember the names for the parameters, which in my case lead to a lot of typos. Now, with the clean browser interface, you have the prompt line at the top of the page and the parameters are buttons and sliders – very easy to use.

Screenshot of web browser interface 1
Screenshot of web browser interface with parameters

All prompting that is done through the web browser can be done through the Discord interface too, don’t worry, just remember to start your prompt with /imagine. When using parameters you need to indicate this with two dashes in front of the parameter itself. For instance, if you want to instruct MJ to create an image with an aspect ratio of 5:7, meaning that the sides are 5 units wide and 7 units high you need to write: –ar 5:7. In the browser you just move the slider to the desired ratio. If you prefer writing dashes and parameters you can still do this in the web browser. Please note, Mac has the habit of combining the two dashes, so don’t worry if in this post you see a long line instead of a double dash.

Having a common baseline is important, thus my prompting will use the default settings which you can see in the second image with the parameter options open. There is one exception, under Model the Mode should be Standard, not Raw, but we’ll come back to this. If I use Raw I will let you know in the prompt.

MJ has quite a few parameters and I won’t discuss them all here, just the ones I think need explaining and may be the most impactful ones. You can familiarise yourself with the ones I am not discussing at MJ’s own Parameter list. Here are the ones I will cover, some in more depth some less:

  • No
  • Tile
  • Character Reference
  • Style Reference
  • Stylize

Raw vs. Standard

a woman sitting in a café, frontal view
a woman sitting in a café, frontal view
a woman sitting in a café, frontal view
a woman sitting in a café, frontal view
a woman sitting in a café, frontal view –style raw
a woman sitting in a café, frontal view –style raw
a woman sitting in a café, frontal view –style raw
a woman sitting in a café, frontal view –style raw

Let’s go back to the Raw vs. Standard Mode question. Raw can be considered another parameter expressed as: –style raw. It is used in MJ’s own words:

to reduce the Midjourney default aesthetic.

If you compare the two sets of images above, you notice the difference, although both sets have the exact same prompt: a woman sitting in a café, frontal view. The first four use the Standard Mode and the last four Raw. For me, Raw is the equivalent to photo-realistic in other words, if you aim for life-like, photo-realistic images (of people) you most likely succeed when using Raw and not the MJ default aesthetic, Standard. It’s not perfect, mind you, always check for mistakes like excessive fingers, limbs etc.

No

Ever wanted to create an image but you couldn’t get rid of an element? Well, the parameter –no works quite well in certain cases, where you have something specific you want to exclude. Just add the parameter and in this case you need to add it manually even in the browser version, since this one as some others is not available as a button. The prompt for the following images is: a delicious hamburger, lush, vibrant colours, french fries –no cheese –style raw.

a delicious hamburger, lush, vibrant colours, french fries –style raw –no cheese
a delicious hamburger, lush, vibrant colours, french fries –style raw –no cheese
a delicious hamburger, lush, vibrant colours, french fries –style raw –no cheese
a delicious hamburger, lush, vibrant colours, french fries –style raw –no cheese

Tile

This parameter is more of an artistic one. It helps you to create a never ending pattern.
The image may stand out on its own, yet its full potential unfolds when used repeatedly to craft a larger composition, integrated onto an object’s surface, or utilised as a desktop background. To get a seamless pattern you need to use for instance Photoshop where you can create a pattern from the image and apply it to whatever you want.

a few simple snow flakes in the style of fractal –tile –stylize 500
a pattern created in Photoshop from the base image above

Character Reference

Now we are getting to the newer and interesting stuff. You may or may not have noticed character reference (–cref + image URL) is not on MJ’s list of parameters. In short, –cref allows you to use a reference image and to tell Midjourney you want for instance this same person with say different clothing and maybe in a different environment. To achieve better results it pays off to define the character more thoroughly, a mere a woman sitting in a café, frontal view will not necessarily yield good results. To highlight this I first used as the reference image the same woman from the café series earlier above – it’s the second image from the left side. After this you will see my second test run with a more elaborate prompt. 


a woman sitting in a café, frontal view –style raw
a woman standing in a street corner –style raw
a woman standing in a street corner –style raw
a woman standing in a street corner –style raw
a woman standing in a street corner –style raw

And here is my second –cref run with a more elaborate prompt: a 30 years Caucasian woman with black hair standing in a street corner. The more you define the character the less you give MJ room to come up with variations, for instance having Caucasian in the prompt you eliminate characters from other ethnic backgrounds, the same goes for the hair colour etc. Well, as you can see it’s not nearly perfect, but this is something you can now work with. Try a couple of Reruns and pick the best matching image from the iterations.

a 30 years Caucasian woman with black hair standing in a street corner –style raw
a 30 years Caucasian woman with black hair standing in a street corner –style raw
a 30 years Caucasian woman with black hair standing in a street corner –style raw
a 30 years Caucasian woman with black hair standing in a street corner –style raw
a 30 years Caucasian woman with black hair standing in a street corner –style raw

Despite a specific prompt, it’s good to understand that prompting and re-prompting, doing variations is essential to achieving good results – especially when you prompt for something particular. Some sort of expectation management is in place here, I believe. It’s illusory to expect that all four images depict the very same character as referenced to and it’s actually impossible, if you think of it, there exists no such person after all!

Style Reference

Similar to character reference, style reference (–sref + image URL) allows you to use a reference image to tell Midjourney you want to transfer the style of the reference image to the new image. Now, style includes stuff like the over all feeling, colour scheme and so on, but not necessarily the style itself and certainly NOT the objects or subjects.

Style as a term in this particular case is problematic as there is the possibility of confusing two different usages of style when working with MJ. When prompting for a specific art style like in the very first image in this blog post: Caravaggio’s painting depicting the cutest white mouse ever, eating cheese on a kitchen table, soft light coming in from a window. You could also prompt: The cutest white mouse ever, eating cheese on a kitchen table, soft light coming in from a window in the style of Caravaggio’s painting. In my (granted, no so extensive) testing I have not been able to transfer Caravaggio’s style as such to new image using –sref. Instead the new image would receive the over all feeling and colours of the reference image. I say this – again – to curb expectations. Furthermore, –sref doesn’t work too convincingly when referencing from Standard mode to Raw mode or vice versa.

My reference image is the snow flake from the Tile section above and I applied it to the second case (woman in her thirties) in the character reference section. As you can see some of the colours and ornamental elements are clearly depicted in the new image.

a 30 years Caucasian woman with black hair, sitting in a café, frontal view –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –stylize 1000

The second image I applied the reference to is the burger. Here too you can see the influence of the reference image, be it in a creative way, for instance in the last one 🫐.

a delicious hamburger, lush, vibrant colours, french fries –stylize 1000
a delicious hamburger, lush, vibrant colours, french fries –stylize 1000
a delicious hamburger, lush, vibrant colours, french fries –stylize 1000
a delicious hamburger, lush, vibrant colours, french fries –stylize 1000

Stylize

And finally, we come to Stylize (–stylize or just –s). In the browser version you have it as a slider called Stylization under Aesthetics. According to MJ Low stylization values produce images that closely match the prompt but are less artistic (https://docs.midjourney.com/docs/en/stylize-1). My interpretation of this is that the lower the value the “more raw” the image becomes and the higher the value the more freedom MJ is given to apply an artistic touch to the image. Compare the following two series.

a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –sylize 0
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –sylize 0
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –sylize 0
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –sylize 0
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –stylize 1000
a 30 years Caucasian woman with black hair, sitting in a café, frontal view –style raw –stylize 1000

How do you interpret the clear difference between these two series? I think it is a significant one. The ones with low stylize values look more ordinary, like you and me as opposed to the ones with a high stylize value that could easily be models, used in commercials, tv ads or on a runway. Now, it seems that our society’s current beauty standards would consider the stylised ones to be more beautiful.

Before I let you go I want to make the following point. It’s crucial to reflect on an important issue within GenAI: bias. Although Midjourney has made notable progress in recognising and addressing diversity, including but not limited to gender, ethnicity, and age. However, true inclusivity extends far beyond these categories. It encompasses understanding and actively working against biases related to generational differences, sexual orientations, religious and spiritual beliefs, disabilities, and socioeconomic backgrounds. When utilising any image creation tool, it’s important to apply the same level of critical scrutiny and questioning that we use with text-to-text GenAI systems (ChatGPT, Claude etc.) to the images and texts we get.

Thank you for reading, happy prompting!

Posted on

Hyytiälä forest station – workshop retreat

Global campus at Hyytiälä forest station

A quick update on the Global campus team.

We visited our university’s Hyytiälä Forest Station for several intensive workshop sessions on March 18 and 19. We concentrated our efforts on fine tuning Serendip‘s first episode called Boreal forest. New to Serendip, our immersive virtual adventure? Make sure to view the trailer and read about it on the site. In addition to our work we also dedicated time to team building and bonding activities.

Global campus at Hyytiälä forest station
Unfortunately due to image size ration two of our team members were “cut out” from the header image – here you have all!

The location and facilities are ideal for this type of work, where the team remains together for an extended period, and all services, including accommodation, food and social activities (sauna!), are organised by Hyytiälä forest station.

I created a scenario type Thinglink called Workshop retreat at Hyytiälä forest station based on the images I had taken with a drone, 360° camera and smart phone.

If you’re new to Thinglink, please take a moment to familiarise yourself with the following guidelines: For the best viewing experience, I recommend using a large screen. The 360° images allow rotation, offering a comprehensive view of the surroundings. Some images include tags with detailed information. To navigate to the next image, click the ‘Proceed’ button located in the upper right-hand corner.

You can access the material through the following links according to your preferences. In any case, remember to go full screen for maximal immersive experience and in case you have a VR headset handy to use it!

With an Accessibility player

View the Thinglink in the web browser

or embedded below, in this post:

Enjoy!

Lake Kuivajärvi, Hyytiälä forest station
Lake Kuivajärvi, Hyytiälä forest station

Posted on

Kwizie – passive video watching bye bye

EDIT 22.2.2024 – DISCLAIMER – Global campus has received full access to Kwizie for testing purposes. The findings have not been influenced by Kwizie.

The Interactive Journey with Kwizie

In an era where digital innovation is at the forefront of educational transformation, Kwizie has the potential to redefine the way we engage with video content for learning. This platform bridges the gap between passive viewing and active, gamified learning experiences. Here’s a quick look at how Kwizie could reshape educational engagement through its innovative features and user-centric approach.

Transforming Passive Videos into Interactive Learning Experiences

Kwizie makes learning more dynamic and interactive. By converting in theory any video into a comprehensive quiz, it introduces a novel way to learn, catering to diverse subjects and languages. This flexibility is a testament to Kwizie’s commitment to making learning accessible and engaging.

Who is it for? Obviously teachers will benefit from this handy and user-friendly tool. In as little as 8 mouse clicks you can prepare an Instant Quiz – just like that. It takes some more clicks if you want to customise the quiz and have more control over the number of chapters and questions.

But a life long learner just as well can benefit from this tool. Say you have a concept you always wanted to learn properly. In my case it was the internal combustion engine, how does it work? I never really cared about this, but think it is part of general knowledge. Now with Kwizie I learn this in minutes. Care to test it yourself? For this purpose I chose a different topic, in the true spirit of sustainability – Global campus’ main theme: How does composting works? Have a go!

Kwizie live tab
Kwizie, live tab

Core Features Unveiled:

  • Multilingual and Multifaceted: With support for numerous languages, Kwizie ensures that learners can access content in their preferred language, breaking down barriers to education.
  • Optimised for Mobile Learning: Recognising the importance of mobile accessibility, Kwizie delivers a pleasant experience across devices, ensuring learners can engage anytime, anywhere.
  • Customisation at Your Fingertips: The platform offers a variety of customisation options, allowing educators (the Quiz Master) to tailor quizzes to their audience’s age and learning objectives, providing a personalised learning experience.
  • Effortless Sharing Mechanisms: Sharing knowledge has never been easier, Kwizie’s uses  QR codes for quiz distribution, fostering a collaborative learning environment.

 

Kwizie – How to, phase 1

Kwizie – How to, phase 2

Kwizie – How to, phase 3

Kwizie – How to, phase 4

Uncovering Kwizie’s Potential

To truly understand Kwizie’s impact, I embarked on a comprehensive testing journey, exploring its capabilities across a spectrum of videos and subjects. From environmental science to theoretical physics, the platform’s versatility was put to the test, revealing insightful nuances about its functionality and user experience. Here are my takeaways:

Insights from the Field:

  • Ease of Use: Creating quizzes is a breeze, thanks to Kwizie’s user-friendly interface that guides you through the process, from video selection to finalising quiz questions.
  • Interactive Learning: The platform’s use of timers adds an element of excitement to quizzes, though the option to pause between questions would enhance user control.
  • Educational Value: Kwizie excels in reinforcing learning objectives, with automated chapter summaries and key concepts highlighting its utility as a robust educational tool.
  • Room for Improvement: While already an impressive tool, I have encountered  a few things that in my opinion need attention. I found found some Youtube videos that did not work as expected eg. Analysing video phase never finished or when in the Live tab Creating a new Quiz was not possible. Accessibility wise there is an issue with failing WCAG AAA, a minor issue in the field of colour contrast I admit, but something easily corrected. I have reported back to Kwizie about my findings.

Reflections on Kwizie’s Educational Impact

Through testing, Kwizie’s role as an innovative educational platform became evident. Its  LMS integration (via API) and the ability for learners to contest answers exemplify its potential to not just educate but to also engage learners in meaningful ways. The call for a broader array of question types and improved accessibility features presents an opportunity for Kwizie to further refine its offerings.

Envisioning the Future of Digital Education with Kwizie

As we look towards the future, Kwizie offers a platform that not only enhances the learning experience but also empowers educators and learners alike. Its ongoing evolution and adaptation to user feedback will undoubtedly continue to shape the landscape of digital education.

Wrapping Up: Kwizie as a Catalyst for Educational Evolution

Kwizie’s digital education platform and its ability to transform video content into interactive quizzes represents a meaningful step forward in educational technology, offering new pathways for learning that are both engaging and accessible. As digital education continues to evolve, platforms like Kwizie will play a central role in shaping the future of learning, making it more dynamic, inclusive, and effective for everyone involved.

A short walk through

Finally I recorded the making of one quiz, so you can see how easy it really is.

Do you want to take this quiz too? No problem, here is the quiz about sauna in English.

 

Posted on

Pushing the boundaries

Imaginative landscape from Abbott's Flatlands. Image generated in Midjourney

When an intriguing call for papers appeared exploring AI co-creation, I felt compelled to test boundaries despite having slim to none academic publishing credentials. The concept resonated instantly, though self-doubt crept in studying full details. Could conversational technology collaborate on speculative scholarly work? Curiosity won out over uncertainty’s paralysis. If nothing else, illuminating ethical application merits investigation.

It was a Special Issue Call by the Irish Journal of Technology Enhanced Learning (IJTEL) with the title: The Games People Play: Exploring Technology Enhanced Learning Scholarship & Generative Artificial Intelligence.

I chose Claude, an AI assistant from Anthropic, entering an intensive weekend iteration. There were three options to choose from 1. Position Paper, 2. Short Report or 3. Book Review. I went with the book review. I fed Claude an 1884 novel called Flatland: A Romance of Many Dimensions by Edwin Abbott. Claude rapidly generated an abstract and book review excerpt about Flatland’s dimensional metaphors. However, hurried passes produced explanations minus critical analysis to create cohesion. Through clear prompting, I pushed Claude to incorporate additional theories, doubling the length of certain passages. It relied completely on my explicit redirects to shape fragments into cogent framing. After ten iterations I felt confident we had a useful book review.

Our accepted article examined generative AI’s promise and pitfalls, affirming Claude’s usefulness accelerating drafting under firm direction. But truly comprehending nuance and context without significant human oversight appears premature. Still, well-defined augmentation roles provide productivity upside versus total autonomy today. In other words, the current sweet spot for AI writing tools involves utilising their ability to rapidly generate content under a researcher’s close direction and oversight, rather than granting them high levels of autonomy to complete complex tasks alone start to finish.

More pressingly, this collaboration underscored ethical questions arising as generative models gain sophistication. If AI meaningfully impacts literature reviews, translation works or even initial thesis drafting one day, how can scholars utilise those productivity benefits responsibly? Tools excelling at prose introduce complex attribution and usage monitoring challenges threatening integrity.

Rather than reactively restricting technology based on risks, proactive pedagogical probes enlighten wise integration guardrails. Insights from transparent experiments clarifying current versus aspirational capabilities inform ethical development ahead.

Imaginative landscape from Abbott's Flatlands. Image generated in Midjourney.
Imaginative landscape from Abbott’s Flatlands. Image generated in Midjourney.

Forward-thinking educators can guide this age of invention toward positive ends by spearheading ethical explorations today. Our thoughtful efforts now, probing human-AI collaboration’s realities versus ambitions, construct vital foundations upholding academic integrity as new tools progress from speculative potential to educational reality.

We have the power to shape what comes through asking tough questions in times of uncertainty. As educators we shoulder the responsibility to model how inquiry protects core values even amidst rapid change. And through ethical leadership, we just might uncover new sustainable and inclusive ways to progress.

Want further reading on this topic?

Ethics of Artificial Intelligence – UNESCO

Ethics guidelines for trustworthy AI – by the EU

Ethics of AI – a MOOC by the University of Helsinki

If you are interested in reading my more personal account about this project you can do so here.

Posted on

Introducing the Visual Consistent Character Creator: A New Era with GPT Builder

Introduction

Welcome to an exciting exploration of the innovative new GPT Builder tool (by OpenAI) and my ambitious first project with it — the Visual Consistent Character Creator, an attempt to unlock the holy grail of the text to image sphere. This pioneering tool represents a major leap forward in AI-assisted creativity, combining GPT Builder’s capabilities with the user’s unique imagination.

What exactly does it do? The GPT Builder assists you in creating your own personal AI assistant tailored to your needs – in a nut shell.

In this case, my aim was not just to create a character generator, but to come up with a tool that allows me to construct aesthetically pleasing and above all consistent looking AI-generated characters. As a cherry on top I had the GPT Builder formulate the prompt in a Midjourney readable syntax.

John, a fictional character generated using a tailored GPT with OpenAI’s GPT Builder. The initial generation of John by DALL-E.

The Innovative Process: Enabling Detail and Consistency

My Visual Consistent Character Creator enables this through three key capabilities:

  1. Comprehensive trait selection — this allows for diverse and highly customised characters.
  2. Sequential, step-by-step trait selection — this ensures (or at least strives to achieve) coherence and precision in line with GPT Builder’s innovative approach.
  3. Quality check through AI-enhanced portraits using DALL-E — as a first step to ensure some level of consistency has been achieved before moving to Midjourney (or other text to image generators).

By combining these strengths, my tool can cater to a wide spectrum of creative needs while maintaining visual consistency and artistic flair. Let’s have a look at the process.

The Innovative Process: Detail and Consistency

1. Comprehensive Trait Selection At the outset, I focused on defining a wide array of character traits, mainly physical attributes (and some secret ingredients I am not revealing). I created a template for this, a matrix of sorts. This was done keeping in mind the need to match the high standards of character portrayal as seen in Midjourney. Every trait was carefully chosen to ensure that my GPT Builder could cater to diverse creative needs while maintaining a high level of detail and visual consistency.

2. Sequential Interactivity for Enhanced Precision A standout feature of the Visual Consistent Character Creator is its methodical, step-by-step trait definition process. Reflecting the innovative approach of the GPT Builder, this process ensures that each character trait is not only distinct but also contributes to a coherent overall portrait. This phase was somewhat tricky as the GPT Builder, although always complying, not always “remembered” my instructions and occasionnally would show signs of hallucinations.

3. AI-Enhanced Portrait Prompt Meeting Midjourney Syntax In the final step, after the Visual Consistent Character Creator has summarised the character’s traits and the user has confirmed them I instructed the tool to generate two types of portraits — a detailed close-up and a full-body image with DALL-E which conveniently sits in this workflow as it is part of OpenAI’s ecoystem, so the user never has to leave the browser window. At the very end the Visual Consistent Character Creator creates a prompt the user can copy&paste into Midjourney.

After some testing however, I have to say that the consistency of the characters is quite impressive, but only when generating with DALL-E. When exporting the prompts to Midjourney, the consistency is less evident. A major advantage of integrated DALL-E in ChatGPT is the possibility to discuss the result with GPT and ask for modifications on the generated image. This is huge!

John, a fictional character generated using a tailored GPT with OpenAI’s GPT Builder. The initial generation of John by DALL-E.

John, a fictional character generated using a tailored GPT with OpenAI’s GPT Builder. The initial generation of John by DALL-E.

The Potential Impacts: Opportunities and Challenges

By significantly enhancing the character design process, a visual character designer assistant like I just build with GPT Builder could revolutionise creative sectors like gaming, animation and graphic novels. The ability to quickly generate consistently looking, detailed and high-quality characters could greatly accelerate production and encourage more experimentation.

However, this also risks reducing human input in creative roles. As AI becomes increasingly capable of mimicking human artistry, important questions around originality and authenticity arise. While AI art tools offer exciting new opportunities, maintaining a balanced perspective regarding their applications will be vital so that we can benefit from their potential while responsibly managing their risks.

Overall, as an innovative new frontier in AI-assisted creativity, this tool promises to take character design into an exciting new era. By harnessing its capabilities thoughtfully, we can unlock immense creative potential.

Screenshot of ChatGPT interface creating iterations of John, a fictional character generated using my tailored GPT with OpenAI’s GPT Builder. Generated by DALL-E.

Posted on

AWEXR 2023, Vienna

AWEXR 2023 - Vienna

 

Yesterday and today (24. and 25.10. 2023) I attended the AWE XR Europe in Vienna. As usual when I travel, I chose to walk as much as possible and avoid public transportation. This time too, I took the train from the airport to Wiener Mitte and from there I made my way to the event venue, a distance of about 6km. Walking allows me to truly experience a city – its beat, vibe, smells, and soundscape. It’s the best way to get a feel for the local culture.  Exploring on foot ensures at least a minimum of movement during conferences and fairs –  and take in the sensory experience of a new place. Ironically, after deliberately immersing myself in the sights, sounds and smells of Vienna, I then spent the much of the event dealing with virtual and augmented worlds.

AWEXR 2023 - Vienna AWEXR 2023 - Vienna

The exhibition area at AWE XR Europe had a different feel compared to Laval VR earlier this year, with fewer exhibitors overall. I felt at AWEXR the focus was primarily on technical and engineering XR applications. There were exceptions like for instance hixr’s Time Travel Berlin and Chronopolis.

AWEXR 2023 - Vienna

Some notable observations

The playground area featured different XR experiences like Artivive and Nettle VR. Interesting exhibitors included Copresence app for virtual collaboration, Cognitive 3D for 3D modeling and Ikarus 3D’s 3D modeling software and product visualisation in general.

For me the one of the standout exhibits at AWE XR Europe was Time Travel Berlin’s, an immersive XR experience that virtually transports users back in time to 1920s Berlin. Unlike most other exhibits focused on technical demonstrations, both Time Travel Berlin and Chronopolis highlighted the humanities applications of XR through an incredibly detailed historical recreation in the case of Time travel Berlin.
Participants are immersed in a vivid simulation of the vibrant Pariser Platz outside the Brandenburg Gate, populated with period vehicles, hotels, and pedestrian crowds. The meticulous attention to accuracy and human-centered storytelling made it feel like walking through history. It showed that while XR enables engineering feats like 3D modeling, it can also profoundly enhance fields like education, heritage preservation, and narrative experiences. This innovative humanities-focused use of immersive technology was a refreshing change from the predominant tech demos. As one of the few exhibits bridging STEM and the humanities, it was undoubtedly a highlight of AWE XR Europe.

Another interesting exhibitor was Holonet, a Croatian startup offering a VR collaboration platform using realistic hologram avatars based on user photos. Their system supports see-through capabilities on glasses that have the feature. This creates a much more immersive experience than conventional VR avatars.

Delta Reality, another Croatian company, focused on delivering exceptional, meticulously crafted XR experiences. Their talented team brings together experts across technology, design and creativity to push the boundaries of immersive content.

AWEXR 2023 - Vienna

Overall, AWE XR Europe provided a great opportunity to see the latest innovations in XR from both startups and established players. The perceived smaller size enabled more intimate networking and discovery compared to other larger events. I was able to connect with key players in the European XR ecosystem and bring back valuable insights for our team. A highlight was reconnecting with James Mifsud of ArborXR, whom I had met earlier this year at Laval XR. It was great to catch up with James and attend the afterparty together. The impromptu dinner with other attendees especially the ones from Croatia was especially enjoyable and inspirational. Events like AWEXR enable these valuable personal connections within the close-knit XR community.

AWEXR 2023 - Vienna AWEXR 2023 - Vienna

AWEXR 2023 - Vienna AWEXR 2023 - Vienna

Posted on

UniPID collaboration

Empowering Education through Artificial Intelligence

Earlier this year, I had the privilege of conducting an online workshop on the use of artificial intelligence (AI). The event, organised by UniPID, is in line with the broader vision of Global Campus to harness the power of AI and bring it to professionals in higher education.

I introduced the attendees to innovative tools like ChatGPT and Midjourney. ChatGPT, for instance, offers a conversational approach that can assist teachers in course design, from creating exercises and tasks to developing syllabi. On the other hand, Midjourney stands out with its unique ability to generate images that closely represent real-life objects, enabling teachers to bring their imaginative ideas to life in visual formats.

We delved into the potential of AI in creating personalised learning experiences, ensuring that students from diverse backgrounds can receive quality education tailored to their needs. Furthermore, we touched on the ethical implications of the use of AI in education. Global Campus emphasises the importance of responsible AI use, and it was enlightening to engage with educators and stakeholders on this critical topic.

This workshop was a reminder of the incredible impact we can achieve when we collaborate, share knowledge, and drive innovation.

In conclusion, my experience at the workshop was both enlightening and inspiring. We’re not just envisioning the future of education; we’re actively shaping it through workshops like this. I’m excited about the possibilities that lie ahead.

Read UniPID’s take on the workshop: Teachers’ workshop: the use of Artificial Intelligence in virtual education.

Posted on

Laval Virtual

My Journey Through Laval Virtual: The Quest for Immersion in the World of Virtual Reality

To give you some context you will find a brief summary about myself at the end of this post.

As an enthusiast of the concept of virtual realities, I was eager to attend Laval Virtual, the premier event showcasing the latest advancements in VR, augmented reality (AR), and mixed reality (MR) in short XR in as someone put it humbly the capital of VR! I couldn’t wait to dive into the world of cutting-edge technology, engaging discussions, and artistic creations. Throughout the event, I found myself constantly questioning: “What truly defines immersive experiences in XR?”
This is my personal journey to and through Laval Virtual as I explored innovative brands, participated in thought-provoking discussions, and found inspiration in the arts, all in the quest for immersion. But before I let you read the text itself I offer you the possibility to immersive yourself in a few 360° photos from various stages of my trip.

You can freely navigate with your mouse within the 360° image, there is even a hotspot you can click on and move on to the next image.

Note: While writing this on Friday 14 April, Laval Virtual is still on-going, but unfortunately I have to catch a train and then plane to make it home still today. Greetings from Paris Charles de Gaulle airport.

Discoveries

During my first day on Wednesday 12 April, I was thrilled to explore the innovative brands and technologies being showcased. As a newish VR enthusiast, I was particularly impressed by Movella‘s Xsense,  found the approach L.A.P.S. is taking interesting and thought Olfy had taken the next logic step in bringing one more of our senses to the XR table, very refreshing. Xsense’s groundbreaking work in motion capture, live avatar performance solutions, and sensory integration expanded, seeing it live, my understanding of what’s possible in VR and AR experiences.
L.A.P.S.’s solutions on the other hand enable real-time facial expression tracking, which allows avatars to mirror the movements and emotions of the performers in real-time. Finally Olfy is a virtual reality system that simulates smells to create a more immersive experience. In their own words: The sense of smell allows virtual reality experiences to be more engaging and immersive. Our goal is to enhance the emotions and effectiveness of virtual experiences by allowing you to experience them 100% (Olfy).

Thought-Provoking Conversations

As I attended the various discussions and keynotes, the Immersive Digital Learning topic stood out as a highlight for me, particularly the engaging panel discussion featuring Anaïs Pierre, Bogdan Constantinescu, Jayesh Pillai, and Thierry Koscielniak. Anaïs passionately emphasised that technology serves as a tool, and we must prioritise learning goals before seeking the appropriate technological solution. The use of tech tools must be purposeful and meaningful.
This is exactly how I feel about the use of technology in general. Content and (in my case usually) learning goals first, otherwise your course or product will not find its full potential.

I listened to Kent Bye’s fast-paced talk on the topic of XR moral dilemmas and ethical considerations. He discussed several crucial concerns that we should all be pondering, including the digital divide in access to XR technology, threats to privacy (e.g., biometric data), apparently there is no legislation on this and similar issues.

Furthermore a new concept I hadn’t heard before: mental privacy which is part of a proposed set of rights called the Neuro rights. Mental privacy refers to the protection of an individual’s thoughts, emotions, and mental processes from unauthorised storing and access, particularly in the context of novel technologies that can potentially monitor or manipulate these aspects of the human experience. This really jump-started my brain and I am still processing all the possible implications and reflecting on the ethical and practical considerations of using XR technologies in education and beyond.

During another panel discussion, I was introduced to the concept of eco-design, which involves reducing the energy footprint of eg. VR headsets. It became clear that as VR technology evolves, we must be aware of and address the energy-intensive nature of these devices to create a more sustainable and ethical future. By incorporating eco-design principles, we can minimise the environmental impact of VR and ensure a more responsible approach to technology.

Finding Inspiration in the Arts

Now of course Recto VRso cannot go unmentioned in this blog post. Recto VRso is a component of Laval Virtual, which this year took place partially at Le Quarante, a cultural center in Laval and L’Espace Mayenne. It showcases innovative uses of XR in the field of art and culture. Attendees can interact with some of the  installations that push the boundaries of what is possible in XR art, providing a platform for networking and inspiration in the field – further fueling my passion for the intersection of technology and creativity.

I’d like to single out one installation called Memory house by artist Jiahe Zhao.
According to the artists own words an my rough translation into English: Numerous items hold cherished memories, ranging from beloved family toys to treasured travel souvenirs. “Memory House” offers a solution for storing these memories in a virtual realm. By using 3D scanning technology, memory objects are brought into the virtual space, and through transformation by AI (artificial intelligence), they are manifested into a one-of-a-kind virtual edifice that is ideal for exploration.

And yet Recto VRso seems to be an exception. Over and over have I noticed that XR related applications are heavily situated in the engineering, training/onboarding and skills learning sectors, very little is to be found in the liberal arts world, sadly. Having an academic background in field of liberal arts myself I was excited to discover the creative applications of XR technology at Laval Virtual. I found inspiration in BavARt, an AR-specialised firm that combines art and technology in a Pokémon Go-style app.

Immersion?

As I left LavalVR, I couldn’t help but reflect on my initial question of what defines immersive experiences in VR. Throughout my journey, I discovered that immersion is not just about cutting-edge technology and realistic visuals. It also involves connecting with our emotions, bridging the gap between the digital and physical worlds, and finding inspiration in the creative fusion of art and technology. The quest for immersion is a never-ending journey, one that continuously ignites my enthusiasm for discovering the boundless opportunities that await in the world of virtual reality.

Final thoughts

This was my first conference I attend on my own, I knew no one at Laval and because I am not really an extrovert, but an observer if you will, I struggle connecting with  strangers. Most attendees were there in larger parties and would therefore communicate and entertain themselves among each other which makes it for outsiders as myself hard to mingle. I am not complaining just stating what I noticed.
Yet at the hotel I couldn’t help but be involved in socialising with the French during breakfast. As people were arriving to the breakfast they were greeting everybody and they were greeted by the ones already there. People would ask each where they’re from and what they’d do and so forth. There was active communication throughout  breakfast. I loved it

And finally, note that the venues for the Laval Virtual are kilometers away from each other and not just quelques pas (a few steps) as was repeated by officials a few times =)  Nonetheless I took this as an opportunity to walk and boy did I do some walking, a total of 27km on two and a hald days between my hotel and the three venues.

Thank you Laval Virtual et à la prochaine!

***

Background

I am fairly new to the XR world, in fact it was only last October (2022) that I had a VR headset on for the first time, imagine that! Having studied languages, folkloristics and other liberal arts subjects (at the University of Helsinki (UH)) I have noticed that I have a somewhat different approach to technology than most of my peers. I have been working in several positions at UH over the last, almost two decades including as an International exchange expert and Edtech specialist before joining the Global campus team. As a lifelong learner I started studying university pedagogy last year at UH believing it will give me a strong understanding of educational technology. Furthermore I am the chair of Una Europa’s Educational design and technology cluster. For more coherent  info on me see the Us section.

Posted on

AI-Powered Course Material Production

Introduction: The Global Campus of the University of Helsinki is committed to exploring innovative methods for enhancing educational experiences. As part of this ongoing mission, our recent “AI methods in course material production” presentation at the university’s Learning Adventure showcased the potential of cutting-edge AI technologies in creating engaging and dynamic course materials. While our primary audience was the university community, we believe these insights hold value for all EdTech enthusiasts.

In this blog post, we’ll share key takeaways from our presentation, which encompassed five sections: Text, Images, Audio, Video, and Q&A.

  1. Text: Harnessing ChatGPT’s Potential. Kicking off our presentation, we introduced ChatGPT, an AI language model developed by OpenAI. By delving into the concept of prompting, we unveiled various techniques, including Chain of Thought (CoT) methods. Highlighting the effectiveness of role prompting, we showcased ChatGPT’s self-criticism and self-evaluation features as a means to generate meaningful responses.
  2. Images: Visualising Ideas with Midjourney. Transitioning to text-to-image (T2I) generation, we presented Midjourney as an exemplary case. Demonstrating seamless integration between Discord and Midjourney, we revealed the process of creating images through prompting in Discord. For a deeper understanding of the Midjourney case, we invite you to read our earlier blog post here.
    It’s worth noting that in addition to Midjourney, there are several other AI-based applications that allow for the creation of images through text. One notable example is DALL-E, which uses a transformer-based neural network to generate images from textual descriptions. And let’s not forget about StableDiffsusion, a new AI-based technology that allows for the generation of high-quality, realistic images by leveraging the power of diffusion models. With so many exciting applications available, the possibilities for creating images through text are truly endless.
  3. Audio: Bringing Text to Life through AI. Our third segment explored the realm of text-to-audio conversion. We shed light on AI tools and techniques capable of generating lifelike speech from written text, making course materials more engaging and accessible to auditory learners.
  4. Video: Creating Dynamic Learning Experiences with AI. In the penultimate section, we investigated AI’s potential in video production. Discussing the role of artificial intelligence in crafting compelling and informative videos, we emphasised the importance of delivering course content in a visually engaging manner. In addition to Synthesia and Colossyan, there are several other noteworthy applications that are worth exploring. One such application is D-ID, which is a deep learning-based technology that allows for the anonymisation and replacing of faces with natural or fantastical looking options in videos using AI-generated imagery. With the increasing demand for video content in today’s digital landscape, these and other AI-based text-to-video applications offer opportunities for teachers and students to create high-quality videos quickly and easily.
  5. Q&A: Encouraging Audience Interaction. To wrap up our presentation, we engaged the audience in group discussions, addressing questions and concerns stemming from the event. This interactive session fostered a deeper understanding of AI’s role in education and promoted collaboration within our university community. Participants were interested in for example if it was possible to produce materials in Finnish language with these new tools and yes, usually that is also possible.

Conclusion: Embracing AI-powered tools like ChatGPT, Midjourney, and other text-to-audio and video production solutions is revolutionising the way we develop and deliver course materials. By adopting these innovations, we can create more engaging, accessible, and dynamic learning experiences for students across the globe.

AI is not taking away your job, it’s the person that adopts AI that is taking away your job!

Posted on

ThingLink basics, tags

View of a laboratory in ThingLink.

 

Greetings, avid reader! Allow me to introduce you to the delightful world of ThingLink. Where the only limit is your own imagination! If you’re looking to add a touch of finesse to your images and videos, look no further than the humble tag. These interactive buttons bring your multimedia content to life and they’re the secret ingredient of ThingLink. In this blog post, I’ll give you a quick rundown of ThingLink in the form of a video.
BTW we used ThingLink for our very first project. You can read about it in our blog entry called Message from the future.

First of all, let’s define what ThingLink is. Simply put, ThingLink is an online platform that allows you to add interactive tags to your images and videos. These tags can include text, images, audio, and video, making your multimedia content much more engaging and interactive. Whether you’re a teacher, a student, or simply someone looking to add a touch of sophistication to your social media posts, ThingLink is the tool for you. Check out the following Miro board where I have put together a very simply yet effective sequence of slides to highlight what ThinLink is.

Now, why is this relevant for our context which is higher education and Edtech and at the end of the day the learners? Well, for one, it is a very intuitive tool and it gives students control over their own learning journey. No more dull lectures or tedious presentations. With ThingLink, students can interact with the material in multiple ways, truly grasping and internalising the information. And let’s be honest, who doesn’t love a bit of control? According to Yarbrough (2019) “more visuals are not only learner preferred, training and educational professionals have identified that visuals support efficient teaching.”

A ThingLink can be considered a version of an infographic. There are ample studies supporting the claim that infographics are very powerful tools. What makes infographics and in an extended way ThingLink too, so useful? Visuals tend to stick in long-term memory, they transmit messages faster and improve comprehension to name a few (Shiftlearning).

Here is a roughly 7 min video walking you through all the tags and how to create them. In Edit mode tags can can be dragged around the base image – you can even pull a line from under a tag and anchor it to a specific point. 

In a next tutorial blog post with video we’ll have a look the settings and dive into the immersive world of 360° images and videos in ThingLink.  

In conclusion, ThingLink is the tool you didn’t know you needed. With its interactive tags and multimedia-rich approach, ThingLink empowers students to take charge of their studies and reach their full potential. So what are you waiting for? Give it a go and see the magic unfold!

BTW when storyboarding the video I had a clear vision of how to implement text to speech (TTS) with an AI voice – little did I know how NOT easy this was  Stay tuned as at some point I will write a how to post about the process of producing the above video. 

Source:

Yarbrough, J. R. (2019). Infographics: In support of online visual learning. Academy of Educational Leadership Journal, 23(2), 1–15.

Shiftlearning. Blog post: Studies confirm the Power of Visuals in eLearning.