My episode on Ethics, AI, and the Future is out on
@MasterClass
!
I’ve been advising business owners and C-suite leaders on this subject for years, including in front of
@LightspeedHQ
top 100 clients during a private event last November in Montreal.
🤯 You can now “remix” your txt2img AI prompts!
One of the biggest issues with AI generated images relies on the artists’ workflow and the controls they can have over their creation. Every new prompt usually creates a unique piece of art, thus making it hard to tweak the result.
Generating consistent characters is now finally possible!
I break down the full process in 4 easy steps in this thread (full tutorial with AI prompts and screenshots):
Trying
#ChatGPT
’s new Advanced Voice Mode that just got released in Alpha. It feels like face-timing a super knowledgeable friend, which in this case was super helpful — reassuring us with our new kitten. It can answer questions in real-time and use the camera as input too!
😍 OpenAI new Dall-E model is even better at hyper realistic images.
It isn’t finished and the research preview is only accessible to a limited number of users, but I wanted to start sharing some of my favorite explorations.
#openai
#dalle
#dalle2
#openaidalle
#openaidalle2
#ai
🤯 This video has been generated and edited using artificial intelligence.
🤖 But we’ve been hearing the word “AI” a lot lately… So what exactly do I mean by that?
I finally got access to the
@LumaLabsAI
beta! A brand new iOS mobile app and web experience to capture and share
#NeRFs
(
#NeuralRadianceFields
).
I made a video breakdown of the process so you can discover the user-friendly interface!
View my
#NeRF
here:
🤖 I asked an A.I chatbot to write a prompt to generate A.I art 🤯
ℹ️ OpenAI just released ChatGPT, a chatbot powered by their davinci-003 model. I obviously wanted to try it as soon as possible, but I was not ready for the quality of its output.
🧵[Thread]
Midjourney Outpainting New Features.
I created this video entirely using my iPhone: Midjourney’s Discord Bot + Capcut. After dozens of iterations, I wanted to showcase the latest capabilities offered by
#Midjourney
’s new outpainting features:
After spending days trying to figure out how to import
#pointcloud
files to Blender AND animate them using
#geometrynodes
WHILE maintaining the original scanned colors… I finally got the result I wanted thanks to some help from my colleague and friend
@erindale_xyz
.
#3dscanning
🤖 Would you interact with A.I tools this way?
• Spatial mind map (hello Jarvis) to navigate through a history of my iterations
• Node-base XR interface (2D/3D/AR/MR/VR compatible)
• With multiplayer to co-create with other artists
#aiart
#artificialintelligence
@ShapesXR
🥽
I don’t consider myself comfortable enough with geonodes yet to follow
@NodevemberIO
this year, but instead I decided to follow
@erindale_xyz
’s tutorials (he is live now on Youtube btw) and learn from scratch 🙌🏽
Here is something I did last night following his process!
New
@Blender
AI addon just dropped:
@ai_render
(with the power of
@StableDiffusion
).
It allows you to turn your 3d scene into an Al generated image.
We just published this video with
@polygonrunway
on IG/Tiktok/Youtube. Follow them to learn 3D illustration (link in their bio).
Navigating in
#RealTime
with a
#Quest2
controller inside animated
#GeometryNodes
in a
#Blender3D
scene rendered with eevee (compared to the final render with cycles). You can also put the
#VR
headset on (great to test 1:1 immersion), or use the WASD keys to move around.
#b3d
I had the honor to introduce Unity’s AI platforms to the world during our
#Unite2023
annual conference keynote which brought 2000 amazing people in Amsterdam and is being broadcasted to millions worldwide.
#Unite2023
🤯 Forcing chatGPT to share images from the internet.
🤖 I tried this clever “hack” from Jon Radoff on LinkedIn to trick OpenAI’s A.I chatbot into accessing the internet (even though by default, it shouldn’t be able to).
👇 Find the full code in the comments and try it yourself!
The company I work for (
@unity
) just published an article on their official blog about my career journey, childhood influences, and passion projects inspired by my cultural roots.
🙏🏽 Thank you for the opportunity to tell my story. I hope it inspires some of you!
Manuel (
@ManuVision
) draws from childhood inspiration in his work as a Design Manager for Digital Twins at Unity and on a very cool dream project. 🏝️
Find out more about Manu's journey and the projects that inspire him in our latest
#FacesOfUnity
:
I’m having a lot of fun discovering new techniques with
#geometrynodes
. I’m not really following
#nodevember
prompts. Instead, I’m diving into
#blender3d
tutorials and courses from my friends
@e
.r.i.n.d.a.l.e &
@curtisjamesholt
to add new strings to my bow 🍜
👋🏾 Hey new followers, my name is Manu!
🎨 I’m a Design Lead at
@Unity
& 3D Artist from Guadeloupe.
🥽 I’ve spent the last 11 years in Montreal researching/teaching on the topics of UX, Human-Computer Interaction, Haptics, and Augmented/Virtual Reality.
👉🏾
I predicted this in my TED talk!
#ChatGPT
can now translate Guadeloupean Creole in Real-Time using the new Advanced Voice Mode that just came out in alpha.
This is huge for accessibility, cultural preservation, and representation! I can even make it work through my
#RayBanMeta
!
I completed the
#chatGPT
game that was trending on reddit. I will publish à visual story about it tomorrow using
@OpenAI
#DallE2
. Here are some previews so you can try it too.
#SharingIsCaring
.
🟣 Tome is the A.I.-powered “storytelling” app (powerpoint competitor?) launched by a startup composed of veterans of Meta’s product teams that hope to build useful business software powered by generative A.I.
Thread:
#ai
#artificialintelligence
#dalle2
#openai
#dalle
#tomeapp
🤖 Text-Based Video Game using chatGPT & Dall•E
ℹ️ Yesterday, I heard about this 11 years old kid who made Reddit’s front page with his chatGPT game idea. So I decided to try it AND visualize it using Dall•E.
#ai
#chatgpt
#dalle2
#openai
#artificialintelligence
#dalle
Breaking news:
@LumaLabsAI
just dropped their text-to-3D generator on iOS and Web (still available on Discord too). Extra tips if you’re on iOS: Export your asset in the .USDZ format to try it in AR (Augmented Reality).
#Microsoft
could soon get a return on its $1 billion investment in
#OpenAI
, creator of the
#ChatGPT
chatbot, which gives humanlike text answers to questions.
🤖 ChatGPT invented a new color called
#Blixium
🤯
ℹ️ I asked chatGPT to think of a color that doesn’t exist and it tried to describe to me with words. Then I asked it what would be the closest HEX code to have an idea of what it could look like in our spectrum.
#chatGPT
#ai
I've been creating an AI-powered comic, and here's the kicker - it's not just any comic.
I've managed to pull off something awesome: consistent, distinct characters, all crafted by AI.
This isn't just a story, it's a breakthrough in AI comics!
I’m extremely proud to have been part of the newly released “Mobile AR Development” learning pathway on the official
@unity
Learn online platform. My focus area was on teaching AR Experience Design, helping creators to learn how to develop delightful AR apps!
#XRDesign
#ARVR
🎓 Tutorial: Creating an AI Generator with
@scenario_gg
.
⚠️ This isn’t a sponsored ad. But I’m available for paid partnerships (contact link in bio)
ℹ️ Scenario allows you to train a generator based on a custom dataset, allowing you to generate infinite AI images with consistency
🏝 Low Poly Stylized Tropical Island
This particular style was achieved with a combination of painted UV textures, a bit of smooth shading, material outlines and clever
#blender3d
modifiers.
Music by
@debrouya_
- Tour du Monde
Collect here:
#LowPoly
#b3d
Step 1: Generate between 5 and 20 images of the same character to train the model.
In this case, I recommend using
@OpenAI
Dall-E 3 (through ChatGPT) as it is great at generating character sheets. Make sure to get varied expressions and angles, and describe the outfit.
Caribéanofuturisme is a museum exhibition by Guadeloupean artist Manuel Sainsily, sponsored by OpenAI and composed of multi-platform immersive experiences, incorporating both traditional and digital elements powered by emerging technology such as mixed reality and AI, […]
Unleashing the future of XR with Generative AI! ✨ Witness rapid, immersive environment creation using ControlNet and skybox diffusion by Blockade Labs 🚀🌐
@BlockadeLabs
Trying out the new
#Blender
3.6 LTS with
#SimulationNodes
. This concept is basically a node group that automatically creates directional curl vectors from gradients on which points are moving. Thanks to
@higgsasxyz
for the setup 🌀
Every 2 weeks, we collaborate in
#VR
using
@GravitySketch
! I challenged myself to create a futuristic chair concept in 1 hour inspired by some
#AIart
research I did. My goal is to bring some of these 3D creations into my upcoming
@Spatial_io
world.
🤖 OpenAI Point-E (text-to-3D on your phone!) 🤯
💭 I wanted to post this video 3 days ago but my internet connexion was not very stable back in Guadeloupe. I’m on a plane currently landing in Montreal and decided to try my luck with the Wi-Fi.
🧵Thread
#ai
#pointe
#openai
Step 3: Using
@Scenario_gg
's custom LoRA model training feature, import each image separately (here I used 12). Be descriptive in the captions, and follow the same settings as in this screenshot.
Show time for our
@Unity
Dev Day in the beautiful Brighton, UK office. Always great to connect with the community outside of social media. We got amazing speakers today such as
@AntoniaRForster
!
We had such a great turn out during our Live
@unity
Muse Demo at
@official_GDC
today! There was a huge line-up to get in as well as for the Q&A section. Lots of great questions from our users and people seemed very excited about the future of
#AI
at Unity! So proud of our team 🫶🏽
Wow, what a year it’s been! I am so grateful to have had the opportunity to collaborate with
@OpenAI
on some truly incredible projects this year, blending AI generative art with futuristic technologies like AR, VR, 3D, and haptics.
[🧵Thread]: 1/3
🤖
#Dreambooth
#Tutorial
in 60sec… on Mobile! 🤯
💬 You asked for a video breakdown, so here it is!
ℹ️ We’re playing with cutting edge technology so this solution will certainly evolve over time, but as of today, you can follow this entire process directly from your phone!
Watch my full
#TEDTalk
on Youtube (link in bio) about how
#AI
can help us achieve our full potential today. I share examples from my Caribbean culture of how AI can be used for good by combining it with nowadays technologies.
Also you can upload any song you want, so obviously… since I’m a Caribbean from the french west indies, I went ahead and uploaded 2 songs just for fun:
-
@KASSAV_OFFICIAL
#ZoukLaS
éSèlMédikamanNouNi
-
@kalash972
ft.
@THEDAMSO
#MwakaMoon
RT if you want me to test another classic
You could of course use any other generative AI tool to generate the character sheet, but after multiple tests, I found Dall-E 3 to be the best fit for this particular use case.
For example, here are some unused images generated with Midjourney using the same prompt.
Thrilled to be an official XR Design teacher at a renowned Canadian university! Join me on a transformative 10-week journey into the future of spatial computing. University reveal, enrolment details, and more information will be announced soon!🥽🎓
#Education
#XR
#SpatialComputing
Step 4: Once the model trained, start generating new images. Have fun with various environments or facial expressions.
And that's it! Keep scrolling for some additional images, tips and credits.
You can now create and animate your avatar using
@readyplayerme
with
@NaomiAnimation
v.1.5.0 on iPad! This is the same interoperable avatar you can use in most VR social apps like
@Spatial_io
!
@devonko_
This is heartbreaking, I still remember following your first C4D course a few years ago… which helped in my transition to learn Blender later (and I believe influenced my XR-oriented career path). I’m sharing this and hope you find a donor soon Devon 🙏🏽😭❤️
But now with this new update by Midjourney, we can finally regain some level of control in the process.
I can’t wait to see what people create with it. On my side, it’s a blessing as I work on my collections and try to maintain consistency across my different pieces.
📖 Storytelling is never going to be the same.
🤖 This video was created using AI. Each frame is generated from a single image input and *no text prompt* (in this case created using Stability AI SDXL 1.0 locally). This is possible thanks to Runway Gen-2.