Gareth Griffiths’ Post

With the simple additional step of importing architecural models into Google earth, we can streamline pretty effective contextual studies with this workflow. Would work well for greenfield sites, but the challenge I still run into is when there is existing infrastructure on the project site. Thoughts or solutions?

What if we turn Google Maps into anime? 🤔 I tried it using Stable Diffusion and ComfyUI. Here is the workflow: - Checkpoint model: meinamix_meinaV11 - Positive Prompt: day, noon, (blue sky:1.0), clear sky - Negative Prompt: (worst quality, low quality:1.4), (zombie, sketch, interlocked fingers, comic) - Resolution: 768 x 512 - ControlNet model: control_vllp_sd15_canny Depending on the Google Maps location, I add a country or city name in the positive prompt (e.g. Japan, New York, Paris, etc.). I used toyxyz's custom webcam node to capture a section of the screen and plug that into a ControlNet canny model. KSampler: - seed: 1 - control_after_generate: fixed - steps: 15 - cfg: 4.0 - sampler_name: euler_ancestral - scheduler: normal - denoise: 1.00 It is possible to optimize this further and make better and faster generations. Perhaps by using an LCM-LoRA, StreamDiffusion, TouchDesigner, or a model based on SDXL-Lightning. Video music from Uppbeat dot IO Space Journey by Hartzmann #AI #ArtificialIntelligence #GenerativeAI #StableDiffusion #ComfyUI #anime

Gmaps styled through Comfy, with workflow detailed...

Real-time image reinterpretation using multiple agents and shaders for an intended output (this example uses node-based comfy-ui). It makes me think about how, in the future, IP could also be a proprietary chaining of curated agents.

Enjoying playing with a new level of AI video and audio generation This is 'Field Day' More can be found on my social media: https://lnkd.in/deDjx5Zr #ai #art #aiart #digitalart #artificialintelligence #nature #landscape

This is one of those mind-blowing moments in AI Picture to Unreal Engine Scene using the new Reverie tool Move across lighting, environmental influence and style = Magic #aitools #digitaltools #unrealengine https://lnkd.in/d-_AnGMd

I was curious about creating interactive 3D scenes on-the-go, so I decided to put the Ray-Ban Stories and #LumaAI Gaussian Splatting to the test! 🕶️📲 Gaussian splat magic with a casual stroll, using Luma AI! The results were astonishing! View the Glasses Tests on your phone or browser! 👓 - https://lnkd.in/gwQ6_mxT 📲 - https://lnkd.in/gnwz2unA TIPS: -Walk around the object at least twice, capturing from different heights for a well-rounded view. -Keep a steady pace to ensure smooth video capture. The RayBan Stories make this a breeze! -Choose an environment with good lighting to highlight the details of your subject. 🌞 -Use your RayBan Stories for recording - it’s hands-free and captures a clear perspective. Upload the clips directly from your phone to Luma AI app for quick and easy Gaussian Splatting. (*Android Uses Browser) PROBLEMS?!: If the scene appears incomplete, it’s likely because some angles were missed. A thorough walk-around can fix this! The blend of RayBan Meta’s sleek recording and Luma AI’s tech has truly revolutionized how we can interact with the world, one Gaussian splat at a time! 🙌 #GaussianSplatting #3DWorlds #CreateWithEase #RayBan #RayBanMeta #3DGaussianSplatting #NERF #OnTheGo

Google presented Their First World Model: Genie, Generative Interactive Environments, an 11B parameter model capable of generating action-controllable virtual worlds described through text, synthetic images, photographs, and even sketches. Genie was trained in an unsupervised manner from unlabeled Internet gaming videos, without any action or text annotations.  Yet, it still infers an interactive world from an unseen image, where users can act in a logical way, on a frame-by-frame basis. Ot is Google’s first foundational world model, as it understands the world through its autoregressive dynamics model and spatiotemporal video tokenizer.  Here’s a kit of 8 cool open-source world model projects. https://lnkd.in/d27pzHmx    #OpenWeaver #OpenWeaverStudio #NoCode #Google #Genie #WorldModel #GenerativeInteractiveEnvironments #AI #DeepLearning 

🍻 The Process of.....A Cow walked into a bar 🎉 It is pretty straight forward. 1) Writing the prompt for the first image. 2) Saving and uploading the image to GPT-4V 3) Asking GPT-4V to describe the image 4) C&P the description into whatever image generator you are using 5) I chose to use Leonardo Ai becuase of the control that you have and the ease of use of it's UI. I will dig into exactly what I did in Leonardo tomorrow. Composition: Blackmagic Design (Davinci Resolve) Music: Created with Soundful #ai #aiimages #innovation #futuretech #leonardoai #gpt #gpt4 #branding #marketing #digitalmarketing

Recently used some AI based software to clean up a music video I produced back in the year 2000. It's amazing what can be done with old footage. It took some tweaking to keep it from looking like a video game, AI really wanted to chew on it. I managed to use the AI bits of software to fix some interlacing issues that were not properly corrected when digitizing the footage many years ago. I used my own intelligence to clean up the rest. 😎 Original video... https://lnkd.in/gMq_jkEU And the remastered version on Vimeo (better compression than LinkedIn) https://lnkd.in/gwFPdEgr