Today in Generative Media

AI image contest jeered; Adobe says genAI won't kill graphic design; genAI isn't cheap

• Photo Competition Slammed for ‘World’s First’ AI Image Contest (PetaPixel)

• Just as cameras didn't end painting, generative AI won't kill graphic design, says Adobe exec (TechSpot)

• The generative AI bill is coming due, and it’s not cheap (FastCompany)

• Image Restoration with DA-CLIP (HuggingFace)

• Skin texture and photos look like photos is something we worked on hard at Adobe in Firefly Image 2. We also worked on "text alignment", which means that if I ask for thumb up, I want just one thumb up. (X) This is probably the best model (other than tinkering with SD and loras) that generate normal people faces and with amazing variations. (X)

• SweetDreamer: Aligning Geometric Priors in 2D Diffusion for Consistent Text-to-3D (project page)

• My new favourite 3D creation workflow 😍

  1. Text-to-image with Stable Diffusion XL (on TPU!)

  2. Image to 3D with Gaussian Splatting

  3. Endless possibilities with the output 3D model (X)

• Another example of how I can turn a sketch into a video in a matter of minutes using sketch to Vizcom, Leonardo AI image to image, Runway or Pika labs. (X)

• Meta’s AI personas are here. Buckle up y’all (X)

• Text-to-vector is something everyone was waiting for! You can find it in Illustrator (beta). But the best part is that you can make them 3D! (X)

• State of the Art on Diffusion Models for Visual Computing (arXiv)

• Stable Diffusion 2022 ➡ SDXL 2023 (X)

• PhotoAI.com posts hitting TikTok (X)