Tootfinder

Google makes its Imagen 2 image generator generally available in Vertex AI, and adds inpainting and outpainting to remove or add elements and expand borders (Kyle Wiggers/TechCrunch)
https://techcrunch.com/2024/04/09/google-releases-imagen-2-a-vi…

Google releases Imagen 2, a video clip generator | TechCrunch
Imagen 2 can now create short, four-second videos from text prompts, along the lines of AI-powered clip generation tools like Runway, Pika and Irreverent Labs.

Fortifying Fully Convolutional Generative Adversarial Networks for Image Super-Resolution Using Divergence Measures
Arkaprabha Basu, Kushal Bose, Sankha Subhra Mullick, Anish Chakrabarty, Swagatam Das
https://arxiv.org/abs/2404.06294

Fortifying Fully Convolutional Generative Adversarial Networks for Image Super-Resolution Using Divergence Measures
Super-Resolution (SR) is a time-hallowed image processing problem that aims to improve the quality of a Low-Resolution (LR) sample up to the standard of its High-Resolution (HR) counterpart. We aim to address this by introducing Super-Resolution Generator (SuRGe), a fully-convolutional Generative Adversarial Network (GAN)-based architecture for SR. We show that distinct convolutional features obtained at increasing depths of a GAN generator can be optimally combined by a set of learnable convex…

Provenance:
https://openart.ai/@kencf0618033

https://openart.ai/@kencf0618033

Kenneth Freeman's Profile and Image Gallery | OpenArt
Free AI image generator. Free AI art generator. Free AI video generator. 100+ models and styles to choose from. Train your personalized model. Most popular AI apps: sketch to image, image to video, inpainting, outpainting, model fine-tuning, real-time drawing, text to image, image to image, image to text and more!

Here’s a charming little wallpaper generator app for iPhone, with the backend written and created on @… ! https://techhub.social/@thatvirtualboy

thatvirtualboy (@thatvirtualboy@techhub.social)
Attached: 1 image I wrote 1,444 words about my latest app, Wall•Genie… in case you’re really strained for something to do. https://thatvirtualboy.com/say-hello-to-wallgenie/ #iOSDev #indieDev #ai #generativeAI

Samsung-Backed AI Image Generator Produces Nonconsensual Porn https://www.404media.co/samsung-backed-ai-image-generator-produces-nonconsensual-porn/

A #Microsoft #AI engineer has spent months testing its AI image-generator, Copilot Designer. He's disturbed by the violent/sexual content it creates & MSFT's decision not to take it offline to investigate.
On Wednesday, he notified the FTC.

Semantic Image Synthesis with Unconditional Generator
Jungwoo Chae, Hyunin Cho, Sooyeon Go, Kyungmook Choi, Youngjung Uh
https://arxiv.org/abs/2402.14395 h…

Semantic Image Synthesis with Unconditional Generator
Semantic image synthesis (SIS) aims to generate realistic images that match given semantic masks. Despite recent advances allowing high-quality results and precise spatial control, they require a massive semantic segmentation dataset for training the models. Instead, we propose to employ a pre-trained unconditional generator and rearrange its feature maps according to proxy masks. The proxy masks are prepared from the feature maps of random samples in the generator by simple clustering. The fea…

My take on the Google #AI image generator controversy, and bias in AI image generation in general:
When you ask for a generative creation (including LLMs) you WANT the AI to fill in a ton of details. Some of those are fine: backgrounds, textures, expressions, etc. Some can intersect with societal issues like race or gender.
LLMs actually can be vague –a benefit of language – but image genera…

Provenance:
https://openart.ai/@kencf0618033

This https://arxiv.org/abs/2310.07842 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csRO_…

DiPPeR: Diffusion-based 2D Path Planner applied on Legged Robots
In this work, we present DiPPeR, a novel and fast 2D path planning framework for quadrupedal locomotion, leveraging diffusion-driven techniques. Our contributions include a scalable dataset generator for map images and corresponding trajectories, an image-conditioned diffusion planner for mobile robots, and a training/inference pipeline employing CNNs. We validate our approach in several mazes, as well as in real-world deployment scenarios on Boston Dynamic's Spot and Unitree's Go1 robots. DiPP…

Israel-based BRIA, a Getty-backed AI image generation startup that licenses 1B images from 18 stock providers, raised a $24M Series A from Publicis and others (Katie Paul/Reuters)
https://www.reuters.com/technology/getty-bac…

At the risk of being crude or flippant...
Well duh! What the fuck did you think people would do with really good LLM image generation, make pictures of cookies? It's a disinformation-generator. That's what it is. That's what you've built. Don't pretend like you didn't see it coming. We know you're not that stupid, just that self-absorbed.

Google pausing AI portrait tool after complaints of historical inaccuracies | CBC News
Google is pausing an AI tool that creates images of people following inaccuracies in some historical depictions the model generated, the latest hiccup in the Alphabet-owned company's efforts to catch up with rivals OpenAI and Microsoft.

Odd things Ubuntu does
Having just booted a new VM up using the official #Canonical #Ubuntu image on #Azure, I noticed that, during the first boot cycle, it makes a call out to "entropy.ubuntu.com"
This is done by the "polliniate" package which describes itself:
pollinate is an entropy as a service client. It will fetch a random seed from one or more remote servers, and feed those into a (pseudo) random number generator. This is often useful in cloud and virtual machine environments with limited sources of entropy.
https://manpages.ubuntu.com/manpages/jammy/man1/pollinate.1.html
There's something oddly jarring about this. Not least because this call-out for "supplementary" entropy is done after generating the host's SSH keys.
I hope, hope very much, that someone with Prof in their title from a very well regarded Computer Science department has written a paper suggesting this is ok.

Israel-based BRIA, a Getty-backed AI image generation startup that licenses 1B images from 18 stock providers, raised a $24M Series A from Publicis and others (Katie Paul/Reuters)
https://www.reuters.com/technology/getty-bac…

This https://arxiv.org/abs/2310.07842 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csRO_…

DiPPeR: Diffusion-based 2D Path Planner applied on Legged Robots
In this work, we present DiPPeR, a novel and fast 2D path planning framework for quadrupedal locomotion, leveraging diffusion-driven techniques. Our contributions include a scalable dataset generator for map images and corresponding trajectories, an image-conditioned diffusion planner for mobile robots, and a training/inference pipeline employing CNNs. We validate our approach in several mazes, as well as in real-world deployment scenarios on Boston Dynamic's Spot and Unitree's Go1 robots. DiPP…

One-Step Image Translation with Text-to-Image Models
Gaurav Parmar, Taesung Park, Srinivasa Narasimhan, Jun-Yan Zhu
https://arxiv.org/abs/2403.12036 https:…

One-Step Image Translation with Text-to-Image Models
In this work, we address two limitations of existing conditional diffusion models: their slow inference speed due to the iterative denoising process and their reliance on paired data for model fine-tuning. To tackle these issues, we introduce a general method for adapting a single-step diffusion model to new tasks and domains through adversarial learning objectives. Specifically, we consolidate various modules of the vanilla latent diffusion model into a single end-to-end generator network with…

One option is to externalize these choices from the image generator. You can imagine adding a "race bit" to the input that says "if you have a character make it race X" and then roll some race dice on input. This is what it feels like Google has done.
Another option is to give the choice to a better equipped AI. Dall-E does this by having GPT rewrite prompts. This still isn't good at distribution, but can use the LLM's superior knowledge to catch things like…

Stability AI makes Stable Diffusion 3 available to developers via an API and a new Stable Assistant Beta content creation platform; the model remains in preview (The Verge)
https://www.theverge.com/2024/4/17/24132927/stable-diffusion-3-…

Stability AI is giving more developers access to its next-gen text-to-image generator
Stability AI has released the API for Stable Diffusion 3, the latest version of its text-to-image model. It’s now available to developers on its Developer Platform.

HEMIT: H&E to Multiplex-immunohistochemistry Image Translation with Dual-Branch Pix2pix Generator
Chang Bian, Beth Philips, Tim Cootes, Martin Fergie
https://arxiv.org/abs/2403.18501

HEMIT: H&E to Multiplex-immunohistochemistry Image Translation with Dual-Branch Pix2pix Generator
Computational analysis of multiplexed immunofluorescence histology data is emerging as an important method for understanding the tumour micro-environment in cancer. This work presents HEMIT, a dataset designed for translating Hematoxylin and Eosin (H&E) sections to multiplex-immunohistochemistry (mIHC) images, featuring DAPI, CD3, and panCK markers. Distinctively, HEMIT's mIHC images are multi-component and cellular-level aligned with H&E, enriching supervised stain translation tasks. To our kn…

Such fun ...
#OpenArtai

Kenneth Freeman's Profile and Image Gallery | OpenArt
AI art generator makes you 10x more creative and productive. 100+ models and styles to choose from. Support txt2img, img2img, ControlNet, inpainting, and more with Face/anime enhancement with 2x and 4x upscaling. Discover 20M+ prompts.

Overdue, but I’ve added the machinery to automatically add opengraph images to my blog entries #hugo
https://blog.lanzani.nl/2024/open-graph-hugo/

Automatically add images for Open Graph in Hugo · Technical inconsistencies blog
While using a static site generator is a low-maintenance endeavor, it also means that complex requirements need to be coded if nobody has done it before. Today, I automated image previews that can be used to display a nice preview through for Open Graph (and Twitter cards!). The Open Graph protocol enables any web page to become a rich object in a social graph. My blog, however, is text-heavy and often misses the images, so I had to come up with something else.

You Only Sample Once: Taming One-Step Text-To-Image Synthesis by Self-Cooperative Diffusion GANs
Yihong Luo, Xiaolong Chen, Jing Tang
https://arxiv.org/abs/2403.12931

You Only Sample Once: Taming One-Step Text-To-Image Synthesis by Self-Cooperative Diffusion GANs
We introduce YOSO, a novel generative model designed for rapid, scalable, and high-fidelity one-step image synthesis. This is achieved by integrating the diffusion process with GANs. Specifically, we smooth the distribution by the denoising generator itself, performing self-cooperative learning. We show that our method can serve as a one-step generation model training from scratch with competitive performance. Moreover, we show that our method can be extended to finetune pre-trained text-to-ima…

The Effects of Generative AI on Design Fixation and Divergent Thinking
Samangi Wadinambiarachchi, Ryan M. Kelly, Saumya Pareek, Qiushi Zhou, Eduardo Velloso
https://arxiv.org/abs/2403.11164

The Effects of Generative AI on Design Fixation and Divergent Thinking
Generative AI systems have been heralded as tools for augmenting human creativity and inspiring divergent thinking, though with little empirical evidence for these claims. This paper explores the effects of exposure to AI-generated images on measures of design fixation and divergent thinking in a visual ideation task. Through a between-participants experiment (N=60), we found that support from an AI image generator during ideation leads to higher fixation on an initial example. Participants who…

This https://arxiv.org/abs/2310.15223 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_…

Fast Forward Modelling of Galaxy Spatial and Statistical Distributions
A forward modelling approach provides simple, fast and realistic simulations of galaxy surveys, without a complex underlying model. For this purpose, galaxy clustering needs to be simulated accurately, both for the usage of clustering as its own probe and to control systematics. We present a forward model to simulate galaxy surveys, where we extend the Ultra-Fast Image Generator to include galaxy clustering. We use the distribution functions of the galaxy properties, derived from a forward mode…

DestripeCycleGAN: Stripe Simulation CycleGAN for Unsupervised Infrared Image Destriping
Shiqi Yang, Hanlin Qin, Shuai Yuan, Xiang Yan, Hossein Rahmani
https://arxiv.org/abs/2402.09101

DestripeCycleGAN: Stripe Simulation CycleGAN for Unsupervised Infrared Image Destriping
CycleGAN has been proven to be an advanced approach for unsupervised image restoration. This framework consists of two generators: a denoising one for inference and an auxiliary one for modeling noise to fulfill cycle-consistency constraints. However, when applied to the infrared destriping task, it becomes challenging for the vanilla auxiliary generator to consistently produce vertical noise under unsupervised constraints. This poses a threat to the effectiveness of the cycle-consistency loss,…

The Effects of Generative AI on Design Fixation and Divergent Thinking
Samangi Wadinambiarachchi, Ryan M. Kelly, Saumya Pareek, Qiushi Zhou, Eduardo Velloso
https://arxiv.org/abs/2403.11164

The Effects of Generative AI on Design Fixation and Divergent Thinking
Generative AI systems have been heralded as tools for augmenting human creativity and inspiring divergent thinking, though with little empirical evidence for these claims. This paper explores the effects of exposure to AI-generated images on measures of design fixation and divergent thinking in a visual ideation task. Through a between-participants experiment (N=60), we found that support from an AI image generator during ideation leads to higher fixation on an initial example. Participants who…

This https://arxiv.org/abs/2309.03244 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_ees…

EGIC: Enhanced Low-Bit-Rate Generative Image Compression Guided by Semantic Segmentation
We introduce EGIC, an enhanced generative image compression method that allows traversing the distortion-perception curve efficiently from a single model. EGIC is based on two novel building blocks: i) OASIS-C, a conditional pre-trained semantic segmentation-guided discriminator, which provides both spatially and semantically-aware gradient feedback to the generator, conditioned on the latent image distribution, and ii) Output Residual Prediction (ORP), a retrofit solution for multi-realism ima…