
Metavershche and AIshche — page 11
Это не новости, это персональный экспертный взгляд на то, как развивается индустрия ИИ, графики, метаверса, крипты, нейротехнологий и в каком направлении катится все это безобразие. Для связи: @SergTsyp
Open in TelegramChannel posts

VideoViggle JST-2: New Model for 3D Generation
Viggle released the JST-2 engine for 3D scene generation. The technology uses Gaussian splats and costs $0.06 per scene, but interaction physics remain limited.
VideoSweeney in generations and new content
Post about Sweeney's stability in generations and funny content mentioning Tesla.
VideoFLUX 3 Max: New Accelerated Model for Video Generation
fal and Black Forest Labs have released the accelerated FLUX 3 Max model. Video generation is now 11.5 times faster with high quality and an Elo rating of 1139.

GPT-6.5 Release Details Uncovered via New Domain
The opensi.lol domain with a timer until November 1 and coordinates for San Francisco has been opened. This indicates an imminent GPT-6.5 release and a presentation at the Fort Mason Center.
VideoSanya and Danya Were Here
Sanya and Danya Were Here It's so disgusting that it's almost good. And all because you don't pray in the morning. The news is contextually from Sanya, or rather from Tini below
VideoCreating a Music Video with Seedance 2.5 and Suno
Breakdown of creating a music video for the song "Forest" using Seedance 2.5 and Suno. Generation settings, lip-sync setup, and bypassing filters.







GPT Image 2.5 Outperforms Open Models in Image Generation
An overview of GPT Image 2.5 capabilities compared to open-source models. Tests of generation quality and a method for automatic image labeling via Python and FFmpeg.


DeepSeek Released New Model and Search API for AI Agents
DeepSeek released the V4.1-Flash-1017 model with a 1 million token context and launched a Search API for AI agents, competing with major tech companies.

OpenAI to Return Sora via New Startup of Hollywood Producers
OpenAI may bring back the Sora model through a new startup led by Jeffrey Katzenberg and Bill Peebles. The project aims to develop technologies for the film industry.
VideoAnthropic to Allocate $48 Billion to Help AI Victims
Anthropic will create a $48 billion fund to help AI victims. An expert highlighted the contrast between the cost of progress and support measures.

Image Generation in Cartoon Style
How a neural network trained on cartoons copes with image generation. An analysis of face and texture quality when working with prompts.

MOSS AI Generates Classical Art Without Censorship
Testing the MOSS AI model reveals its uncensored approach to generating classical art interpretations, including a detailed marble sculpture of Jessica Rabbit.
VideoInstead of competing in penises, let's compete in penises
Instead of competing in penises, let's compete in penises Saturday, exhale, beavers.
VideoRisk Metric for AI Regulation in the US
Danya proposed a bill for the US: blocking powerful AI models is possible only if there is a threat of damage exceeding $1 billion or the death of more than 50 people.
VideoNeuro-roasting: Creating Promo Videos for Albums
Analysis of the process of creating a promo video for Endshipil and TumaniYo's album using Midjourney and Seedance neural networks.








New Benchmarks for Image Generators: Flux Klein Overview
Overview of Flux Klein capabilities: analysis of advantages and limitations of the image generator, including prompt handling, face quality, and parameter influence.

Jeva has gained vision
Jeva has gained vision. Now he can make quick decisions not only on text but also on images. It seems that Cloudflare with its CLEF already has this.
Video
Video
Flux 3. Klein, heavily censored
Flux 3. Klein, heavily censored. Heavily censored. I mean, he thinks naked people look like this. Krea too, by the way.
VideoFLUX 3 Image Klein
Black Forest Labs released FLUX 3 Image Klein with 4K support, bounding boxes, and local deployment from 8 GB VRAM. Review of specifications and licensing.
