BREAKING NEWS
LATEST POSTS
-
Transformer Explainer -Interactive Learning of Text-Generative Models
https://github.com/poloclub/transformer-explainer
Transformer Explainer is an interactive visualization tool designed to help anyone learn how Transformer-based models like GPT work. It runs a live GPT-2 model right in your browser, allowing you to experiment with your own text and observe in real time how internal components and operations of the Transformer work together to predict the next tokens. Try Transformer Explainer at http://poloclub.github.io/transformer-explainer
-
Henry Daubrez – How to generate VR/ 360 videos directly with Google VEO
https://www.linkedin.com/posts/upskydown_vr-googleveo-veo3-activity-7334269406396461059-d8Da
If you prompt for a 360° video in VEO (like literally write “360°” ) it can generate a Monoscopic 360 video, then the next step is to inject the right metadata in your file so you can play it as an actual 360 video.
Once it’s saved with the right Metadata, it will be recognized as an actual 360/VR video, meaning you can just play it in VLC and drag your mouse to look around. -
Black Forest Labs released FLUX.1 Kontext
https://replicate.com/blog/flux-kontext
https://replicate.com/black-forest-labs/flux-kontext-pro
There are three models, two are available now, and a third open-weight version is coming soon:
- FLUX.1 Kontext [pro]: State-of-the-art performance for image editing. High-quality outputs, great prompt following, and consistent results.
- FLUX.1 Kontext [max]: A premium model that brings maximum performance, improved prompt adherence, and high-quality typography generation without compromise on speed.
- Coming soon: FLUX.1 Kontext [dev]: An open-weight, guidance-distilled version of Kontext.
We’re so excited with what Kontext can do, we’ve created a collection of models on Replicate to give you ideas:
- Multi-image kontext: Combine two images into one.
- Portrait series: Generate a series of portraits from a single image
- Change haircut: Change a person’s hair style and color
- Iconic locations: Put yourself in front of famous landmarks
- Professional headshot: Generate a professional headshot from any image
-
AI Models – A walkthrough by Andreas Horn
the 8 most important model types and what they’re actually built to do: ⬇️
1. 𝗟𝗟𝗠 – 𝗟𝗮𝗿𝗴𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹
→ Your ChatGPT-style model.
Handles text, predicts the next token, and powers 90% of GenAI hype.
🛠 Use case: content, code, convos.
2. 𝗟𝗖𝗠 – 𝗟𝗮𝘁𝗲𝗻𝘁 𝗖𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝗰𝘆 𝗠𝗼𝗱𝗲𝗹
→ Lightweight, diffusion-style models.
Fast, quantized, and efficient — perfect for real-time or edge deployment.
🛠 Use case: image generation, optimized inference.
3. 𝗟𝗔𝗠 – 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗔𝗰𝘁𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹
→ Where LLM meets planning.
Adds memory, task breakdown, and intent recognition.
🛠 Use case: AI agents, tool use, step-by-step execution.
4. 𝗠𝗼𝗘 – 𝗠𝗶𝘅𝘁𝘂𝗿𝗲 𝗼𝗳 𝗘𝘅𝗽𝗲𝗿𝘁𝘀
→ One model, many minds.
Routes input to the right “expert” model slice — dynamic, scalable, efficient.
🛠 Use case: high-performance model serving at low compute cost.
5. 𝗩𝗟𝗠 – 𝗩𝗶𝘀𝗶𝗼𝗻 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹
→ Multimodal beast.
Combines image + text understanding via shared embeddings.
🛠 Use case: Gemini, GPT-4o, search, robotics, assistive tech.
6. 𝗦𝗟𝗠 – 𝗦𝗺𝗮𝗹𝗹 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹
→ Tiny but mighty.
Designed for edge use, fast inference, low latency, efficient memory.
🛠 Use case: on-device AI, chatbots, privacy-first GenAI.
7. 𝗠𝗟𝗠 – 𝗠𝗮𝘀𝗸𝗲𝗱 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹
→ The OG foundation model.
Predicts masked tokens using bidirectional context.
🛠 Use case: search, classification, embeddings, pretraining.
8. 𝗦𝗔𝗠 – 𝗦𝗲𝗴𝗺𝗲𝗻𝘁 𝗔𝗻𝘆𝘁𝗵𝗶𝗻𝗴 𝗠𝗼𝗱𝗲𝗹
→ Vision model for pixel-level understanding.
Highlights, segments, and understands *everything* in an image.
🛠 Use case: medical imaging, AR, robotics, visual agents. -
Introducting ComfyUI Native API Nodes
https://blog.comfy.org/p/comfyui-native-api-nodes
Models Supported
- Black Forest Labs Flux 1.1[pro] Ultra, Flux .1[pro]
- Kling 2.0, 1.6, 1.5 & Various Effects
- Luma Photon, Ray2, Ray1.6
- MiniMax Text-to-Video, Image-to-Video
- PixVerse V4 & Effects
- Recraft V3, V2 & Various Tools
- Stability AI Stable Image Ultra, Stable Diffusion 3.5 Large
- Google Veo2
- Ideogram V3, V2, V1
- OpenAI GPT4o image
- Pika 2.2
-
ComfyUI-CoCoTools_IO – A set of nodes focused on advanced image I/O operations, particularly for EXR file handling
https://github.com/Conor-Collins/ComfyUI-CoCoTools_IO
Features
- Advanced EXR image input with multilayer support
- EXR layer extraction and manipulation
- High-quality image saving with format-specific options
- Standard image format loading with bit depth awareness
Current Nodes
Image I/O
- Image Loader: Load standard image formats (PNG, JPG, WebP, etc.) with proper bit depth handling
- Load EXR: Comprehensive EXR file loading with support for multiple layers, channels, and cryptomatte data
- Load EXR Layer by Name: Extract specific layers from EXR files (similar to Nuke’s Shuffle node)
- Cryptomatte Layer: Specialized handling for cryptomatte layers in EXR files
- Image Saver: Save images in various formats with format-specific options (bit depth, compression, etc.)
Image Processing
- Colorspace: Convert between sRGB and Linear colorspaces
- Z Normalize: Normalize depth maps and other single-channel data
FEATURED POSTS
-
Infinigen – a free procedural generator of 3D scenes
https://github.com/princeton-vl/infinigen
Infinigen is based on Blender and is free and open-source (BSD 3-Clause License). Infinigen is being actively developed to expand its capabilities and coverage.
-
Photography basics: Exposure Value vs Photographic Exposure vs Il/Luminance vs Pixel luminance measurements
Also see: https://www.pixelsham.com/2015/05/16/how-aperture-shutter-speed-and-iso-affect-your-photos/
In photography, exposure value (EV) is a number that represents a combination of a camera’s shutter speed and f-number, such that all combinations that yield the same exposure have the same EV (for any fixed scene luminance).The EV concept was developed in an attempt to simplify choosing among combinations of equivalent camera settings. Although all camera settings with the same EV nominally give the same exposure, they do not necessarily give the same picture. EV is also used to indicate an interval on the photographic exposure scale. 1 EV corresponding to a standard power-of-2 exposure step, commonly referred to as a stop
EV 0 corresponds to an exposure time of 1 sec and a relative aperture of f/1.0. If the EV is known, it can be used to select combinations of exposure time and f-number.Note EV does not equal to photographic exposure. Photographic Exposure is defined as how much light hits the camera’s sensor. It depends on the camera settings mainly aperture and shutter speed. Exposure value (known as EV) is a number that represents the exposure setting of the camera.
Thus, strictly, EV is not a measure of luminance (indirect or reflected exposure) or illuminance (incidentl exposure); rather, an EV corresponds to a luminance (or illuminance) for which a camera with a given ISO speed would use the indicated EV to obtain the nominally correct exposure. Nonetheless, it is common practice among photographic equipment manufacturers to express luminance in EV for ISO 100 speed, as when specifying metering range or autofocus sensitivity.
The exposure depends on two things: how much light gets through the lenses to the camera’s sensor and for how long the sensor is exposed. The former is a function of the aperture value while the latter is a function of the shutter speed. Exposure value is a number that represents this potential amount of light that could hit the sensor. It is important to understand that exposure value is a measure of how exposed the sensor is to light and not a measure of how much light actually hits the sensor. The exposure value is independent of how lit the scene is. For example a pair of aperture value and shutter speed represents the same exposure value both if the camera is used during a very bright day or during a dark night.
Each exposure value number represents all the possible shutter and aperture settings that result in the same exposure. Although the exposure value is the same for different combinations of aperture values and shutter speeds the resulting photo can be very different (the aperture controls the depth of field while shutter speed controls how much motion is captured).
EV 0.0 is defined as the exposure when setting the aperture to f-number 1.0 and the shutter speed to 1 second. All other exposure values are relative to that number. Exposure values are on a base two logarithmic scale. This means that every single step of EV – plus or minus 1 – represents the exposure (actual light that hits the sensor) being halved or doubled.Formulas
(more…)
-
Zibra.AI – Real-Time Volumetric Effects in Virtual Production. Now free for Indies!
A New Era for Volumetrics
For a long time, volumetric visual effects were viable only in high-end offline VFX workflows. Large data footprints and poor real-time rendering performance limited their use: most teams simply avoided volumetrics altogether. It’s similar to the early days of online video: limited computational power and low network bandwidth made video content hard to share or stream. Today, of course, we can’t imagine the internet without it, and we believe volumetrics are on a similar path.
With advanced data compression and real-time, GPU-driven decompression, anyone can now bring CGI-class visual effects into Unreal Engine.
From now on, it’s completely free for individual creators!
What it means for you?
(more…)