Promotional graphic for ComfyUI Depth Estimation update v1.3.5. Features large, glossy 3D text reading 'REAL 3D' in cyan blue, centered against a blurred purple and orange sunset cityscape background. Below the main title, white text reads 'COMFYUI v1.3.5' followed by the key features: 'Camera Intrinsics + Depth Anything V3'. Dec 08 2025 · 0 comments · 3D Modeling and Design, Advanced Digital Art, AI, AI-Generated Art, machine-learning, Tools ·

ComfyUI Depth Estimation Node v1.3.3 Released

I’m excited to announce the release of **ComfyUI Depth Estimation Node v1.3.3**! This update brings key improvements to 3D integration workflows and enhances stability for your daily generations.

🌟 What’s New?

1. Camera Intrinsics Export
The standout feature in this release is the new support for exporting **Camera Intrinsics**. This is a game-changer for anyone taking their depth maps into 3D software like Blender or Houdini. You can now get more accurate 3D reconstruction by matching the camera parameters directly from the estimation node.

2. Improved Stability
We’ve squashed several bugs that were causing intermittent issues during batch processing. The node is now more robust, ensuring your long render queues finish without hiccups.

3. Performance Tweaks
Minor optimizations have been made to the depth calculation process, shaving off valuable milliseconds from your generation time.

GIF

🔧 How to Update

The easiest way to get v1.3.3 is via the **ComfyUI Manager**:
1. Open ComfyUI.
2. Click on “Manager”.
3. Select “Fetch Updates”.
4. If the node is already installed, click “Update All” or find “ComfyUI Depth Estimation” in the list and update specifically.
5. Restart ComfyUI to apply the changes.

Alternatively, you can do a `git pull` in your `custom_nodes/ComfyUI-Depth-Estimation` directory.

🔗 Links
* GitHub Repository: Link to your GitHub Repo
* ComfyUI Manager: Link to Comfy Registry

Promotional graphic featuring 3D neon text 'Multi-Vision Toolkit v2' and 'Qwen3-VL & Flash Attention' against a dark background with a cybernetic eye illustration. Dec 02 2025 · 0 comments · AI, machine-learning, Tool, Tools, VLM ·

Vision-Toolkit Demo

php

Vision-Toolkit Demo

Advanced Computer Vision & Dataset Curation

🔍 Primary Functionality

Automated image captioning and dataset curation using state-of-the-art Vision Language Models (VLM). Streamline your LoRA training data preparation with local, privacy-focused AI.

⚡ Key Capabilities

  • Multi-Model Support: Seamless switching between Florence-2 (Speed) and Qwen3-VL (Accuracy).
  • Non-Blocking UI: Asynchronous processing ensures the interface remains responsive during heavy inference.
  • Batch Processing: Rapidly tag and caption thousands of images with custom prefixes and suffixes.

🚀 Main Benefits

  • Privacy First: All inference runs locally on your GPU—no data leaves your machine.
  • Efficiency: Reduce dataset preparation time by up to 80% compared to manual tagging.
  • Precision: High-fidelity captions specifically optimized for generative AI training.

Download Vision-Toolkit

Open Source • Local Execution • Python Based

 

 

Futuristic saurian oracle character with cybernetic enhancements and purple-blue lighting, generated using Saurian Oracle Qwen LoRA Oct 11 2025 · 0 comments · Advanced Digital Art, AI-Generated Art ·

SAURIAN ORACLE QWEN LORA

I’m excited to release the Saurian Oracle Qwen LoRA – a specialized AI model for generating futuristic saurian oracle characters with mystical and technological elements.

Saurian Oracle LoRA example 1
Futuristic saurian oracle with cybernetic enhancements

What is it?

This LoRA model enables you to generate unique reptilian sage characters combining ancient mysticism with futuristic technology. Use the trigger word saurian_oracle in your prompts.

Sample Gallery

Example 1
Example 2
Example 3
Example 4

Key Features

  • 282 MB model size – efficient and fast
  • Trained for 3,000 steps on 23 curated images
  • Optimized for 768×768 resolution
  • Apache 2.0 license – free for any use
  • Works with Qwen image generation

Quick Start

1. Download from Hugging Face
2. Load into your Qwen pipeline
3. Add saurian_oracle to your prompts

Recommended settings: Guidance 3.5-4.5, Steps 20-25

Download

Download on Hugging Face

Repository: https://huggingface.co/Limbicnation/saurian-oracle-qwen-lora

Released under Apache 2.0 License

Flux Aug 04 2024 · 0 comments · Short Films and Mini-Movies ·

New Update! TransformDepth V2

TransformDepth V2 Update: Now Featuring DepthAnything V2 Support!

We’re excited to announce that TransformDepth now includes support for DepthAnything V2, offering more accurate depth estimations and improved post-processing options for even better results. This update brings a significant upgrade, allowing for more precise and refined outputs.

Nov 23 2023 · 0 comments · Tutorials and How-To Guides, YouTube Content and Creators ·

Speed Revolution in AI: How Latent Consistency Models Turbocharge Image Generation

🌟 Unveiling the game-changer in image generation: Latent Consistency Models (LCM) 🖼️! Our new video simplifies how LCM slashes image generation steps, making Stable Diffusion lightning-fast ⚡! Say goodbye to the old 25-50 steps and hello to just 4-8. We’re talking about SDXL models zooming on a 4090 in about 0.75 second—yes, you heard that right, 0.75 second! 🚀 And for Mac users? A whopping 10x speed increase! Get the scoop on how this magic happens without the usual distillation hassle. Hit play for an epic tech revolution! 💻 #StableDiffusion #LCM #AI #TechTrends #SpeedyAI