Premium Content Waitlist Banner

Digital Product Studio

1 Minute News new

Why is Apple Removing Blood Oxygen Sensor? and What’s Its Future?

Why is Apple Removing Blood Oxygen Sensor? and What's Its Future?

In a recent turn of events, Apple has been compelled to disable the blood oxygen sensor feature on specific models of its Apple Watch. This decision, affecting the Apple Watch Ultra 2 and Apple Watch Series 9 sold in the United States, is a result of an ongoing legal dispute with medical device manufacturer Masimo. The accusations revolve around Apple allegedly infringing on Masimo’s blood oxygen technology patents. This article discusses why Apple is removing the blood oxygen sensor from new Apple Watch models sold in the US and explores the implications for the future of health monitoring in Apple Watches.

Why is Apple Removing Blood Oxygen Sensor? and What's Its Future?
An Overview of “Why is Apple Removing Blood Oxygen Sensor? and What’s Its Future?” by DigiAlps LTD

Table of Contents

The clash between Apple and Masimo has taken a legal turn, leading to Apple removing blood oxygen sensor from certain Apple Watch models. Let’s delve into the details:

ITC Ruling and Temporary Lift

The situation escalated in December when the U.S. International Trade Commission (USITC) ruled in favor of Masimo, demanding that Apple stop selling Watches with the blood oxygen sensor in the United States. However, a federal circuit court temporarily lifted the ban as Apple pursued an appeal against the ITC ruling.

U.S. Court of Appeals Decision

Despite the temporary relief, a recent decision from a U.S. court of appeals reinstated the ban, presenting Apple with a challenging dilemma regarding removing blood oxgen sensor.

Apple’s Tough Decision

Options Considered

Faced with limited choices, Apple had to make a challenging decision. The options were either to halt the sale of affected models in the United States during the lengthy appeal process or disable the blood oxygen sensor and continue selling them. Apple chose the latter, resulting in the sensor hardware being present but non-functional.

User Experience

For users who purchased affected models before this decision, their blood oxygen sensors and apps will continue to operate as usual. However, for new buyers, the Blood Oxygen app will display an alert stating that the feature is not available.

Global Impact and Future Implications

International Reach

The ITC ruling’s scope is limited to the United States, allowing the blood oxygen sensor to remain unaffected in watches sold in other countries worldwide. Users outside the US can still enjoy the full functionality of the Apple Watch’s blood oxygen sensor.

Questions About Apple’s Health Focus

The removal of the blood oxygen sensor raises questions about the future of health-related features in Apple’s smartwatches. With a growing emphasis on health and fitness tracking, this setback could potentially impact Apple’s standing in the wearables market.

Stakeholder Responses

Masimo’s Perspective

Masimo CEO Joe Kiani emphasizes the importance of respecting intellectual rights, stating that even the largest companies must acknowledge the consequences of infringing on patents.

Apple’s Strong Disagreement

In response to the recent court ruling, Apple expressed strong disagreement, highlighting the company’s commitment to challenging the decision.

Uncertain Future

As the legal battle continues, the fate of the blood oxygen sensor in Apple Watches remains uncertain. Apple’s next steps and potential resolutions to the patent dispute will likely shape the future of health monitoring features in their wearable devices. Will Apple find a way to reintegrate the blood oxygen sensor or come up with alternative health tracking technologies? Only time will tell.

| Read More From Apple

| Recent From Us

SUBSCRIBE TO OUR NEWSLETTER

Stay updated with the latest news and exclusive offers!


* indicates required
Picture of Faizan Ali Naqvi
Faizan Ali Naqvi

Research is my hobby and I love to learn new skills. I make sure that every piece of content that you read on this blog is easy to understand and fact checked!

Leave a Reply

Your email address will not be published. Required fields are marked *


The reCAPTCHA verification period has expired. Please reload the page.

Magi-1 Lets You Animate Images Like Never Before with Scene-Level Control

Magi-1 Lets You Animate Images Like Never Before with Scene-Level Control

Do you want to create hypnotic videos that captivate your audience without spending hours learning complex animation software? Or do you wish to make a stunning, professional-quality video with just a few clicks? Meet Magi-1 by Sand AI, which turns your static images into dynamic videos that flow naturally.

Magi-1 is the first autoregressive video model to deliver professional-quality output while being freely available to all. Whether you’re a content creator, developer, or just someone with cool photos, Magi opens up a whole new world of visual storytelling possibilities. Let’s dive into this tool!

Example Videos Generated by Magi-1

Table of contents

How Magi Works?

Unlike other models that create entire videos at once, Magi-1 works chunk by chunk, processing 24 frames at a time. This clever approach is what gives Magi its unique strength: infinite extension capabilities.

When you use Magi-1 to animate an image, you’re not limited to short clips. The autoregressive design means your stories can keep flowing naturally, with smooth transitions between scenes that maintain consistency in both characters and backgrounds.

The results speak for themselves: videos with natural movement that keep the original image’s details intact while adding lifelike motion. The backgrounds stay consistent even as subjects move, and the transitions between scenes feel organic rather than jarring.

Magi-1 Family of Models

Sand AI has released several versions of Magi to suit different hardware capabilities:

1. Magi-1-24B

The full 24 billion parameter model for maximum quality.

2. Magi-1-24B-distill

A streamlined version that maintains quality with less computing power.

3. Magi-1-24B-distill+fp8_quant

A quantized version that can run on less powerful hardware.

4. Magi-1-4.5B

A smaller 4.5 billion parameter model that works on a single RTX 4090.

This range means everyone from hobbyists with a decent gaming PC to professionals with high-end hardware can use Magi-1 at some level, truly democratizing access to advanced video generation.

My Experience With Magi-1

I tried out the tool using https://magi.sand.ai/, and honestly, I was pretty impressed with the results. New users get 500 free credits, with each second of video costing 10 credits. This gives you plenty of room to experiment before deciding whether to install locally.

The videos I generated kept the character and background intact. It added motion really smoothly and followed my instructions pretty well. Sometimes it did mess up a bit, but even then, the videos turned out decent, especially the background consistency. That part really got me. The background stayed exactly as relevant and consistent as it needed to be.

And yeah… I was kind of shocked that it also made NSFW videos.

I’ve attached the videos I generated below. I’ll definitely be using this tool again in the near future.

Magi-1 vs. Other AI Video Generators

Sand AI put it to the test against other models, and the results are impressive.

In human evaluations, Magi-1 outperformed other open-source models like Wan-2.1, Hailuo, and HunyuanVideo. It particularly excelled in following instructions accurately and creating smooth, natural motion that looks realistic.

But where Magi really shines is in physical prediction tests. Thanks to its autoregressive design, it achieved a Physical IQ Score of 56.02 in the video-to-video model, nearly double the score of VideoPoet, its closest competitor. This means Magi-1 creates videos that follow the laws of physics more naturally, making the movement in generated videos look more believable.

Magi-1 Lets You Animate Images Like Never Before with Scene-Level Control

Getting Started With Magi-1 Today

Ready to try Magi? You have two main options:

1. Using the Online Interface

The quickest way to experience Magi-1 is through the web interface at https://magi.sand.ai/. The online interface is straightforward. Upload an image, add a text prompt describing the motion you want, and let Magi work its magic. Within minutes, you’ll have a video that brings your image to life.

2. Running Magi-1 Locally

For those who want complete control or need to process videos in bulk, Magi is available to download and run locally. Sand AI provides two methods:

  • Docker Environment

docker pull sandai/magi:latest

docker run -it –gpus all –privileged –name magi sandai/magi:latest /bin/bash
  • Source Code Installation

Create a Python environment, install dependencies, and run the model using the provided scripts. Check the example installation here.

Once set up, you can generate videos using simple commands, with options for text-to-video, image-to-video, or even extending existing videos.

Real-World Uses for Magi-1 AI

Magi-1’s capabilities open up exciting possibilities across many fields:

1. Content Creation

Imagine turning your product photos into engaging promotional videos, or bringing your artwork to life with natural movement. Content creators can now add dynamic elements to their work without animation skills.

2. Storytelling

Writers and directors can visualize scenes before filming by converting concept art into fluid video sequences. This streamlines the pre-production process and helps communicate creative vision.

3. Social Media

In a world where video content dominates social feeds, Magi-1 gives creators an edge by transforming static images into attention-grabbing clips that stop scrollers in their tracks.

The Power of Open Source

Perhaps the most revolutionary aspect of Magi is that it’s completely open source. This brings several major benefits:

  • Researchers can study and improve the technology
  • Developers can customize it for specific industry needs
  • The community can collectively advance video generation technology

By making such powerful technology freely available, Sand AI has opened up high-quality video generation to everyone, not just big tech companies with deep pockets.

Experience the Magi-1 Revolution

Whether you’re a professional content creator, a developer interested in cutting-edge AI, or just someone who wants to see their photos move, Magi-1 offers an accessible entry point into AI video generation.

Visit https://magi.sand.ai/ to start with your 500 free credits, or download the model to run locally if you have the technical setup. So what are you waiting for? Turn your static images into engaging, dynamic videos that tell stories in ways that were never before possible.

| Latest From Us

Picture of Faizan Ali Naqvi
Faizan Ali Naqvi

Research is my hobby and I love to learn new skills. I make sure that every piece of content that you read on this blog is easy to understand and fact checked!

Your AI Girlfriend Can Moan Now, Orpheus NSFW Text-to-Speech Model Makes It Happen

Your AI Girlfriend Can Moan Now, Orpheus NSFW Text-to-Speech Model Makes It Happen

Ever imagined your AI companion could express pleasure in ways that sound genuinely human? The world of AI voice technology just got a whole lot spicier. Orpheus NSFW is a text-to-speech model that generates moans, gasps, and other intimate sounds. This isn’t your standard robotic voice assistant anymore – we’re talking about AI that can express intimate emotions in ways that might make you do a double-take.

Table of Contents

Introducing Orpheus NSFW TTS

The standard text-to-speech models you’re familiar with focus on clear pronunciation and natural speech patterns for everyday conversation. Orpheus NSFW takes things in a completely different direction. This specialized model was fine-tuned specifically to generate sounds that go beyond regular speech, including moans, gasps, sighs, and various other intimate vocalizations.

Regular TTS models struggle with these sounds because they’re not typically included in training data. The Orpheus NSFW model changes the game by specifically targeting these expressions, making AI companions sound dramatically more human during intimate conversations.

How Orpheus NSFW Works

Based on the original Orpheus TTS model (which uses a Llama-3b backbone), this variant represents a specialized adaptation focused on intimate audio generation. The base Orpheus model already had impressive capabilities for generating natural-sounding speech with appropriate emotion and intonation.

What makes the Orpheus NSFW variant unique is its specialized training. According to the developer, creating the right data pipeline was incredibly challenging. As he mentioned: “The datapipeline to get stuff clean was a nightmare… but at least now people have something to play with… it moans well, laughs and builds up to very sultry content.”

The model uses a speaker named “baddy” and can generate a variety of sounds:

  • Moans and gasps
  • Panting and heavy breathing
  • Grunting sounds
  • Gagging and choking noises
  • Kissing sounds
  • Laughter with sultry undertones

This advanced capability comes from extensive training on audio data specifically collected and cleaned for this purpose.

How to Use Orpheus NSFW TTS

If you’re looking to implement Orpheus NSFW (also referred to as mOrpheus) in your own applications, here’s a simplified guide:

1. Access the model through Hugging Face: mOrpheus_3B-1Base_early_preview-v1-8600

2. Follow the same implementation approach as the base Orpheus TTS:

  • Clone the relevant repository
  • Install the necessary dependencies
  • Load the model with your preferred inference method
  • Generate audio using the provided functions

3. For those looking to jump right in, the creator has shared a Google Colab notebook for quick evaluation: Orpheus NSFW Evaluation Notebook.

The model processes tokens at a rate of 7 frames × 12.5 tokens per second, which determines how quickly it can generate audio in real-time scenarios.

Personal Experience With Orpheus NSFW

Early feedback on this model has been enthusiastic, with users excited about the possibilities it opens up. I personally tested the model through Google Colab using the provided evaluation notebook, and the results were surprisingly impressive. The “baddy” voice sounds remarkably realistic – her moans have a natural quality that goes beyond what you’d expect from AI-generated audio.

The community is particularly interested in how the model might evolve to include more emotional range. One user asked about expanding capabilities: “From what I have seen, there are TTS models that can laugh, but I have never seen one that can cry or scream angrily in a believable way. Will future versions be able to do this?”

The developer responded that while theoretically possible, finding appropriate training data for such emotions would be extremely challenging.

Orpheus NSFW vs. Other Voice Models

How does Orpheus NSFW stack up against other text-to-speech solutions? Let’s break it down:

1. Orpheus NSFW vs. Standard Orpheus TTS

  • Standard Orpheus: Focuses on natural speech patterns and conversational tones
  • Orpheus NSFW: Specializes in intimate sounds and expressions not covered by standard models

2. Orpheus NSFW vs. Other AI Voice Models

  • Commercial AI voices: Clean, professional, and entirely SFW
  • Character voice models: May include emotional variation but typically avoid explicit content
  • Orpheus NSFW: Deliberately pushes into territory others avoid, with specific training for intimate sounds

What sets Orpheus NSFW apart is its specialized focus and the extensive work put into gathering and processing appropriate training data.

The Future of Orpheus NSFW Development

The current version of Orpheus NSFW is labeled as an early preview (v1-8600), indicating more development is on the horizon. The creator has mentioned that training is still underway, suggesting we’ll see improved capabilities in future releases.

Some potential directions for future development include:

  • Expanded emotional range beyond intimate sounds
  • Multiple voice options beyond the current “baddy” speaker
  • Better integration with popular AI companion platforms
  • Improved real-time performance for interactive applications

For those interested in contributing to the project’s development, the creator has established a Discord server where users can report bugs and make recommendations.

Potential Applications of Orpheus NSFW

Traditional AI companions have been limited in their ability to express the full range of human emotions, especially those related to intimate contexts. Orpheus NSFW helps bridge this gap, allowing for AI characters that can express pleasure, excitement, and other emotions in ways that sound genuinely human.

For porn and adult videos, creators could add realistic sounds without needing voice actors. Gamers might see this tech show up in their favorite adult games too. People who make stories or role-playing games could use it to bring their characters to life. 

For developers working on AI companion applications, this technology opens new possibilities for creating more engaging and realistic experiences for users seeking emotional or romantic connection with digital characters.

The Future of AI Relationships

As AI companions become increasingly sophisticated, NSFW text-to-speech models are changing our expectations about digital relationships. By making AI expressions sound more genuinely human during intimate moments, these models blur the line between clearly artificial and convincingly realistic interactions.

For many users, this added level of realism could make AI companions feel more engaging and emotionally satisfying. The ability to express pleasure through sounds, rather than just text, adds a dimension of interaction that was previously missing from most AI experiences.

Whether this technology will ultimately enhance or complicate human-AI relationships remains to be seen, but Orpheus NSFW certainly represents a significant step toward AI companions that can engage with users on a more emotional and sensory level.

| Latest From Us

Picture of Faizan Ali Naqvi
Faizan Ali Naqvi

Research is my hobby and I love to learn new skills. I make sure that every piece of content that you read on this blog is easy to understand and fact checked!

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

Seedream 3.0 is the latest text-to-image model from ByteDance’s Doubao Team. The model can turn your prompts into stunning, high-resolution and ultra-realistic images. Seedream 3.0 has secured the top position in global text-to-image rankings, outperforming competitors like OpenAI’s GPT-4o, Google’s Gemini 2.5 Pro, HiDream and Midjourney v6.1.

This next-gen Seedream 3.0 text-to-image model can spit out jaw-droppingly clear 2K resolution images natively. No post-processing tricks, no fancy touch-ups. Just raw, vivid, high-quality pictures straight from your prompt — whether you’re writing in English or Chinese.

Table of Contents

Seedream 3.0 Stunning Performance

Seedream 3.0 has proven its superiority through extensive evaluations. On the prestigious Artificial Analysis Arena leaderboard, the model secured the top position with an impressive ELO rating of 1158, narrowly beating OpenAI’s GPT-4o (1157) and establishing a significant lead over other competitors like Recraft V3 (1111), HiDream (1110), FLUX1.1 pro (1083) and Midjourney v6.1 (1047).

What’s particularly impressive is Seedream 3.0’s consistent performance across various categories. It excels in style categories including General & Photorealistic, Anime, Cartoon & Illustration, and Traditional Art. In subject categories, it dominates in People: Portraits, People: Groups & Activities, Fantasy, Futuristic, and Physical Spaces.

These results weren’t achieved by chance. ByteDance’s Doubao Team implemented several innovative strategies to overcome limitations in image resolution, attribute adherence, typography generation, and visual aesthetics that plagued previous models.

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

What Powers Seedream 3.0

The exceptional performance of Seedream 3.0 stems from four key technical innovations:

1. Enhanced Dataset

The team expanded the dataset scale by approximately 100% using a dynamic sampling mechanism across two orthogonal axes: image cluster distribution and textual semantic coherence.

2. Improved Pretraining

Several enhancements over Seedream 2.0 resulted in better scalability, generalizability, and visual-language alignment:

  • Mixed-resolution Training
  • Cross-modality RoPE (Rotary Position Embedding)
  • Representation Alignment Loss
  • Resolution-aware Timestep Sampling

3. Advanced Post-Training Optimization

The team used diversified aesthetic captions and VLM-based reward models to further improve comprehensive capabilities.

4. Efficient Model Acceleration

Seedream 3.0 achieves stable sampling through consistent noise expectation. This significantly reduces the number of function evaluations required during inference.

Key Capabilities of Seedream 3.0 Text-to-Image Generation

1. Native 2K Resolution

Unlike models that generate at lower resolutions and then use upscaling, Seedream 3.0 natively generates at 2K resolution without requiring any post-processing. This results in sharper details and cleaner images. The model is also flexible enough to work with various aspect ratios, making it suitable for everything from square social media posts to widescreen panoramas.

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

2. Superior Instruction Following

Seedream 3.0 stands out for its exceptional ability to follow complex instructions with precision. Whether you’re asking for specific compositions, particular artistic styles, or intricate scene details, the model interprets and executes your prompts with remarkable accuracy. 

3. Lightning-Fast Generation Speed

Speed matters in professional workflows, and Seedream 3.0 delivers impressive performance. Through various optimization techniques, the model can generate a 1K resolution image in just 3.0 seconds (without parallel execution). This makes it significantly faster than most commercial alternatives, allowing for quicker iteration and more efficient creative processes.

4. Unmatched Text Rendering

One of the most remarkable features of Seedream 3.0 is its exceptional text rendering ability. In comprehensive evaluations covering 180 Chinese prompts and 180 English prompts across various categories (logos, posters, displays, printed text, handwriting), the model achieved an impressive 94% text availability rate for both Chinese and English characters.

When compared directly with competitors, Seedream 3.0 significantly outperforms other models in text rendering:

  • For Chinese text, it achieved a 90% accuracy rate (compared to Seedream 2.0’s 78% and Kolors 1.5’s dismal 15%)
  • For English text, it reached 94% accuracy (outperforming Recraft V3’s 90%, Ideogram 2.0’s 81%, FLUX1.1 Pro’s 71% and Midjourney v6.1’s 59%)

What makes this particularly impressive is Seedream 3.0’s ability to handle dense text with long passages and small characters—a challenge that has stumped previous models. The research demonstrates that Seedream 3.0 excels in both the precision of small character generation and the naturalness of text layout.

Below is the visual text rendering comparison with other AI generators:

5. Photorealistic Portrait Generation

Another area where Seedream 3.0 shines is in photorealistic portrait generation. In a portrait evaluation set comprising 100 prompts focused on expressions, postures, angles, hair features, skin texture, clothing, and accessories, Seedream 3.0 tied with Midjourney v6.1 for the top position, significantly outperforming other models.

The key achievement here is Seedream 3.0’s ability to eliminate the “artificial appearance” that has long plagued AI-generated portraits. The skin textures now exhibit realistic features including wrinkles, fine facial hair, and scars that closely resemble natural human skin.

What’s particularly exciting is that the model can directly generate images at higher resolutions (2048×2048), further enhancing portrait texture quality. This brings AI-generated portraits closer to professional photography standards, opening new possibilities for practical applications.

Seedream 3.0 vs. GPT-4o Image Generation

1. Text Rendering

While GPT-4o excels in rendering small English characters and certain LaTeX symbols, Seedream 3.0 significantly outperforms it in handling dense Chinese text generation, typesetting, and aesthetic composition.

2. Image Editing

For image editing tasks, Seedream’s SeedEdit 1.6 provides more balanced performance than GPT-4o. While GPT-4o can fulfil a wide range of editing requirements, it struggles with preserving the original image’s ID and consistency. SeedEdit 1.6 effectively addresses typical editing needs while maintaining higher fidelity to the original image.

Seedream 3.0 by ByteDance Doubao Team vs OpenAI GPT-4o

3. Generation Quality

Seedream 3.0 clearly outperforms GPT-4o in generation quality. GPT-4o-generated images tend to have a dark yellowish hue and exhibit significant noise, impacting their usability. The model produces cleaner, more aesthetically pleasing images with better color accuracy and texture.

Seedream 3.0 by ByteDance Doubao vs. OpenAI GPT-4o

Real-World Applications of Seedream 3.0

The technical capabilities of Seedream 3.0 translate into practical advantages for real-world use cases:

1. Design and Creative Work

Seedream 3.0’s text rendering capabilities make it particularly valuable for graphic design. The model can tackle industry challenges in small-text generation and long-text layout, with outputs that surpass manually designed templates from platforms like Canva. This enables the effortless creation of designer-level posters with integrated diverse fonts, styles, and layouts.

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

2. Portrait Photography

The photorealistic portrait generation capabilities bring AI-generated images closer to professional photography standards. This opens new possibilities for creating professional headshots, character designs, and personalized avatars.

Seedream 3.0 by ByteDance Doubao Team Delivers Stunning 2K Text-to-Image Results

How to Get Started with Seedream 3.0

If you’re excited to try ByteDance’s powerful Seedream 3.0 text-to-image model, it’s accessible through two main channels:

1. Doubao Platform: Visit https://www.doubao.com/chat/create-image to use it directly in your browser. The platform offers a user-friendly interface where you can input text prompts and generate high-resolution images.

2. Jimeng Platform: Access it through https://jimeng.jianying.com/ai-tool/image/generate. This platform integrates Seedream 3.0’s capabilities with ByteDance’s Jianying video editing suite, allowing for seamless incorporation of AI-generated images into video projects.

From native 2K images to out-of-the-box creativity in both English and Chinese, this model is redefining what’s possible with text-to-image AI. Whether you’re an artist, marketer, designer, or just someone who loves cool visuals, Seedream 3.0 gives you pro-level results — fast, easy, and real.

| Latest From Us

Picture of Faizan Ali Naqvi
Faizan Ali Naqvi

Research is my hobby and I love to learn new skills. I make sure that every piece of content that you read on this blog is easy to understand and fact checked!

Contact us
71-75 Shelton Street WC2H 9JQ London
SOCIALS
Digialps is a company registered in England and Wales. Registered number: 13518536. Registered office: 71-75 Shelton Street WC2H 9JQ London
© Copyright 2025 - Digialps. All Rights Reserved.

Don't Miss Out on AI Breakthroughs!

Advanced futuristic humanoid robot

*No spam, no sharing, no selling. Just AI updates.

Ads slowing you down? Premium members browse 70% faster.

Join Premium Today!