# Exciting Updates to Weights Image Training and Creation!

![Image](https://generated-images.weights.gg/cm2l81z2700pee08shphbam0q/image_1.webp)

We’re thrilled to introduce some amazing new features designed to make your creative process smoother and more powerful than ever.

**Free Image Training:** Free users can now train 1 image model per day—at no cost.

**Premium Image Training:** Unlock unlimited image training with our Premium plan, plus enjoy 1 extra fast training per day for quicker, high-quality results.

**Upgraded AI Images:** Our AI-generated images are now even higher in quality and faster to complete than ever before.

**Input Image Control:** Take your creativity to the next level by using an input image as a starting point. This gives you more control over your final creation and makes it easier to bring your vision to life.



New Image Training Options Now Available




# Free Model Training is Back on Weights!
![Image](https://generated-images.weights.gg/cm2l6dnw201vbyjjvhb2d2l51/image_1.webp)

Free users can now train up to **7 models** each week, completely free of charge.
It's a great opportunity to get hands-on without spending a dime.

For those with **Premium** subscriptions, there's even more to celebrate—unlimited model training with zero restrictions.

Whether you're exploring new techniques or fine-tuning your favorite models, Weights ensures everyone can get back to training.

Stay tuned for more updates and happy training!

Free Training is Back!


# Introducing Flux and LoRA's: A New Era of AI Image Generation 🎨🚀

We're thrilled to announce two game-changing updates to our AI image generation platform: **Flux**, the new industry-leading model for high-quality image generation, and **LoRA** (Low-Rank Adaptation) training capabilities. These innovations are set to revolutionize the way you create and customize AI-generated images.

## Flux: Elevating Image Quality for All Users 🌟

![Before and After Flux](https://assets.weights.gg/31375793-eba0-4b94-8712-fb2eb4a52c1c.png)

Flux is our latest breakthrough in AI image generation, offering unprecedented quality and capabilities:

- **Enhanced Detail**: Create intricate, lifelike scenes with stunning clarity.
- **Realistic Text Integration**: Seamlessly incorporate text into your images with natural-looking results.
- **Higher Resolution**: Enjoy sharper, more detailed outputs for all your creative projects.

The best part? Flux is **completely free** for all Weights users! To use Flux, simply select the "Quality" option when creating a new image.

## Introducing LoRA's: Personalized AI Image Models 🏞️

![LoRA Training UI](https://assets.weights.gg/d7b41d54-bf8c-4381-b97e-91e94456eb84.png)

LoRA (Low-Rank Adaptation) technology allows you to fine-tune AI models for specific styles, characters, or objects. This means you can create truly unique and personalized images that perfectly match your vision.

### What can you do with LoRA's?

1. **Train custom styles**: Create models that replicate your favorite artist's style or a specific aesthetic.
2. **Design unique characters**: Develop models that consistently generate your own characters or personas.
3. **Specialize in objects**: Fine-tune models to excel at generating specific items or environments.

### LoRA Availability

- **Premium Users**: Full access to LoRA training capabilities. Create and use your own custom models!
- **Free Users**: Discover and use LoRA's created by our community. Explore a world of unique styles and creations!

> 💡 **Tip**: Upgrade to Premium to unlock the full potential of LoRA training and take your AI art to the next level!

## Getting Started with Flux and LoRA's

**Using Flux**:
   - Start a new image generation project
   - Select the "Quality" option in the model settings
   - Enjoy your high-quality Flux-powered creations!

**Exploring LoRA's** (All Users):
   - Browse our LoRA library in the "Image Models" section
   - Apply LoRA's to your image generations for unique stylizations

**Creating LoRA's** (Premium Users):
   - Go to Train -> Image Model
   - Upload your dataset of images
   - That's it, we made it easy

## What's Next?

We're constantly working to improve and expand our AI image generation capabilities. Stay tuned for:

- More LoRA training options
- LoRA training for free users
- More advanced image creation tools

Thank you for being part of our creative community. We can't wait to see what you'll create with Flux and LoRA's!

Happy Creating! ✨🎨

---

*Have questions or feedback? Join the discussion in our [Discord](https://discord.gg/weights) or reach out to our support team at team@weights.gg.*

Flux & LoRA's come to Weights!


# Please check out the latest news in Voice Trainings as of 10/30/2024 [Voice Training](https://www.weights.gg/updates/cm2l7lyos0184osahosdhnhnk)

 Important Update: Changes to Voice Training Limits

Dear Weights Community,

We're implementing changes to our Voice Training limits, effective immediately. We recognize this decision may disappoint some users, but it comes after careful consideration and is driven by financial necessity.

## Our Journey
Since Weights began, we've aimed to make AI accessible, user-friendly, and free where possible. We've expanded from model hosting to offering the comprehensive AI creation suite available today. However, Voice Training has presented unique challenges.

## The Voice Training Dilemma
Voice Training is exceptionally resource-intensive, often occupying a high-end GPU for hours per job. This has led to unsustainable costs that our current Premium subscription model can no longer offset. Despite significant efforts to optimize performance and reduce expenses, we need to adjust our Voice Training offerings to ensure Weights' continued operation.

## New Voice Training Limits
* __Free Voice Training will no longer be available__
* Weights Premium subscribers will receive __15 voice trainings per month__

_Note: Existing Premium subscribers with higher limits will retain their current plans until the next billing period_

## Alternative Options
For those who prefer not to subscribe to Weights Premium, individual Voice Training Items can be purchased through the Weights Item Shop. This pay-per-use model offers flexibility at a lower price point.

## Moving Forward
While we regret the need for these changes, they are crucial for Weights' long-term sustainability and our ability to continue innovating in AI creation. We remain dedicated to our core mission of making AI technology accessible and user-friendly.

We appreciate your understanding and continued support. Thank you for being part of our community!

## Update - Rewarding Your Loyalty: Introducing Free Premium Training Items!

We value your dedication to Weights and have exciting news for our most consistent users. We want to introduce a way for free users to be able to train, while still managing our costs. Starting now, you can earn free Premium Training items by maintaining your daily streak:

* Every 5-day streak: Receive 1 free Premium Training item
* Continue your streak to earn more rewards
* Example: Come to Weights and make a creation for 5 days straight, and you'll receive a premium training item. If you come back for 5 more days after that (making your streak 10), you will receive another one, and so on.


Thanks,


__The Weights Team__

An Update on Voice Training



### Q: What is Weights?

A: Weights is a cloud-based platform that lets you create AI-generated **voice models**, **images**, and **videos**. You can train custom voice models, create high-quality images and videos, and even interact with AI characters that _remember your conversations_.

### Q: What’s new on Weights?

A: October 20, 2024 updates include:

- **Free Training:** Free users can now train up to 7 voice models weekly and 1 image model per day.
- **Premium Perks:** Premium users enjoy unlimited model training, faster image processing, high-quality video creation, and enhanced character chat features.
- **Improved Chat:** Characters now _remember details_, adapt their mood, and respond in their voice. Premium users can add custom Image LoRA models to characters for accurate visuals.

### Q: What benefits do Premium users get?

A: Premium users enjoy **unlimited training** for both voice and image models, **faster processing**, high-quality video creation, and dynamic AI chat interactions. Premium users can also organize creations in custom collections and share them with the community.

---

# 💡 **VOICE MODELS**

---

### Q: How do I train a voice model?

A: Upload a clean audio sample of the voice you want to replicate (between _5-20 minutes long_). Start training, and you’ll receive your voice model once processing is complete.

### Q: How many voice models can I train?

A: Free users can train up to 7 voice models weekly, while Premium users have _unlimited training_.

### Q: What’s the Advanced Cover Creator?

A: The **Advanced Cover Creator (Beta)** allows you to create multi-voice performances, including duets. With advanced controls, you can customize voice blending and timing to enhance your covers.

---

# 💡 **IMAGE & VIDEO MODELS**

---

### Q: How does image model training work?

A: To train an **image LoRA model**, users can upload reference images that help the model learn specific styles, objects, or facial characteristics of people. Free users can train up to 7 models per week, while Premium users have access to _unlimited model training_ with faster processing options.

### Q: Can I combine multiple LoRAs in image creation?

A: Yes! **Multi-LoRA support** allows you to combine up to three LoRAs in a single image, enabling _style mixes and unique character compositions_.

### Q: How do I create high-quality AI videos?

A: Free users can create 1 video per day, and Premium users can generate up to 10 videos from image models in high quality. This applies to both _videos and image generation_. To get started, visit the **Video Creation** tab under "My Creations."

---

# 💡 **CHAT FEATURES**

---

### Q: How does the chat feature work?

A: Users can chat with AI characters who _remember past interactions_, facts about you, and adjust their mood dynamically. Chat voices now respond with _emotion and personalized memory_, making conversations more engaging.

### Q: Can I personalize my chat character’s appearance?

A: Users can add **Image LoRA models** to chat characters, ensuring consistency and accuracy when they share images.

---

# 💡 **CONTENT MANAGEMENT**

---

### Q: How can I organize my creations on Weights?

A: You can now **Like** and **Save** content across voice models, images, and videos with one click. Organize your creations in _custom collections_ and share them with the community if you’d like.

### Q: Can I manage content in bulk?

A: Yes! Use **bulk editing** to select multiple items for deletion or organization. This feature works across **My Creations**, **Collections**, and **Likes**.

---

# 💡 **FREE VS. PREMIUM**

---

### Q: What are the main differences between Free and Premium plans?

A: Free users can train up to 7 voice models per week and 1 image model per day, while Premium users enjoy _unlimited training_, high-quality video creation, dynamic chat voices, and faster processing speeds.

### Q: How do I upgrade to Premium?

A: To upgrade, click your profile picture in the upper right corner of the site. In the dropdown menu, select **Settings**, and then go to **Premium Subscriptions** to upgrade.

### Q: Can I cancel my Premium subscription anytime?

A: Yes, you can cancel anytime. Just click your profile picture, go to **Settings**, and select **Premium Subscriptions** to manage or cancel your plan. _Access to Premium features_ will continue until the end of your billing period.

---

# 💡 **SAFETY & CONTENT**

---

### Q: How does Weights ensure content safety?

A: Weights enforces strict content guidelines and uses advanced automated filters to prevent inappropriate or harmful content from being created or shared. All publicly visible content must pass rigorous _content filters_ to ensure it meets community standards and keeps the platform safe for everyone.

### Q: Can I share my creations?

A: Yes! Creations can be _shared with the community_, and Weights also offers options for _private creations_ visible only to you.


Frequently Asked Questions (FAQs)

In our fast-paced digital world, having the right tools can make all the difference. Whether you're a content creator, musician, or tech enthusiast, upgrading to Premium can boost your productivity and creativity. Here's why:

## Train More Models

Premium users are entitled to train up to 15 models per month.

## Priority Processing

Skip the line with priority processing and get your results faster.

## More Jobs

Queue up to 25 jobs at once, compared to 10 for free users.

## Longer Audio & Bigger Uploads

Process audio up to 15 minutes and upload files up to 50 MB.

## Extended TTS Capabilities

Use up to 2500 characters for Text-to-Speech, versus 250 for free users.

## Larger Datasets

Train with up to 40 minutes of data, double the free plan's allowance.

## Better Quality

Enjoy higher quality outputs and longer AI-generated videos.

## Exclusive Features

Get early access to new features, improvements, and content types.

## Verified Badge

Show off your Premium status with a special badge.


*Make the switch today! The Free plan gives you a taste, but Premium unlocks everything you need for just $9.99 a month.*


![Image](https://assets.weights.gg/0a3e7978-e79e-434c-bb90-425afbd2bad5.jpg)


Premium Weights Account

Looking to make an AI cover? Weights makes it extremely easy and free. You can make unlimited AI covers with RVC models in as little as 3 clicks on Weights, from any device! Let's get started:

# Create a Weights Account
This should only take 2 seconds. Just click `Log in` if you don't already have an account, and choose your preferred account type.

![Log in](https://assets.weights.gg/69bcbae1-f232-4671-a01e-fe31beb35ecf.png)

# Find a Voice
The first step to creating a cover is to find a voice you want to use. You can do this by looking at the Featured Models, or by browsing all 100k+ models on Weights


![How to find all Models](https://assets.weights.gg/b756b7a0-536e-4d89-b698-34966a1cb6ab.png)

From there, you can browse, search, and filter to find the perfect voice. Clicking the `Play` button on any of the models will also play a sample of each one. 

Once you've found the perfect model, click the **use** button.

![How to use a model](https://assets.weights.gg/cdb3670d-1f5b-4ca1-b7a8-af2bb9f5a9ec.png)

# Creating your cover
Now that you've selected your Voice, you've basically done the hardest part! You will see a modal open, which will take you through the 3 steps to create your cover.


![First step of creating a cover](https://assets.weights.gg/49b87d61-67ef-4311-87d9-8d17dd6c371e.png)

On this first step, you'll see the Voice you've selected. 
> **Pro tip**: You can also search for models directly from here without having to browse. Just select the `Voice Cover` option from the **Quick Create** on the homepage.

From here, you can hit Next, or select another model if you'd like to create a duet (more than one model singing a song).

## Choosing your audio
Now it's time to chose the audio you want to use. You can either upload an audio file, search for a YouTube video, use Text-to-Speech, or record your own voice.

![Audio Selection Step](https://assets.weights.gg/b584090e-8f72-4017-99b6-ece0435f9ba6.png)

Once you've selected your chosen audio, it's time to hit Next.

## Options
The final, optional step is to select your options for the cover.

![Options Selection](https://assets.weights.gg/3cff526f-914a-40aa-aed9-471d7bacaed3.png)

* Pre-Stemmed: Use this option if the input track has no instrumentals. It's not required, but can make it come out cleaner
* Pitch: Use if the pitch of the input voice and chosen voice don't match. A good rule of thumb is to use +/-12 for converting between male and female voices. You can create as many covers as you want on Weights though, so don't worry about getting this right the first time!

Once you've selected your options, just hit **Create**, and you're done.

# Wrap-up
Congrats! You've made your first cover. To find the results, head to your `My Creations` page:

![How to find your Creations](https://assets.weights.gg/5e800272-195f-4d54-be88-acb24c190dac.png)

From there you'll see everything you've made on Weights!

Happy Creating!



How to Create an AI Cover

Want to make AI Covers locally, for free? Replay has you covered! Here's how to get started:

---

## Installing Replay

1. **Download Replay**: Visit the [Replay download page](https://tryreplay.io) and download the installer.
2. **Open the Installer**: Once the download is complete, open the installer and follow the on-screen instructions.

   > **On Windows**: If you see a security warning, click **More info** and then **Run anyway**.
   ![Windows Installer Warning](https://assets.weights.gg/2c448a77-0dfc-4545-8a2b-4e0c70582282.png)

---

## Using Replay

1. **Open Replay**: If Replay isn't already open, launch the application. You should see a screen similar to the one below.
   ![Replay Welcome Screen](https://assets.weights.gg/3eb547f2-b784-42b5-925c-b780626c1a08.png)

   > Click download and allow some time for all the required files to be downloaded. This may take a while depending on your internet speed.

2. **AI Server Download**: Once the "AI Server" is done downloading, Replay should fully launch, and you will see a screen similar to the one below.
   ![Replay Main Screen](https://assets.weights.gg/092291d0-f2c0-4e57-80e0-68ddef359f47.png)

   > You have now successfully installed Replay! You can now locally Train, Create AI Covers, and Create Music.

---

## Replay AI Cover Settings

- **Index Ratio**: `0.7`
- **Consonant Protection**: `0.5`
- **Volume Envelope**: `0.5`

  ---

- **De Echo & Reverb**: `You Choose`
  > Depends if the song you are processing has a lot of Reverb & Echo.

  ---

- **Relative Pitch**: `You Choose`
  > Depends on if the song you are processing is sung by a Female or a Male.
  - If processing a song with a Female voice model and sung by a Male, change the pitch to `+8` or `+12`.
  - If processing a song with a Male voice model and sung by a Female, change the pitch to `-8` or `-12`.

  ---

- **Instrumental Pitch**: `Leave at 0`
  > You may change this if necessary, but it usually results in deformed or out-of-tune sounding instrumentals.

  ---

- **Stem Method**: `UVR-MDX-NET Voc FT (pth)` or `UVR-MDX-NET-Inst_full_292`

  ---

- **F0 Method**: `RMVPE` or `Fcpe`
  > You can use Pm if you want fast processing.

  ---

- **Render Device**: `Cuda` or `MPS`
  > Never use CPU unless necessary, as it is the slowest render option.

---

## Replay Training Settings

### Nvidia GPUs Only

- **Auto Epochs**: `On`

  ---

- **Stem Method**: `UVR-MDX-NET Voc FT (pth)` or `UVR-MDX-NET-Inst_full_292`

  ---

- **Render Device**: `Cuda`
  > Never use CPU unless necessary, as it is the slowest training option.

---

## Replay Music Settings

### Nvidia GPUs Only

- **Render Device**: `Cuda`
  > Never use CPU unless necessary, as it is the slowest training option.

  ---

- **Song Duration**: `15s` to `30s` for optimal processing times
  > You can change this number to any duration you prefer, but longer times will result in extended creation times.

  ---

- **Model Name**: `facebook/musicgen-medium` or `facebook/musicgen-melody` (if using a Melody)

Replay Guide

# What is Weights?
Weights is a social platform for creating with all types of AI. You can create:
- **Voice Covers**
- **Duets**
- **Text-to-Speech**
- **Images**
- **Videos**
- **Music**
- **Chat with AI Characters**

all for free! And once you've created something, you can share your masterpiece with the community to inspire others.


# Your first Creation
Let's make your first AI Creation!

First, you'll need to create an account. If you're not signed up already, just click the **Log In** button at the top of the Home page. This only takes a few seconds.

Next, let's select an option from the _Quick Create_ section at the top of the Home Page. Let's click on *Image*.

![Quick Create on Mobile](https://assets.weights.gg/3842f834-fb5c-441b-aaa9-6e9782d83281.png)

This will open a modal, with your options for creating an image:
![Image Creation Modal](https://assets.weights.gg/c6129eee-2783-40af-a132-ddba7ca8eb95.png)

From here, you can either describe the image you want to make, or simply choose one of the Inspiration ideas. Once you've either described or selected your image prompt, hit **Generate**. After a second or two, you'll see your generated image!

![Example of a Generated Image](https://assets.weights.gg/284f242a-8fb7-404e-b377-467f4353dbfe.png)

Congrats! You've made your first AI Creation. See, wasn't that easy?

# Accessing Your Creations
Everything you make on Weights lives in **My Creations**. To get there, simply click on Create -> My Creations in the Navigation Bar on Desktop, or the _My Creations_ tab on Mobile. From here, you can view everything you've made on Weights.

# Sharing Your Creations
Make something you're really proud of? Share it with the community! To create a post, you'll click the **Create Post** button from _My Creations_.

![Create Post Button](https://assets.weights.gg/61fc907d-c317-401f-88d4-1c98e8c6ff52.png)

After this, just fill out a fun description for your post, and hit _Post_! Just like that, your Creation will be shared with the Community to enjoy! You'll automatically be notified when people leave you comments.

# Wrap-Up
That's about all you need to know to get started on Weights! There's a ton more to explore, so get out there are start creating!

If you need more help, check out our other Guides or [Join our Discord Community](https://discord.gg/weights).

Happy Creating!

Weights Updates

New Image Training Options Now Available

Free Training is Back!

Flux & LoRA's come to Weights!

An Update on Voice Training

Frequently Asked Questions (FAQs)

Premium Weights Account

How to Create an AI Cover

Replay Guide

Weights Starter Guide

links

legal

community

get the app