This name is actually a combination of two parts: Runway ML is a neural network, and Gen-3 Alpha is its latest model, which became publicly available in August. However, there are some limitations, but more on that later.
The neural network itself was known long before the announcement of Sora AI as one of the most advanced video generators. The third generation of the model has become the most technologically advanced. It is important to note that there are two versions of it:
- Gen-3 Alpha is a full-fledged model that allows you to generate videos from images and text descriptions. It has been significantly improved compared to the previous version, Gen-2, which has improved accuracy, consistency, and movement dynamics. Available by subscription only.
- Gen-3 Alpha Turbo — a stripped-down version with the ability to generate video based only on ready-made images — Image to Video. It can be tested for free, albeit with limitations.
Users also have access to the Gen-2 model — the previous generation with less impressive generation quality. It is only interesting to try it out for comparison with the more advanced Gen-3 Alpha.
How to access the neural network
Just go to the official Runway ML website. It is open to users from Russia without any tricks.
How to access Runway ML Gen-3
To get started, you’ll need to create an account. You can do this via email or by using your Apple or Google accounts.
How to generate videos in Runway ML Gen 3
As we mentioned earlier, the free Gen-3 Alpha Turbo model only works with source images. Essentially, you upload the first frame and tell the neural network what to do next. Or you can just give it an image and give it complete freedom.
To get started, open the Runway ML editor. By default, it selects Gen-3 Alpha Turbo as the newest model.
Runway ML Gen-3 Alpha neural network editor.
Upload an image to the upper left section of the editor. You can click on it to open the file explorer, or drag and drop the image directly into the upload field. If you want this frame to be the last one instead of the first, click the Last button below the image.
Runway ML Gen-3 Alpha neural network editor
In the field below, describe the action you want to see in the finished video. It can be a simple plot, such as A man playing with puppy, or a complex prompt that includes details of the characters’ and camera’s movements.
It is better to write your request in English.
If you leave the field blank, the neural network will come up with the action itself. Usually, even the most basic prompt gives a more adequate result than the AI’s own invention.
Various presets are offered below the prompt input window. These are essentially text prompts that add a string of words to your request and allow you to achieve the desired effect — from dynamic action to slow motion and first-person mode.
To see all the options, click on the Custom Presets button on the left panel, under the T icon. In the Custom tab, you can create your own presets if you find a good combination of keywords that gives you the results you want.
Selecting presets in the Runway ML Gen-3 Alpha neural network
Clicking on the settings icon opens up a limited set of options. The only free option is to pin a seed, which is a string of numbers that helps the neural network refer to a specific generation and its parameters. This option is useful if you have previously generated a very successful video and want to make a new one in the same style.
The generation speed lives up to the “Turbo” in the name: it usually takes just a few seconds to create a video, and the queues are minimal. You can watch the video in the built-in player, download it to your device, and evaluate the quality of the neural network’s work.
There is also a Lip Sync button below the finished video. If you click on it, the neural network will analyze the person’s face in the frame (if there is one, of course) and prepare to overlay the voice with lip synchronization.
You can voice the text or use voice input — say the necessary phrases yourself or upload a ready-made audio file. If the source is text, you can also choose the voice.
If you are interested in generating a video without an image, you will have to downgrade the model to Gen-2. It is more erratic with movements and transitions than the third version, but it is available on free accounts.
How much does it cost?
Every second of generation with Gen-3 Alpha Turbo or Gen-2 costs 5 credits, so the initial bonus is used up very quickly. And considering that credits are not replenished, you will either have to pay or register a new account.
Gen-3 Alpha is only available by subscription, there are no free options.
Runway ML Gen-3 Alpha subscription price
Subscription prices start at $15 per month for unlocking all features and 625 credits on top. The maximum cost is $95 per month for unlimited generation with any model and 2,250 credits for additional features such as Lip Sync.