Luma Dream Machine Tutorial

Luma Labs Dream Machine is an AI powered text-to-video and image-to-video generator which you can use to create some jaw-dropping video clips free of charge. This quick tutorial will walk you through the process in a nutshell and answer some frequently asked questions about using Luma AI.

Basic Steps to Generate Luma Videos

To get started creating your first Luma videos, follow these simple steps:

1. Head over to https://lumalabs.ai/dream-machine click "Try Now" and create an account using your email. Might be best to avoid using a VPN or you might get blocked.

2. Open the "Create" section. This is where you'll be hanging out https://lumalabs.ai/dream-machine/creations. It will later contain all your generated videos, but initially just going to offer you a text/image fields and an up arrow button to submit your video generation requests.

3. Give AI your prompt. Prompt is like an instruction or an order. What would you like to see?
   

   

   Next to the up arrow is the number of generations you have left. Mine screenshot shows 0 as I've currently spent them all.

   Your prompting options are:

   • Fill out text prompt only. If you need prompt examples, check this post.
   • Upload an image (or 2 images) only. For using images as keyframes check this guide.
   • Upload both an image/2 images and also supply some text.

   Additionally, once you start making any changes, you'll see 2 checkboxes pop up: Loop & Enhance prompt. 

   Loop will create some sort of a scenario where end of your clip is connected to the beginning again. For instance:

   

   Enhance prompt is enabled by default, and you might want to leave it there if you're new to this, letting AI figure out how best to animate your text and/or image prompts. Unchecking it basically means you'll describe everything yourself, particularly camera movements and other dynamic visual effects. Unchek it if instead of doing what you're calling for in your prompt, AI is doing something completely different. (Though be prepared sometimes it will not help.) Follow this prompt structure especially if you've unchecked prompt enhancement: Camera Shot + object or subject, camera motion + description of events {optional: cinematic style, vibe, any additional details}

4. Once you're finished entering your prompt, click the up arrow button and sit tight - your video is now added to generation queue.

   

Camera Motions in Dream Machine

In the beginning of September 2024 LumaLabs added a feature of selecting camera motions from a handy drop-down suggestions list.

Just type the word "camera" anywhere in your prompt to trigger the list of camera motions.

Following camera motions are available:

• Move Left/Right
• Move Up/Down
• Push In/Out
• Pan Left/Right
• Orbit Left/Right
• Crane Up/Down

LumaLabs Dream Machine Questions Answered

What kind of videos can I create with Luma Dream Machine?

You can create a variety of short, high-quality videos, including realistic scenes of people, animals, and objects interacting with the physical world. LumaLabs Machine supports text-to-video and image-to-video generation.

Is Luma AI Free to Use?

Luma Labs has introduced both free and paid options. The free plan allows you to try the model's full capabilities: same resolution, all the same functionalities as paid users, but you're only allowed 30 generations per month, and no more than 10 a day. Meaning, if you'd rather burn your free monthly allowance in just 3 days, you can generate 10 videos today and 10 tomorrow and 10 the day after - and that's it, you have to wait 27 days more for your 30 credits to renew. Update: now it's 4 generations per day for free tiers.

How long does 'in queue' status last in Luma Dream Machine?

Once you've submitted your video for generation, it first gets in Dream Machine queue. How long it will sit there depends on whether you're on a free or paid plan. From my personal experience, free plan submissions can be held in waiting for hours, whereas on a Standard Plan it usually takes up to a few minutes to get your video generated. I've generated over 200 videos and personally have never had a single generation fail, but some users report on reddit that occasionally a generation might get 'stuck' in queue forever, while their newer ones get through fine. So just be aware there could be bugs.

When do credits renew?

Open the Account section at https://lumalabs.ai/dream-machine/account to check when your free and paid credits renew.

How to get highest resolution, high quality videos from Luma?

Use the highest quality images. Animating high resolution images produces high resolution videos, whereas text prompts can yield somewhat lower resolution results. Source your high res images from AI generators like Midjourney, DALL·E, Leonardo. Upload your image(s) to Luma and click the up arrow button to bring them to life.

Why is my prompt not working?

There can be many reasons. It could be the way you've phrased it, or it can be current AI's limitations. It's hard to tell without seeing your exact prompt. Sometimes some prompts will not work, and you're best to move on. Or you might keep burning your credits trying to implement the same idea which might not be possible right now. If you'd like to see how many of my prompts went completely different direction, you can watch this following video.

What are Luma AI's limitations?

While impressive, video generating AIs, including Luma, have their limitations, things they struggle with or bugs.

For example, all visual AI's, image and video generators, currently struggle with human fingers. Body disfigurement and anatomical anomalies are still common. Typically, best bet is slow motion movements. So perhaps aim to create visually captivating effects while keeping movements nice and smooth. In many cases, adding "slow motion" to the prompt will produce a more reliable result.

How to prevent any camera movement (get static shot)?

1. Uncheck 'Enhance Prompt'. If left on, AI basically has more freedom to add whatever it sees fit to your prompt, including in terms of the camera movement.
2. Explicitly State "Static Shot": This sets the expectation that there will be no camera movement. Works both at the start or at the end of the prompt.
3. Avoid Descriptive Camera Movements: Avoid terms like "pan," "zoom," "dolly," "tilt," or any mention of camera motion. These imply movement.

For example, you could structure the prompt like this: Static shot of [subject] centered in the frame, with no camera movement. The focus is on [specific actions or expressions], maintaining a steady and fixed perspective throughout. or Static wide shot of a man, focused on his subtle breathing and blinking, conveying a calm and contemplative atmosphere.

Can i change the dimensions of the video in Lumalabs?

Currently this is not possible for text-to-video, only for image-to-video generations. In the case of an image prompt, Luma will stick to the dimensions of your image file. So one workaround is to generate an image first in your preferred AI image generator with the desired orientation and aspect ratio and then upload it to Dream Machine for animation.

Can I delete my Luma video?

If you'd like to remove any generated video in Luma, you have to open its dedicated page by clicking on it first. there, you'll find options to extend, download or delete.

What are Dream Machine Alternatives?

To generate videos from text or images, you can try such alternative solutions as Kling AI, Runwayml, Haiper, and Pika. These companies offer free AI video generation under various conditions. Leonardo AI also offers 4-second long videos, but strictly from images created on its platform.

If you'd like a more information with examples, please browse around this site under #dreammachine tag, and you can also find Luma Labs own guide here.

Last modified 05 September 2024 at 10:21