Luma Dream Machine is a next generation video model for creating high quality videos with text prompts
1 min. read
Published on
Share this article
Improve this guide
Read our disclosure page to find out how can you help MSPoweruser sustain the editorial team Read more
Key notes
- Dream Machine also supports array of fluid, cinematic and naturalistic camera motions matching the emption and contents of the scene.
Luma Labs today announced Dream Machine, an Gen AI model that can generate high quality and realistic videos based on text prompts and images. Dream Machine video model was built on a multimodal transformer architecture and it was trained directly on videos. Luma Labs mentioned that Dream Machine can generate physically accurate, consistent and action-packed scenes. The best part is that anyone can try the model for free here.
Dream Machine is an incredibly fast video generator. It can generate120 frames in 120s allowing to iterate faster. Dream Machine can generate 5s shots with a realistic smooth motion, cinematography, and drama. Also, since Dream Machine understands how people, animals and objects interact with the physical world, you can create videos with great character consistency and accurate physics.
Luma Labs mentioned that that Dream Machine is one of their frontier generative models. Unlike the several existing image-animation models, Dream Machine is a true video generation model. To train this model, Luma Labs used Amazon Web Services and its NVIDIA H100 training infrastructure SageMaker HyperPod.
