Uploaded October 2022 | Updated September 2026, 2 weeks ago
Text-to-image models are a novel machine learning technology that can generate high-quality, photorealistic images from a simple text prompt. Within Google Research, our scientists and engineers have been exploring text-to-image generation using a variety of AI techniques.
Imagen and Parti, Google’s text-to-image models, have the ability to generate photorealistic images but use different approaches. Imagen is a Diffusion model, which learns to convert a pattern of random dots to images. Parti’s approach takes advantage of existing research and infrastructure for large language models such as PaLM and is critical for handling long, complex text prompts to produce high-quality images.
In this third episode of #MeetAGoogleResearcher, Drew Calcagno speaks with Google Research Scientists Mohammad Norouzi - who worked on Imagen - and Jason Baldridge - who worked on Parti. They’ll talk about their role in creating each of these models, how they’re making the art process more accessible to artists and non-artists alike, and the future of text-to-image generation.
Watch and see how these models are bridging quantitative and qualitative fields, while giving us insight to our own knowledge of the real world.
Resources:
Check out Imagen → https://goo.gle/3yvFNsS
Check out Parti → https://goo.gle/3Vk5wi2
Check out the “How AI creates photorealistic images from text” blog post → https://goo.gle/3CJQ5YW
What is PaLM? Check out our blog post → https://goo.gle/3ehvXE8
Follow Mohammad on Twitter → https://goo.gle/3CqtJdw
Follow Jason on Twitter → https://goo.gle/3CoTbjD
Follow Drew on Twitter → https://goo.gle/3BTvKR8
Follow Drew on YouTube → https://goo.gle/3PaLrGo
Chapters:
0:00 - Intro
1:22 - Speaker intros
2:15 - How did you get into this text-to-image field? How dis this field come to what it is today?
2:51 - What is Imagen and Parti? Which one did you contribute to?
6:41 - What got you into this field?
10:10 - What are the guardrails of Imagen and Parti?
11:52 - Favorite images you’ve made?
13:54 - What is the ultimate goal of Imagen and Parti?
14:59 - What’s the future of text-to-image models?
18:11 - How will Imagen and Parti help people?
20:05 - Outro
Watch more episodes of Meet A Google Researcher → https://goo.gle/MeetAGoogleResearcher
Subscribe to the Google Research Channel → https://goo.gle/GoogleResearch
#MeetAGoogleResearcher #imagen
Text-to-image models are a novel machine learning technology that can generate high-quality, photorealistic images from a simple text prompt. Within Google Research, our scientists and engineers have been exploring text-to-image generation using a variety of AI techniques.
Imagen and Parti, Google’s text-to-image models, have the ability to generate photorealistic images but use different approaches. Imagen is a Diffusion model, which learns to convert a pattern of random dots to images. Parti’s approach takes advantage of existing research and infrastructure for large language models such as PaLM and is critical for handling long, complex text prompts to produce high-quality images.
In this third episode of #MeetAGoogleResearcher, Drew Calcagno speaks with Google Research Scientists Mohammad Norouzi - who worked on Imagen - and Jason Baldridge - who worked on Parti. They’ll talk about their role in creating each of these models, how they’re making the art process more accessible to artists and non-artists alike, and the future of text-to-image generation.
Watch and see how these models are bridging quantitative and qualitative fields, while giving us insight to our own knowledge of the real world.
Resources:
Check out Imagen → https://goo.gle/3yvFNsS
Check out Parti → https://goo.gle/3Vk5wi2
Check out the “How AI creates photorealistic images from text” blog post → https://goo.gle/3CJQ5YW
What is PaLM? Check out our blog post → https://goo.gle/3ehvXE8
Follow Mohammad on Twitter → https://goo.gle/3CqtJdw
Follow Jason on Twitter → https://goo.gle/3CoTbjD
Follow Drew on Twitter → https://goo.gle/3BTvKR8
Follow Drew on YouTube → https://goo.gle/3PaLrGo
Chapters:
0:00 - Intro
1:22 - Speaker intros
2:15 - How did you get into this text-to-image field? How dis this field come to what it is today?
2:51 - What is Imagen and Parti? Which one did you contribute to?
6:41 - What got you into this field?
10:10 - What are the guardrails of Imagen and Parti?
11:52 - Favorite images you’ve made?
13:54 - What is the ultimate goal of Imagen and Parti?
14:59 - What’s the future of text-to-image models?
18:11 - How will Imagen and Parti help people?
20:05 - Outro
Watch more episodes of Meet A Google Researcher → https://goo.gle/MeetAGoogleResearcher
Subscribe to the Google Research Channel → https://goo.gle/GoogleResearch
#MeetAGoogleResearcher #imagen
