Emu Video - Historique des versions

Pitpitt le 30 avril 2024 à 01:05

2024-04-30T01:05:22Z

← Version précédente		Version du 29 avril 2024 à 21:05
Ligne 14 :		Ligne 14 :
	[https://emu-video.metademolab.com/ Source : Emu Video]		[https://emu-video.metademolab.com/ Source : Emu Video]

			[[Catégorie:ENGLISH]]

			[[Catégorie:GRAND LEXIQUE FRANÇAIS]]
	[[Catégorie:~~publication~~]]

Imeziani le 29 avril 2024 à 15:29

2024-04-29T15:29:24Z

← Version précédente		Version du 29 avril 2024 à 11:29
Ligne 1 :		Ligne 1 :
	== Définition ==		== Définition ==
	Méthode simple de génération de texte en vidéo basée sur des modèles de diffusion, factorisant la génération en deux étapes :		Méthode simple de génération de texte en vidéo basée sur des modèles de diffusion, factorisant la génération en deux étapes :
	Génération d'une image en fonction d'une invite textuelle		Génération d'une image en fonction d'une invite textuelle
	Génération d'une vidéo en fonction de l'invite et de l'image générée.		Génération d'une vidéo en fonction de l'invite et de l'image générée.

	== Français ==		== Français ==

Imeziani le 29 avril 2024 à 15:29

2024-04-29T15:29:08Z

← Version précédente		Version du 29 avril 2024 à 11:29
Ligne 1 :		Ligne 1 :
	~~==en construction==~~

	== Définition ==		== Définition ==
	~~XXXXXXXXX~~		Méthode simple de génération de texte en vidéo basée sur des modèles de diffusion, factorisant la génération en deux étapes :
			Génération d'une image en fonction d'une invite textuelle
			Génération d'une vidéo en fonction de l'invite et de l'image générée.

	== Français ==		== Français ==
Ligne 9 :		Ligne 9 :
	== Anglais ==		== Anglais ==
	''' Emu Video '''		''' Emu Video '''

	Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.

	==Sources==		==Sources==

	[https://~~arxiv~~.~~org~~/~~abs/2311.10709~~ Source : ~~arxiv~~]		[https://emu-video.metademolab.com/ Source : Emu Video]



	[[Catégorie:~~vocabulary~~]]		[[Catégorie:publication]]

Pitpitt : Remplacement de texte : « ↵↵↵==Sources== » par «  ==Sources== »

2024-01-31T15:03:14Z

Remplacement de texte : « ↵↵↵==Sources== » par «  ==Sources== »

← Version précédente		Version du 31 janvier 2024 à 11:03
Ligne 11 :		Ligne 11 :

	Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.		Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.


	==Sources==		==Sources==

Pitpitt : Remplacement de texte : « ↵↵↵↵ » par «   »

2024-01-29T17:36:03Z

Remplacement de texte : « ↵↵↵↵ » par «   »

← Version précédente		Version du 29 janvier 2024 à 13:36
Ligne 16 :		Ligne 16 :

	[https://arxiv.org/abs/2311.10709 Source : arxiv]		[https://arxiv.org/abs/2311.10709 Source : arxiv]




	[[Catégorie:vocabulary]]		[[Catégorie:vocabulary]]

Pitpitt : Remplacement de texte : « ↵ » par «  ==Sources== »

2024-01-28T02:22:47Z

Remplacement de texte : « ↵<small> » par «  ==Sources== »

← Version précédente		Version du 27 janvier 2024 à 22:22
Ligne 12 :		Ligne 12 :
	Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.		Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.

	~~<small>~~
			==Sources==

	[https://arxiv.org/abs/2311.10709 Source : arxiv]		[https://arxiv.org/abs/2311.10709 Source : arxiv]

Pitpitt : Page créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' Emu Video''' == Anglais == ''' Emu Video ''' Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to direct... »

2023-12-30T14:43:41Z

Page créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' Emu Video''' == Anglais == ''' Emu Video ''' Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to direct... »

Nouvelle page

==en construction==

== Définition ==
XXXXXXXXX

== Français ==
''' Emu Video'''

== Anglais ==
''' Emu Video '''

Emu Video, a text-to-video generation model that factorizes the generation into two steps: first generating an image conditioned on the text, and then generating a video conditioned on the text and the generated image. We identify critical design decisions--adjusted noise schedules for diffusion, and multi-stage training--that enable us to directly generate high quality and high resolution videos, without requiring a deep cascade of models as in prior work. In human evaluations, our generated videos are strongly preferred in quality compared to all prior work--81% vs. Google's Imagen Video, 90% vs. Nvidia's PYOCO, and 96% vs. Meta's Make-A-Video. Our model outperforms commercial solutions such as RunwayML's Gen2 and Pika Labs. Finally, our factorizing approach naturally lends itself to animating images based on a user's text prompt, where our generations are preferred 96% over prior work.

<small>

[https://arxiv.org/abs/2311.10709 Source : arxiv]

[[Catégorie:vocabulary]]

Emu Video - Historique des versions

Pitpitt le 30 avril 2024 à 01:05

Imeziani le 29 avril 2024 à 15:29

Imeziani le 29 avril 2024 à 15:29

Pitpitt : Remplacement de texte : « ↵↵↵==Sources== » par « ==Sources== »

Pitpitt : Remplacement de texte : « ↵↵↵↵ » par « »

Pitpitt : Remplacement de texte : « ↵ » par « ==Sources== »

Pitpitt : Remplacement de texte : « ↵↵↵==Sources== » par «  ==Sources== »

Pitpitt : Remplacement de texte : « ↵↵↵↵ » par «   »

Pitpitt : Remplacement de texte : « ↵ » par «  ==Sources== »