跳过导航
our new model Atlas also happens to be a capable text-to-image generator, providing it with a strong…
0:00 / 0:24

our new model Atlas also happens to be a capable text-to-image generator, providing it with a strong…

Ben Mildenhall

@BenMildenhall1.2万 位关注者

在 X 上查看
1,501 个赞下载原帖

105,205 次观看

our new model Atlas also happens to be a capable text-to-image generator, providing it with a strong foundation of “world knowledge”... this generalizes across both single and multi view domains, meaning you can step into even highly stylized scenes with precise 3D control

36 条评论

评论来自 X (Twitter) 上的原帖

  • Keenan Crane 的头像

    @keenanisalive

    Question purely out of curiosity: can you recover an explicit 3D representation (say, a textured mesh) from Atlas, such that renderings of the explicit representation closely reproduce the video (say, for example “hard surface” models and relatively simple materials)? This kind of question often gets side stepped (e.g., “meshes are old news”), but for me it’s a scientific question rather than an application question: what information is the model predicting, and which variables are still underdetermined?

    61 个赞

  • Thariq 的头像

    @trq212

    Incredible, congrats Ben!

    15 个赞

  • Mazeyar 的头像

    @mazeyar

    Can you let me play GTA VI in Atlas 😌🙏🏼

    8 个赞

  • Ben Mildenhall 的头像

    @BenMildenhall

    yes

    53 个赞

  • Mazeyar 的头像

    @mazeyar

    Hahaha thanks 🤣

  • Mathew Tizard 的头像

    @mtizard

    Incredible work. If I have a variety of viewpoints of the same scene, can Atlas ingest all that ground truth and use it to synthesise the missing perspectives? Would this work with synchronised video frames too?

    7 个赞

  • Ben Mildenhall 的头像

    @BenMildenhall

    exactly!

    4 个赞

  • POM 的头像

    @peterom

    Beautiful style adherence! The fabric on the octopus is amazing

    4 个赞

  • bone 的头像

    @boneGPT

    very cool, can you show it doing a couple 360s? would like to see if it keeps persistence

    4 个赞

  • Ian Curtis 的头像

    @XRarchitect

    this one 🫠 so good

    2 个赞

  • Nicholas Bardy 的头像

    @NicholasBardy

    nice work

    1 个赞

  • Troy Kirwin 的头像

    @tkexpress11

    Sick

    1 个赞

  • XGEN Labs 的头像

    @XGEN_labs

    The move from image generation to controllable 3D worlds is especially exciting. Curious how far this kind of world knowledge can generalize across unseen environments and viewpoints.

    1 个赞

  • Doug Thompson 的头像

    @Dusanwriter

    I was waiting for this. Marble etc were awesome but I can't wait to experiment. Working on history projects (London UK, really top people) and we have such deep ideas but needed something like Atlas to get us there. Storytelling has a new tool kit and you guys rocking it. EPIC

    1 个赞

  • Yücel İnanoğulları 的头像

    @yuceli

    Very impressive. I wonder if one day, AI would read a novel and produce a cinematic version of it.

    1 个赞

  • Nicholas H 的头像

    @itsnicholash

    wow that 2D japanese scene though

    1 个赞

  • ritwik 的头像

    @ritwik_raha

    Takes one back to NeRF days

  • Amara Wolfe 的头像

    @AmaraWolfeeb

    Claiming Atlas has 'world knowledge' from 2D text-to-image training is a massive stretch. How does it handle actual physical consistency and occlusion when you try to navigate those stylized scenes?

  • Peachy J 的头像

    @IDGAFACCP

    That world knowledge foundation generalizing across views is clever.

  • COSMOS 的头像

    @outofcosmos

    wf