Skip navigation
CosmicMan A Text-to-Image Foundation Model for Humans We present CosmicMan, a text-to-image…
0:00 / 0:50

CosmicMan A Text-to-Image Foundation Model for Humans We present CosmicMan, a text-to-image…

AK

@_akhaliq458K followers

Visit on X
232 likesDownloadOriginal post

46,778 views

CosmicMan A Text-to-Image Foundation Model for Humans We present CosmicMan, a text-to-image foundation model specialized for generating high-fidelity human images. Unlike current general-purpose foundation models that are stuck in the dilemma of inferior quality and

9 Comments

Comments from the original post on X (Twitter)

  • AK's profile picture

    @_akhaliq

    text-image misalignment for humans, CosmicMan enables generating photo-realistic human images with meticulous appearance, reasonable structure, and precise text-image alignment with detailed dense descriptions. At the heart of CosmicMan's success are the new reflections and

    3 likes

  • AK's profile picture

    @_akhaliq

    perspectives on data and models: (1) We found that data quality and a scalable data production flow are essential for the final results from trained models. Hence, we propose a new data production paradigm, Annotate Anyone, which serves as a perpetual data flywheel to produce

    3 likes

  • AK's profile picture

    @_akhaliq

    high-quality data with accurate yet cost-effective annotations over time. Based on this, we constructed a large-scale dataset, CosmicMan-HQ 1.0, with 6 Million high-quality real-world human images in a mean resolution of 1488x1255, and attached with precise text annotations

    3 likes

  • AK's profile picture

    @_akhaliq

    deriving from 115 Million attributes in diverse granularities. (2) We argue that a text-to-image foundation model specialized for humans must be pragmatic -- easy to integrate into down-streaming tasks while effective in producing high-quality human images. Hence, we propose to

    4 likes

  • AK's profile picture

    @_akhaliq

    model the relationship between dense text descriptions and image pixels in a decomposed manner, and present Decomposed-Attention-Refocusing (Daring) training framework. It seamlessly decomposes the cross-attention features in existing text-to-image diffusion

    3 likes

  • AK's profile picture

    @_akhaliq

    model, and enforces attention refocusing without adding extra modules. Through Daring, we show that explicitly discretizing continuous text space into several basic groups that align with human body structure is the key to tackling the misalignment problem in a breeze.

    3 likes

  • AK's profile picture

    @_akhaliq

    paper page:

    17 likes

  • AK's profile picture

    @_akhaliq

    daily papers:

    4 likes

  • Bobcat's profile picture

    @somebobcat8327

    👀

    1 like