Form: Variance / Convergence

Form: Variance / Convergence begins with a condition concerning the limits of the image: when multiple individuals attempt to approximate the same fictional character through cosplay, the character gradually detaches from narrative and becomes a recognizable and measurable form.

Following this shift, the project approaches images as part of an ongoing process of standardization. Across cosplay, digital retouching, and AI generation, bodies are not simply represented, but continuously adjusted, converging toward a stable and recognizable visual model.

Each work is structured as a dual matrix: a grid of human performers and a corresponding grid of AI-generated images. In the human matrix, the character disperses across bodies that vary in age, gender, and physical condition, forming a field of differences that resists complete unification. In the AI matrix, these differences are no longer maintained, but translated into consistent and repeatable outputs.

Within this process, the human images function as a form of “donor.” Through structural conditioning and algorithmic denoising, they are gradually transformed toward a stabilized visual state. This operation approaches a kind of digital taxidermy: surfaces are smoothed, details compressed, and backgrounds reformatted into decontextualized, uniform fields. The image no longer operates purely as representation, but as a processed form of preservation.

Yet this stabilization is not seamless. Minor inconsistencies remain visible—slight distortions, residual green spill along the edges. These are not merely errors, but traces of an incomplete conversion, where certain differences have not been fully absorbed.

What emerges is not simply the appearance of a character, but the formation of a visual standard. This standard is not produced by a single source; it is continuously shaped through the interaction of individual desire, cultural circulation, and computational processes.

As images become more stable, visibility itself becomes conditional. Certain bodies are more easily recognized and reproduced, while others drift beyond what can be consistently rendered. These differences do not disappear, but their visibility becomes uncertain.

Form does not resolve this mechanism, but makes it observable. As images approach a predictable form, what we encounter may no longer be the character itself, but a consensus under compression—and within that process, something begins to recede from view.

Form does not document this transformation; it exposes its mechanism. The question is no longer how closely one can resemble the image, but who must disappear for the image to remain stable.

《Form：差異 / 趨同》
《Form：差異 / 趨同》起始於一個關於影像極限的條件：當眾多個體試圖透過 Cosplay 趨近同一個虛構角色時，角色逐漸脫離敘事，轉化為一種可被辨識、比較與量化的形式。

本計畫沿著這個轉化，將影像視為一套持續運作的標準化機制。在cosplay、數位修圖與AI生成的交互作用中，身體並非單純被再現，而是在反覆調整中，逐步朝向一種可被穩定辨識的視覺模型收斂。

每件作品由雙矩陣構成：真人影像與AI生成影像並列。在真人矩陣中，角色分散於不同年齡、性別與身體條件之間，這些差異構成一個難以被完全統一的場域；而在AI矩陣中，這些差異不再被保留，而是被轉譯為一致且可重複的結果。

在這樣的轉換過程中，我將真人影像作為一種「供體」，透過演算法的形體鎖定與去噪提純，使其逐步趨近於一種穩定的影像狀態。這個過程近似於一種數位標本化：皮膚被平滑、細節被壓縮，背景被格式化為去脈絡的純色空間。影像不再僅是再現，而更接近於一種經過處理後的保存形式。

然而，這種穩定並非完全無縫。在AI影像中仍可觀察到細微的不一致——略微偏移的結構、邊緣滲出的綠色殘影。這些並非單純的錯誤，而更像是轉換過程中的殘餘訊號，顯示某些差異尚未被完全消化。

在此脈絡下，影像所呈現的不僅是角色的外觀，而是某種逐漸形成的標準。這種標準並非來自單一來源，而是在個體慾望、文化流通與計算過程之間被不斷生成與強化。

當影像趨於穩定，可見性也開始變得有條件。某些身體更容易被辨識與重複，而另一些則逐漸偏離這個可被接受的範圍。這些差異並未完全消失，但其位置變得模糊，甚至難以被清晰看見。

《Form》並不試圖對此機制下定結論，而是將其展開：當影像越來越接近一種可被預測的形式時，我們所看到的，究竟是角色本身，還是某種被持續壓縮後的共識？而在這個過程中，又有哪些部分，逐漸變得不可見？