English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Classifier-free guidance is an effective sampling technique in diffusionmodels that has been widely adopted. The main idea is to extrapolate the modelin the direction of text guidance and away from null-text guidance. In thispaper, we demonstrate that null-text guidance in diffusion models is secretly acartoon-style creator, i.e., the generated images can be efficientlytransformed into cartoons by simply perturbing the null-text guidance.Specifically, we proposed two disturbance methods, i.e., Rollback disturbance(Back-D) and Image disturbance (Image-D), to construct misalignment between thenoisy images used for predicting null-text guidance and text guidance(subsequently referred to as \textbf{null-text noisy image} and \textbf{textnoisy image} respectively) in the sampling process. Back-D achievescartoonization by altering the noise level of null-text noisy image viareplacing $x_t$ with $x_{t+\Delta t}$. Image-D, alternatively, produceshigh-fidelity, diverse cartoons by defining $x_t$ as a clean input image, whichfurther improves the incorporation of finer image details. Throughcomprehensive experiments, we delved into the principle of noise disturbing fornull-text and uncovered that the efficacy of disturbance depends on thecorrelation between the null-text noisy image and the source image. Moreover,our proposed techniques, which can generate cartoon images and cartoonizespecific ones, are training-free and easily integrated as a plug-and-playcomponent in any classifier-free guided diffusion model. Project page isavailable at \url{https://nulltextforcartoon.github.io/}.

Null-text Guidance in Diffusion Models is Secretly a Cartoon-style Creator