GPT-Image-2 Case · Denoising Diffusion Forward/Reverse Chain

A GPT-Image-2 reference case selected from 'Research Paper Illustrations,' suitable for exploring generative effects in the 'denoising diffusion forward/reverse chain' direction.

Author: AI Plus Lab

✦Results

◌Case Background

来自 GPT-Image2-Skill README 的精选展示条目，适合作为“研究论文图示 / 去噪扩散正/逆向链”方向的站内参考案例。

⌘Prompt Content

16:9 Academic Illustration, showing diffusion forward + reverse chains, with two horizontal chains stacked vertically.

Top chain (left → right) labeled "Forward diffusion q(x_t | x_{t-1})": Five images "x_0", "x_{T/4}", "x_{T/2}", "x_{3T/4}", "x_T", transitioning from a clear scene of hills and sun to pure Gaussian noise. Arrows between images labeled "+ β_t ε".

Bottom chain (right → left) labeled "Reverse denoising p_θ(x_{t-1} | x_t)": Similarly five images in reverse order, with a small hexagonal block ε_θ(x_t, t) between each pair.

Curved arrow "T diffusion steps" on the far right connects the top right and bottom right; curved arrow "sample x_0" on the far left connects the bottom left and top left.

Title: "Denoising Diffusion: forward corruption and learned reverse". Subtitle: "Ho et al., 2020".

✎Outcome Notes

This case has been migrated from the GPT-Image2-Skill README to the AIPlusLab `/prompts` for easy in-site search, browsing, and reuse.