Research map / Image generation / Autoregressive and latent generation
High-Resolution Image Reconstruction With Latent Diffusion Models From Human Brain Activity
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use both decoded image latent z and decoded text latent c in reconstruction (ablation) | Stable Diffusion image reconstruction · NSD test images | identification accuracy | Helped |
| Use decoded image latent z alone, without text conditioning c (ablation) | Stable Diffusion image reconstruction · NSD test images | semantic fidelity | Hurt |
| Use decoded text latent c alone, without image latent z (ablation) | Stable Diffusion image reconstruction · NSD test images | visual consistency | Hurt |
| Use categorical image information instead of full-text annotations for c (ablation) | Stable Diffusion image reconstruction · NSD test images | semantic fidelity | Hurt |
| Use semantic maps instead of original images for z (ablation) | Stable Diffusion image reconstruction · NSD test images | semantic fidelity | Helped |
| Use semantic maps instead of original images for z (ablation) | Stable Diffusion image reconstruction · NSD test images | visual similarity | Hurt |
| Increase noise added to z when generating zc (ablation) | Stable Diffusion image-to-image reconstruction · NSD, higher visual cortex | unique variance explained by zc | Helped |
| Use zc from the middle rather than early denoising stage (ablation) | Stable Diffusion image-to-image reconstruction · NSD, higher visual cortex | unique variance explained by zc | Helped |
| Use the early U-Net layer rather than the bottleneck layer as denoising progresses (ablation) | Stable Diffusion U-Net · NSD, early visual cortex | fMRI prediction performance | Helped |
| Use the bottleneck U-Net layer as denoising progresses (ablation) | Stable Diffusion U-Net · NSD, higher visual cortex | fMRI prediction performance | Helped |
Closest papers by meaning
- Controllable Mind Visual Diffusion Model (AAAI 2024 · similarity 0.88)
- Brain decoding: toward real-time reconstruction of visual perception (ICLR 2024 · similarity 0.87)
- MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data (ICML 2024 · similarity 0.87)
- Neuro-Vision to Language: Enhancing Brain Recording-based Visual Reconstruction and Language Interaction (NeurIPS 2024 · similarity 0.87)
- NeuroClips: Towards High-fidelity and Smooth fMRI-to-Video Reconstruction (NeurIPS 2024 · similarity 0.86)
- Reanimating Images using Neural Representations of Dynamic Stimuli (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.