Research map / Image generation / Text-to-image diffusion
Follow-Your-Mind: Towards Inversion-Free Brain-Driven Visual Context Synthesis and Editing
Haodong Jing, Panqi Yang, Rongchao Zhang, Zhipeng Liu, Yajun Liu, Xuehai Bai, Yongqiang Ma, Nanning Zheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add RoPE temporal encoding instead of absolute position embedding (ablation) | MinD-Flow · Neuro-Creation | CLIP-I: 0.6415 -> 0.6723 | Improved |
| use adaptive gated fusion instead of concatenation (ablation) | MinD-Flow · Neuro-Creation | CLIP-I: 0.7356 -> 0.7618 | Improved |
| use adaptive gated fusion instead of cross-attention fusion (ablation) | MinD-Flow · Neuro-Creation | CLIP-I: 0.7489 -> 0.7618 | Improved |
| use decoupled flow guidance instead of classifier-free guidance (ablation) | MinD-Flow · Neuro-Editing | LPIPS: 0.3685 -> 0.3195 | Improved |
| use decoupled flow guidance instead of CFG (ablation) | MinD-Flow · Neuro-Creation | CLIP-I: 0.7105 -> 0.7618 | Improved |
| add semantic CLIP alignment loss L_sem (ablation) | MinD-Flow · Neuro-Creation | CLIP-I: 0.7512 -> 0.7618 | Improved |
| switch from MSE loss to flow-matching loss (ablation) | MinD-Flow baseline config · Neuro-Creation | CLIP-I: 0.6415 -> 0.6723 | Improved |
| inversion-free 5-step MinD-Flow vs Null-Text inversion 50 steps | MinD-Flow · Neuro-Editing | CLIP-I: 0.7453 -> 0.7638 | Improved |
| inversion-free 5-step MinD-Flow vs Null-Text inversion 50 steps | MinD-Flow · Neuro-Editing | L1: 0.1394 -> 0.1356 | Improved |
| reduce Null-Text inversion from 50 to 10 steps | Null-Text Inversion · Neuro-Editing | L1: 0.1394 -> 0.1917 | Hurt |
| neuro-adaptive soft mask vs no mask or hard mask (ablation) | MinD-Flow · Neuro-Editing | structure vs semantics | Improved |
| MinD-Flow vs MindCustomer on Neuro-Context | MinD-Flow · Neuro-Context | L1: 0.1583 -> 0.1427 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MindCustomer — MindCustomer: Multi-Context Image Generation Blended with Brain Signal (ICML 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MindCustomer: Multi-Context Image Generation Blended with Brain Signal (ICML 2025 · similarity 0.87)
- MindPainter: Efficient Brain-Conditioned Painting of Natural Images via Cross-Modal Self-Supervised Learning (AAAI 2025 · similarity 0.87)
- Neural-Driven Image Editing (NeurIPS 2025 · similarity 0.86)
- Mind Artist: Creating Artistic Snapshots with Human Thought (CVPR 2024 · similarity 0.86)
- NEvo: Neural-Guided Evolutionary Video Synthesis for Dynamic Visual Selectivity (NeurIPS 2026 · similarity 0.86)
- MoRE-Brain: Routed Mixture of Experts for Interpretable and Generalizable Cross-Subject fMRI Visual Decoding (NeurIPS 2025 · similarity 0.85)
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.
