InstantFusion: A Shared Latent Interface Connecting Heterogeneous Image Generators
Shanghai-AI-Laboratory · hf · 2026-10-12
Diffusion/flow models are the de facto image generation paradigm, yet cross-model alignment of latent representations and consistency of denoising dynamics remain underexplored. The authors uncover a cross-model compatibility: independently trained generators show similar latent spatial organization, and directly handing off intermediate states between unaligned models still preserves coarse visual content.
They propose InstantFusion, a lightweight shared latent interface connecting denoising trajectories of heterogeneous generators:
- Model-specific Latent AutoEncoders (LAEs) project intermediate states into a shared latent space and decode them into a target generator's latent space.
- Two complementary objectives constrain the shared space: aligning intermediate states from the same image at matched noise levels, and synchronizing local denoising velocities.
Once trained, denoising transitions seamlessly across frozen generators without per-pair mappings. It demonstrates versatility in acceleration, multi-preference composition, cross-architecture on-policy distillation, and cross-model composition.
More from Multimodal
- Anthropic's bet on code over imagegen is being vindicated, observer argues — IndraVahan · 2026-10-12
- AI-made 5-min Indian history doc uses Claude, Gemini and Flux — deedydas · 2026-10-12
- MixTok hybrid tokenizer hits gFID 1.95 on ImageNet with just ~131 tokens per image — ZhejiangUniversity · 2026-10-12
- AI-generated mud hut photo goes viral: the tiny clay oven is the most convincing object in it — misovalko · 2026-10-12
- Lux In Music adds Yue2 model and multi-track MP3/FLAC separation, fully local — alexcovo_eth · 2026-10-12
- One prompt, one shot: Capsule agent produces a 15s kinetic-type launch video — DavidBennett__ · 2026-10-12