A video model trained on 5-second windows can now generate 4-minute clips
imjustnewatai · x · 2026-07-27
A video model trained on only 5-second windows can now generate up to 4 minutes of video.
- The method, called Self Gradient Forcing, lets future frames teach the model how earlier frames should store memory.
- It reportedly preserves the person, background, and layout much better than previous approaches over long generations.
- The post frames this as a step toward generated worlds that do not quickly drift, melt, or forget what happened.
Related event: Video Model Trained in 5 Seconds Generates 4-Minute Clips(2 posts)→
More from Multimodal
- Third-party test: Claude Opus 5.5 renders finer 3D scenes but costs 13x more than GPT-6 Sol — testingcatalog · 2026-09-23
- ComfyUI trick: aux preprocessor + Qwen transfers poses across characters with one prompt — Acceptable-Work8202 · 2026-09-23
- Same portrait prompt across Midjourney V6.1, V7 and V8.2: do older models look better? — tisch_eins · 2026-09-23
- Testing AI character consistency across a 20-image travel sequence — SiennaVaire · 2026-09-23
- Midjourney v8.2 Faces: New Portrait Generation Samples Shared — azed_ai · 2026-09-23
- One-sentence prompt generates lifelike dog video, shown side-by-side with the real one — wgrathwohl · 2026-09-23