Same Prompt, 3 Years Apart: AI Video Quality Leap
chrisfirst · x · 2026-08-19
The post compares a video generated from the same prompt in August 2023 versus August 2026. The original video was created with Pika, while the latest one uses Seedance 2.5. The comparison demonstrates an insane advancement in AI video generation quality over just three years.
More from Multimodal
- PNS paper: trainable neural subdivision augments Loop subdivision with bounded corrections — ssh4net · 2026-08-19
- Creating Vampire Hunter D Style Character with T2V and R2VA — SIR_NVAX_A_LOT · 2026-08-19
- Seedance 2.5 video model test shows significant boost in consistency — SimplyAnnisa · 2026-08-19
- MLLM-Guided Semantic Correction for Text-to-Video Generation — kwangmoo_yi · 2026-08-19
- H3 video generation struggles with cross physics in 'Passion of the Sponge' — BuffaloDesperate8357 · 2026-08-19
- Help: Adding Video LoRAs to LTX 2.3 for fight scene animations — SayethTwat · 2026-08-19