Tried EdenLive 5B on my game footage with an RTX 5090. Backgrounds looked great, but moving objects fell apart.
I tried EdenLive 5B AR through LightX2V on footage from a game I'm developing after seeing the official demos.
Proof1 community thread
Key points
- The background transformation was surprisingly good.
- I used standard SDPA with compilation and CUDA Graphs disabled, so the processing time isn't comparable to the official benchmarks.
- Has anyone tried EdenLive on game footage with lots of moving objects?
- I'm curious whether different settings or input footage make a noticeable difference.
Sources (1)
- [1]Tried EdenLive 5B on my game footage with an RTX 5090. Backgrounds looked great, but moving objects fell apart.r/StableDiffusion (top, day) · Oct 22, 04:36 PM
I tried EdenLive 5B AR through LightX2V on footage from a game I'm developing after seeing the official demos.
The background transformation was surprisingly good.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 22, 2026Kitten-TTS 2: New Open-source SOTA 1.58 bit TTS (CPU-only with voice cloning, multilingual, emotion control)
- Oct 21, 2026Mica v0.5: The smartest 4B decision model is also the fastest one. 20 ms per answer, 25x faster than Jev
- Oct 20, 2026Gradium Open-Sources Phonon, a 31M-Parameter On-Device TTS Model Under Apache 2.0
- Oct 20, 2026[AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
- Oct 20, 2026Do Diffusion LLMs Scale? Testing Mercury 3 + Decide
- Oct 20, 2026MPS in PyTorch 2.15
