Hello NVIDIA team
I use RTX Video Super Resolution (RTX VSR) every day for watching local videos (mostly 1080p → 1440p, and soon 1080p → 4K). I want to start by saying thank you — RTX VSR has genuinely improved my viewing experience.
When the source is good (1080p or 720p with a high bitrate), the image on my 1440p monitor with RTX VSR enabled is extremely close to 4K. I even paused the video on the same frame and compared the pixels side‑by‑side, and the difference was almost nonexistent. It’s really impressive.
With lower‑quality sources (YouTube, older videos, low bitrate), VSR doesn’t manage to fix the artifacts as well, so I’m curious whether improvements in this area are planned.
I also want to mention that RTX VSR works very well with high‑quality local sources. With good 1080p WEB‑Rip files or Blu‑ray REMUX videos, the upscaling results are consistently clean and detailed. When playing a 1080p REMUX on a 4K TV at a normal viewing distance, it becomes difficult for most people to tell that the original source is only 1080p. The improvement with VSR enabled is subtle but meaningful, and it brings the image noticeably closer to a native 4K presentation.
I also want to say that I’m honestly impressed by your engineering work. RTX VSR is one of the most useful and impressive AI features I’ve ever used, and I really appreciate the effort behind it. I hope this technology continues to grow, because it has huge potential.
I have two questions I’d really like to ask:
First, are you currently working on new improvements or new AI models for RTX VSR? VSR 1.5 is already great, but there is still room for improvement, especially with low‑quality sources, motion handling, and fine detail reconstruction. I know that RTX VSR already works with AV1 and that RTX Video HDR can run together with VSR, but I’m wondering if future updates might improve the way HDR and VSR interact — especially in scenes where fine details can become slightly over‑brightened. I’m also curious whether higher VSR levels, like Level 5 or even Level 6, are something you’re considering for future releases.
Second, are you planning to add AI frame interpolation to RTX Video?
For example, converting 23.976 fps video to 50 or 60 fps during playback. Since NVIDIA already has Optical Flow Accelerator, DLSS Frame Generation, Maxine Video Effects, and other motion‑related AI technologies, it would be amazing to see real‑time AI FPS interpolation for video playback.
I also have a small suggestion:
It would be great if RTX VSR could be available as an FFmpeg filter or API for offline video processing. Many users would love to use VSR not only during playback, but also for high‑quality offline upscaling (1080p → 4K) with larger AI models.
Thank you again for your work — I use RTX VSR every day and I’m genuinely impressed by it. I’m looking forward to future improvements.
Best regards,