not a shader person, but I would think something like an overexposure fade type thing might work. like if the pixels go white they can't immediately go black or whatever, 500ms fade to normal function. Not exactly sure how it would affect the image. also would depend on situation
I know I follow some people who are all about shader tech, so I'm posing this question: If you had to limit the potential for photosensitivity triggers entirely in post-processing, on an engine using TAA, without the option to keep RT history, what would you do?
-
-
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
Do you by chance have a concise, accurate description of what specifically causes problems? I know strobing/flashing but that isn't good enough to act on :)
-
generally it is a frequency thing. the pokemon issue was the scenes in question alternated solid red and blue background shots every frame
- Näytä vastaukset
Uusi keskustelu -
-
-
Super cool problem. Without history is also tough (if you have TAA you do have history). De-strobing timelapse photos seem to focus on exposure consistency. Wonder about luminance history and zonal tonemap/adaptation (instead of 1 lum value, have 4x4)
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
local autoexposure + tonemapping with a fast adaptation rate might be a good starting point. https://bartwronski.com/2016/08/29/localized-tonemapping/ …
-
Thanks for the suggestion! Will look into it. Also, apropos of nothing, you look *exactly* like a genius coder roommate of mine from like 2010-2012, also named Kara. She was a kickass genius.
Keskustelun loppu
Uusi keskustelu -
-
-
Top-of-my-head, Maybe downres the scene then, for each sample point (pixel) run it through a high pass IIR. Then square that to get high frequency energy and feed that into a low pass IIR. When output for a pixel exceeds threshold, you could censor the area around it.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
I suspect TAA doesn’t keep enough information. Might be easier to do with the low res info from eye adaptation auto-exposure. Run a bandpass filter on it and see what comes out. (Assuming its well known frequencies anyway)
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.