Hacker News Reader

Comment by magicalhippo | Hacker News Reader

I just skimmed through parts of the video as I'm about to head to bed, but at least the bits I listened to sounded more like arguing for why 24bit audio isn't necessary for playback, 16bit will do just fine.

Back in the days I made ray tracers and such, and going from an internal SDR representation to an internal HDR representation was a complete game changer, especially for multiple reflections. That was a decade or more before any consumer HDR monitors were released, so it was all tonemapped to SDR before displaying.

That said, I would really like to see his two monitors display something with really high dynamic range. From the stills I saw in the video, they all seemed quite limited.

Anyway, something to watch fully tomorrow, perhaps he addresses this.

View on HN

I think he argues more that if 24-bit audio would have been brought to the market like HDR it would have been advertised as being able to be louder. The new stereos would lack the volume control as 24-bit audio would encode absolute volume levels and older formats would play quietly from the new amazing hardware. Songs would be mastered to have more dynamic range changes so that the sound would pop over to loud volumes every now and then. This would be attributed to the higher bit depth of the signal format even though it would be just a different style of producing music. In order to play old songs loud in the new amazing stereo you would need to buy remastered version which would not simply play louder but would overall be quieter to emphasize the volume spikes added in remastering.

The difference being that you don't need more bits to make music louder or give it more dynamic range, but you do need HDR (and more bits) if you want immersive bright spots. It's not reasonable to dim your scene until it looks like charcoal and tell everyone to adjust their display to 25x brightness for your videos. You need a paradigm shift to where "white" is just a calibration level, not the maximum.

Fair point about the absolute scale. It was something that was introduced way later than when I was working on HDR.

I think it really should be named HDRA or something, similar to how we have dBA[1], to distinguish it from relative HDR aka "just" HDR.

[1]: https://www.engineeringtoolbox.com/decibel-d_59.html

For pretty much any graphics, you shouldn't be using a capped display space at all. You should be doing all of your math with 32 bit floats (or integers) in a linear color space (wrt photons or human color perception depending on what type of effect you're performing). You should only be converting to a display color space at the end.

Reply on news.ycombinator.com