Influence of bits per channel on how tools work and on the final output

  • 1
  • Question
  • Updated 4 years ago
  • Answered
Hello there!
I am trying to wrap my head around all this: what is the effect of using 8, 16 or 32 bits per channel on photoshop and how to use it smartly for a given context (output for print or web or else for example)

Questions:
1*What tools benefit from this change: influence on brushes, filters, gradient tool, etc? Is there any kind of datasheet where all the differences are listed?

2*Are there tools that become available or become disactivated on the different bits/channel modes (which could be on the same above datasheet)?

3*Do you win at using always the highest depth and converting it to the lowest afterward? Are there different methods and parameters to make this transformation? Since we work on screens that are in 8bits per channel, it is a bit hard to figure that out ^^

4*Are this modes working better with if you have a special screen (like the fancy color management ones that i dream of, which seem to have a 10 bit depth if i am not mistaken?)

I already found this link where people say what they think is best, but not the demonstration of why it is best.
http://www.dpreview.com/forums/post/1...
Photo of yoann.3d

yoann.3d

  • 38 Posts
  • 4 Reply Likes

Posted 4 years ago

  • 1
Photo of Chris Cox

Chris Cox

  • 20280 Posts
  • 823 Reply Likes
8 bit/channel is the most common usage. All tools and filters work in 8 bit/channel. This runs fastest because it is moving the least memory around and the math is relatively simple. Consumer cameras produce 8 bit/channel data.

16 bits/channel provides more detail, and won't show artifacts from editing quite as easily7 - but takes more memory and disk space, and operations run a bit slower because they have to move more memory and deal with more complex math. A few filters and tools will not work in 16 bit/channel. Professional digital cameras produce 12 to 16 bit/channel data.

32 bits/channel is High Dynamic Range (HDR) with floating point values - unless you know what that is and why you'd use it, you probably shouldn't be using it. Most filters and some tools will not work in 32 bit/channel. This uses the most memory and disk space, and tends to run slowest because it is moving much more data, and the math is slower.
Photo of yoann.3d

yoann.3d

  • 38 Posts
  • 4 Reply Likes
great answer chris, thank you for that :)

Are the brushes falloffs also influenced by the bits/channels?
Is there always a gain in working in 16 and converting in the end in 8?
Photo of Chris Cox

Chris Cox

  • 20280 Posts
  • 823 Reply Likes
The brush falloff is the same, just mathematically smoother because of increased precision.

The benefits depend a lot on the workflow. If you do a lot of compositing or adjustments, the starting in 16 and converting to 8 at the end will usually improve quality.

Some workflows (HDR, 3D render compositing, etc.) benefit from working in 32 bit/channel - partly from increased precision, partly because it is forced to be gamma 1.0 - and then converting down to the final precision at the end.
Photo of yoann.3d

yoann.3d

  • 38 Posts
  • 4 Reply Likes
fully understood, thank you infinitely!