‘Breaking Pixels’ | High Bit Files & Why They Matter

Insights & Thoughts May 12th 2014 12:06 PM 5 Comments

So you’ve heard photographers insist on shooting in RAW versus JPEG because the files are unprocessed, uncompressed, and thus contain just that much more information than their merely mortal counterparts. Still, you may not know what that really means, or you’ve dabbled with processing a RAW image and seen a little how it can pull more details from shadows and highlights etc, but nothing major. Here’s an idea of how it all works, if  12, 14, 16, 24 bit files actually do better than 8.

[RELATED: RAW VS JPEG (JPG) – THE ULTIMATE VISUAL GUIDE]

Bit? Bit-depth?

Very quickly, each pixel in any image has its tonal value and color stored within it. Since we’re dealing with digital files aka computer files, understand computers store things in code of zeroes and ones. Bit depth is simply a reference to the number of digits the tonal info is stored in. Example, if you had a bit depth of 2, there would be two digits for use and corresponding 4 values. 00, 01,10,11. So basically four different colors. A bit depth of 8 (JPEG standard) is 2 to the 8th power, which is 256 values. That’s a lot more tones per channel. Consider a RAW file which has 4096 per channel.

 I’m sure from this, though you may not care, or care to care, you can see how more bits will carry more info, and can deduce that working in those files will result in less loss. Even if you ‘lose’ the same amount of data, losing it from 4000, is better than losing it from 256. So my advice would be to edit in RAW as much as you can, then go into the rest of the editing when you can do no more in RAW.

If you want to see a difference by looking at your histogram, we can do that, which will give you a visualization of what’s going on. I’ve opened up two images (sorry, didn’t have JPEGs of the RAW file for better comparison), one is RAW, and one is JPEG. I’ve given each some light manipulation and take a look at the histograms. The 8-bit JPEG has crazy spikes in it even after slight tonal adjustments in curves, and the RAW (which actually had more adjustment done), has none of that. Those spikes are lost info – and in this case, lost tone value. Broken pixels, if you will.

8-Bit JPEG as shot
breaking-pixels-histogram-highbit-bit-raw-jpeg-1

8-Bit JPEG With Slight Curves (see spikes in historgam)
breaking-pixels-histogram-highbit-bit-raw-jpeg-2

RAW File With Tonal & Color Adjustments (no spikes)
breaking-pixels-histogram-highbit-bit-raw-jpeg-3

[REWIND: Why You Should Multiply Aperture By Crop Factor When Comparing Lenses]

But Hang On…

I’ve just done the typical argument for high bit files, and shooting in RAW. Now, I’m still going to advocate shooting in RAW because I like knowing the info is there to play with if I want to, and frankly, I like the more subdued, somewhat granular look an uncompressed RAW file has. Yet, there’s a big hairy ‘but’ coming. I’ve tried to explain the virtues of this info to new photographers (who judged the final prints) and non-photographers (who didn’t care and judged the final ‘prints’), and most didn’t notice when one was a high-bit file or not. When I showed them the histogram they understood it, but at a small size, or even larger sizes, they just couldn’t tell.

There were a few types of shots where the differences were more pronounced and they understood some value of what was being presented. Those types were usually shots with high contrast and highlights where any manipulation would show a sort of gradient. Otherwise, however, especially when photos were actually printed (not professionally), the difference can be negligible. Oh, and in case your’e wondering, in regards to CMYK workflow and printingmost printers are going to work in 8-bit. So the choice is yours and I’d suggest, if you’re not going large, and are low on computer processing power and storagemaybe JPEGs are the way to go for casual shooting.

[RELATED: CMYK VS. RGB AND WHY YOU SHOULD CARE]

Advertisement

About

Kishore is a photographer and writer based in Miami, though he can often be found at dog parks, and airports in London and Toronto. With a passion for beauty and aviation photography his work is all at once focused and eclectic. He is also a tremendous fan of flossing and the happiest guy around when the company’s good.

5 Comments

  1. Romain Menke

    nothing in this post is really wrong but there is still a distinction not made.
    16bit vs 8bit has an effect on how photoshop or any other post processing app calculates color values and it defines how much tones you can capture from your raw file.

    But it is still limited by the color profile.
    The color profile (standard photoshop uses sRGB or adobe 1998) also defines what gets stored and what gets compressed. This is very noticeable in highly saturated and bright colours. These lose detail and color information. Nothing 16bit can do about it, because there are no values in sRGB or adobe 1998 to represent these sensor values. ProPhoto is a profile built for this, to capture everything a sensor ever could record.

    So maybe do a post about color profiles and why they are so important ;)

  2. Mark Koolen

    You miss the biggest point in editing, banding.
    In 8 bit there is a lot of banding when you edit photo’s with a even colored back drop.
    16 bit solves the problem.

    • Romain Menke

      Those spikes in the histogram are banding if you have even backgrounds. Quick fix for 8-bit files: add noise

    • Maarten de Boer

      ROMAIN MENKE: “Quick fix for 8-bit files: add noise”. Or just don’t push em too far, anyway what you’re getting at is dithering ;)

    • Romain Menke

      I didn’t know that dithering would work. Makes sense since it also gives a noise pattern.

      I haven’t had banding in a long time (switched to ProPhoto 16-bit when I still worked with a 5D mark 1) and it didn’t really matter when there was some since I always add DXO grain.

      Beside that the developers behind photoshop and stuff like ProPhoto color profiles make incredible tools that allow us to push very far. It’s just up to us to know how

Leave a reply

Advertisement