Colour bit depth 8, 12, 14 or 16?

[nightpiper]

The question is simple. Do u think the extra bits help in reproducing better colours?

I thought i saw better colours from Nikon & Canon in 14 bits mode however, one thing still bugs me. Are we getting all those 14bits quality or were the bits being down-sampled to 8bits?

So far i've tested a few RAW converters (bibble, old nikon capture, canon DPP, Adobe ACR, phase one capture, Olympus Studio) in their respective file format not DNG, i noticed that the colour tweaking in RGB curves only shows from 0 - 255 levels, that means only 8bits per colour. I've tried converting RAW files into aRGB 16bit TIFF but still the RGB curves only has 255 levels. So all along we've been seeing & adjusting 8bit colours? :think: Cud this be the reason why we can't get really rich colours from films like Fuji Provia, Velvia & Kodak Ektachrome 25, 64? When i look at my old films, the colours is simply exceptional.

And no, i dun think the extra bits has anything to do with DR or noise floor. Its more in respect to the granularity of quantizing/digitizing the signal thats already been limited by the threshold noise floor & peak saturation.

What are ur thoughts? Any RAW converters on the market adjust in 16bits, ie, 65k RGB curve levels?

 

[wuyongzheng]

I think 12 and 14 bit don't have much difference except when your photos are underexposured and post-processed in PC later. 14 bit also helps in making HDR photos. If the photo is correctly exposured, the extra bit is just noise.

But on the other hand, storage is cheap, so no harm to store 14bit RAW. So my sudggestion is to store the highest depth possible if you want post-processing.

 

[shojibake]

0-255 is probably just a legacy thing. If not they'll have to change the UI everytime there's a photo with a different bit depth.

 

[nightpiper]

Dun think its a legacy thingy le. U can't possibly adjust 65k levels by juz using 256 levels, thats basically 'lumping' the 16 bits into becoming 8 bits (downsample).

On the other hand, u can adjust 8 bit levels in a 16 bit environment cos the MSB (most significant bits) can be juz 0s. There's no need to change any UI, it can be consistently stay at 65k levels, the software will take care of the 0s in the MSB. Its juz like 32bits apps running in 64bit OS environment, 64bits shud be able backwards compatible.

 

[shojibake]

You're talking about padding, it's a ui issue. That's legacy.

 

[nightpiper]

Care to explain in more details? How's that an UI issue? Legacy? Kind of lost here. No prob with the tech talk, think i can handle it, if not i'll ask my friend mr google

 

[shojibake]

it's just user interface. Anyhow, the company is too lazy to develop custom user interface components that can dynamically update it's resolution based on the bit depth.

 

[nightpiper]

Why wud they need to do that? Why can't they juz stick to a higher bit depth if they r lazy to update the interface? Hmm... :think: Really?

 

[calebk]

Why wud they need to do that? Why can't they juz stick to a higher bit depth if they r lazy to update the interface? Hmm... :think: Really?

If I'm not mistaken, Photoshop, for instance, is used by most people who would work in 8-bit colour bit depth if working on web design for instance. It is mostly us photographers who end up working with the expanded bit depth, such as 14-bit RAW.

Seeing as Photoshop is a very widely used tool, and photographers are only a small group of these users, they may just stick to 255,255,255.

I do believe when you shoot in RAW you are getting your full 14-bits, but when you export to JPG for web or print (unless you go to TIFF) you'd probably downsample to 8-bits.