Colour depth of GF1

I'm confused about bit depth

After a quick look around I couldn't find color depth specs on any cameras. I did this because it seemed to me that 24 bits was extremely high, not low. Then, I ran into this page:

http://www.luminous-landscape.com/tutorials/bit-depth.shtml

which made it all clear—we're really talking about 8 bit depth, but in RGB so we end up with 24 bit. And you're right, the better dSLRs are currently at 48 bit for RAW, and I'd guess that some of the studio backs (Leaf, Sinar, etc.) are above that.

What this doesn't explain to me is what happens in the camera; the better dSLRs (EOS 1, Nikon 3x) list their AD converters as 14 bit. Typically, good AD design specifies 2 bits in excess of the final bit rate (sometimes more) to allow for anomalies. Yet these cameras have 48 bit RAW files (at least in terms of their color depth) which I think implies 16 bit depth in each of the 3 channels. Can anyone explain this or send a link to a good site?

I can say that in operation, the GF1 files I work with (at 16 bits in Photoshop) seem to compare with my 5D shots in terms of color subtlety (but not noise or dynamic range). Then again, I don't think my monitor is more than 12 bits (36??) so there may be a difference there, and I can't see it.

 

Can you tell the difference when looking at the pic?

I can't so i don't even think about it.

 

That's a great question. So many talk about color and warmth etc. Has anyone played with settings to see which ones bring out the best in the GF1 - there are so many choices - almost too many - any how will they play out in Raw anyways.

I won a M9 and do believe that the images look a little more pleasing with the Leica (at a sign. higher cost of course). The Lumix GF1 is great and with the 20mm lens always in my pocket - yet... what would you all suggest to be the best settings (I know that there may be a million answers - but that's the fun of it..)

BTW - how do I automatically get email updates from this site to learn about new posts etc. - I don't seem to have been able to subscribe yet...)

 

Favorite Settings

Your right—I'm sure you'll get as many different answers as people that reply. :smile: I've settled (at least for now) with jpgs set at Dynamic color, with contrast, sharpness, and noise reduction all at -1. Then on AWB, I have that set to +2 towards green, and -1 towards amber. This looks about right to me. On the rare picture that I like, I end up working with the RAW that I capture simultaneously.

The RAW files are not affected by any of this as far as I can tell. (I did test the film settings and know they do not affect the RAW, but didn't check the AWB settings. These should not either, at least according to "RAW."

 

This is great - just what I wanted - I will try that - shoot JPEG and Raw simultaneously .... I'll try out your settings....

 

Most computer monitors are only capable of displaying 24 bit colour depth.

Wikipedia has some decent info on this : http://en.wikipedia.org/wiki/Color_depth#32-bit_color

There are some higher end graphics workstations / monitors that can do better, the only one I know of that's "consumer" off hand happens to be a laptop, the Lenovo Thinkpad W700 I believe.

So my interpretation is ... for a typical person on a computer monitor at home, 24bit will look the same as anything 24bit and higher like 36bit. So... raw images will look the same (with respect to colour range) as jpegs (ignoring compression losses).

Of course I could be all wrong...

 

I take it you mean won an auction rather than won it in a competition? Now that would have been sweeeeeet!!

 

Hi robixx, welcome to the forum and thanks for raising this interesting question.

I think the bit depth depends on the quality of Analogue to Digital conversion... not sure how significant the difference is between 8 bit per channel and 12 bit per channel - when you're converting to sRGB for instance.

Cheers

Brian

 

I wonder whether the greater bit depth is actually delivering increased colour range / quality though? does DxO labs measure this sort of thing?

This isn't something I've researched in any depth (no pun intended )

Cheers

Brian

 

Actually it is indirectly related to colour range. What you say is true in recovering details in what otherwise looks like black or white on jpegs, but the problem comes into play when you want to see the full proper range of shades.

As an example, if you take a colour chart with 1000 shades of red, and you take a picture of it. No matter what you do, you can not display those colours properly on an 8 bit monitor, nor can you store those colours in a JPEG. No amount of raw conversion will fix that, the best you can do is blotch together 2 of the colour samples.

So, if you had a 12 bit monitor and a 12 bit camera with 12 bit raw photos, yes you can see a greater colour range. If any one of the above items is 8 bit, you will see at max a 256 colour - range. Until everything turns to 12 bit, the only real usage for the 12 bits is to effectively change the exposure of certain parts of a picture in attempts to "recover lost detail". (ignoring the other usage which maybe in printed material where they are not restricted to the 256 colour limits). You've got to wonder what effect it has on a photo as a whole though, the ability to selectively change the exposure of individual portions of a picture...

 

Robixx is correct here and the rest just don't understand his point. No worries, nobody was born a genius. :smile:

Please read this article to understand why is 48-bit information useful. GF1 RAW files are not as edit-friendly, but you can't get everything for this price.

Alan, 14-bit AD means that for example bright zone in the photo will have 8192 levels (brighter parts of the picture "take" much more levels), while 12-bit would take only 2048. This brightest zone of the photo can take 50% of all levels in the photo. That tells you how the photo is TAKEN. However once you get the photo on your computer, the information is written differently. The same brightest part of the photo that was taken with 50% of all available levels now takes 25-30% of all levels. Instead of 8192 possible levels (2048 with 12bit AD) the SAME brightness range will use only around 70 levels in an 3 x 8-bit photo. That is of course a huge drop. If the information is however stored as 3 x 16-bit, maximum number of levels in that zone is over 17000. So no information is lost.

 

Maybe looking around could help us solving the problem...

48bit vs. 24bit means approx. double RAW file size.

Nikon D300 uses 14bit AD converter and 48bit RAW. The sensor size is 12.3MP. RAW file size is around 12-13MB, loss-less . GF1 RAW file size with same size of the sensor is what - 14MB? It looks like it has to be 48bit?

 

I was talking about RAW files, they use 48bit color info. As far as I know, there are 24bit and 48bit RAW files and 30/36/42/48 bit AD converters. I have never seen anything in between 24 and 48 (which doesn't mean it doesn't exist, but I have never seen it). If you open a photo in a 48bit editing program it will say either 8 or 16 bits per channel and there are no other values in between.

And even with AD converters there are 3x16=48bit cameras, for example Leica.

Ok, filesize does vary though, depending on the photograph. Canon does use lossless compression as far as I know and for GF1 file it could be tested in a 48bit editing program.

That doesn't say much. 12 bit is AD converter, but it doesn't say how is info stored in RAW file. Useful information nevertheless, AD converter is 12 bit, the same as Nikon D200 (Nikon D300 already 14bit) and Canon 30D (Canon 40D already 14 bit)