16 bit vs 8 bit

[monkey44]

I'm shooting RAW with a 7D2 and 5DM3 ... when I do the conversion, it asks me which one to convert into ... 16 bit Tiff or 8 bit Tiff ---

Can anyone explain which one and what is the difference - I mean, I can see 8 - 16 is twice the size (or is it depth) ... I'm not very tech smart, so please explain this in lay terms, basically -- thanks ... BD

Why would I chose one over the other?

 

[Marsu42]

I'm shooting RAW with a 7D2 and 5DM3 ... when I do the conversion, it asks me which one to convert into ... 16 bit Tiff or 8 bit Tiff ---

Imho your best bet is not to convert raw at all until the last possible moment. For this, use ACR (Adobe Camera Raw) which is Lightroom and part of Photoshop. The advantage of this workflow is that it's non-destructive and you can revert anything at any moment. Last not least, you might profit from future developments in the raw conversion process (esp. denoising).

Can anyone explain which one and what is the difference - I mean, I can see 8 - 16 is twice the size (or is it depth) ... I'm not very tech smart, so please explain this in lay terms, basically -- thanks ... BD

The current Canon resolution is 14bits, i.e. with 16bits you're wasting two bits - doesn't matter if you compress the tiff file though. The largest difference you'll see is in gradients, for example an evening sky going from somewhat blue-ish to something red-ish. Using only 8 bits will introduce nasty color steps and make your precious dslr look like a mobile phone.

Note that 16bit files need more memory when editing and more hard disk space, but in the year 2015 of our lord (other year number for followers of other lords) this shouldn't matter unless you're stacking a lot of layers in Photoshop.

 

[mackguyver]

Note that 16bit files need more memory when editing and more hard disk space, but in the year 2015 of our lord (other year number for followers of other lords) this shouldn't matter unless you're stacking a lot of layers in Photoshop.

Good advice, Marsu, and I'm still working to kill all 64GB of my RAM. I was up to 59GB the other day when I was processing photos in DxO. stitching a pano on PS, editing in Premiere, correcting a windy/shaky video in After Effects, and working on a logo in Illustrator. My poor VRAM (3GB) is actually what began to limit me...but doing the above just a few years ago was just a dream ;D

 

[Marsu42]

My poor VRAM (3GB) is actually what began to limit me...

What do you know, this is nearly the amount of total RAM I have ... and the power of two cores . It's not that bad when I'm not in a hurry, and it forces me to do other things like cleaning my home or pet my well, pets, when Lightroom is rendering files (It'll switch off because it overheats if I run a 2nd app ).

The only hard limit I've ever experienced is rendering panoramas, the stitching software flat out refuses to render with panos of 20+ shots no matter the virtual hard disk swapping space and no matter if I'm prepared to let it work overnight. That's why I know using 8-bit files processing is really not worth the tradeoff even on limited hardware.

 

[monkey44]

Well, just to follow up that comment above -- I always do as much processing as possible in DPP (I know, some don't like it, but I do like it) and then convert to .tiff and polish it up if necessary in Photoshop.

I always save the RAW file as the original file - without saving any changes. I like to have that "untouched" RAW file right out of the camera. Then, after processing and tweaking, save one original .tiff file with the RAW file in an original folder.

Then, I pull a copy up whenever I need an image and work that copy - never the original file, which remains in the originals folder all its life.

 

[Maui5150]

I'm shooting RAW with a 7D2 and 5DM3 ... when I do the conversion, it asks me which one to convert into ... 16 bit Tiff or 8 bit Tiff ---

Imho your best bet is not to convert raw at all until the last possible moment. For this, use ACR (Adobe Camera Raw) which is Lightroom and part of Photoshop. The advantage of this workflow is that it's non-destructive and you can revert anything at any moment. Last not least, you might profit from future developments in the raw conversion process (esp. denoising).

Can anyone explain which one and what is the difference - I mean, I can see 8 - 16 is twice the size (or is it depth) ... I'm not very tech smart, so please explain this in lay terms, basically -- thanks ... BD

The current Canon resolution is 14bits, i.e. with 16bits you're wasting two bits - doesn't matter if you compress the tiff file though. The largest difference you'll see is in gradients, for example an evening sky going from somewhat blue-ish to something red-ish. Using only 8 bits will introduce nasty color steps and make your precious dslr look like a mobile phone.

Note that 16bit files need more memory when editing and more hard disk space, but in the year 2015 of our lord (other year number for followers of other lords) this shouldn't matter unless you're stacking a lot of layers in Photoshop.

I disagree with how you look at it.

For me, if I am shooting 14-bit and my editing platform handles 16-bit, chances are quite high that every color my camera shot is represented in the editing program

If I downsample to 8-bit, then 6-bits of colors need to be mapped to a different color.

I just don't think I am throwing away 2-bits of color, rather I have 2-bits of color more to assure I am as accurate as I can be (i.e. better to have more room)

 

[Marsu42]

Well, just to follow up that comment above -- I always do as much processing as possible in DPP (I know, some don't like it, but I do like it) and then convert to .tiff and polish it up if necessary in Photoshop.

It's not only about like or not, but once your library increases you need to decide for some management system - and for cataloging, tagging, sorting and bulk operations Lightroom blows everything else I know out of the water and then some.

I always save the RAW file as the original file - without saving any changes. I like to have that "untouched" RAW file right out of the camera. Then, after processing and tweaking, save one original .tiff file with the RAW file in an original folder.

It might not matter to most, but this essentially quadruples the data storage - ~25mb raw + ~100mb tiff or psd. With my limited hardware, I like that the ACR workflow just adds a small .xmp sidecar file and I only duplicate data on export - and then in a downsize resolution, 8bit jpeg.

I just don't think I am throwing away 2-bits of color, rather I have 2-bits of color more to assure I am as accurate as I can be (i.e. better to have more room)

I didn't want to argue against 16bit, just note that in the future there will be 16bit straight out of the camera - and even now, with Magic Lantern's dual_iso or hdr tonemapping you can use the full 16bits. Actually, in the latter case you need to go floating-point tiff or exr which further increases storage and memory requirements.

 

[jrista]

I'm shooting RAW with a 7D2 and 5DM3 ... when I do the conversion, it asks me which one to convert into ... 16 bit Tiff or 8 bit Tiff ---

Imho your best bet is not to convert raw at all until the last possible moment. For this, use ACR (Adobe Camera Raw) which is Lightroom and part of Photoshop. The advantage of this workflow is that it's non-destructive and you can revert anything at any moment. Last not least, you might profit from future developments in the raw conversion process (esp. denoising).

Can anyone explain which one and what is the difference - I mean, I can see 8 - 16 is twice the size (or is it depth) ... I'm not very tech smart, so please explain this in lay terms, basically -- thanks ... BD

The current Canon resolution is 14bits, i.e. with 16bits you're wasting two bits - doesn't matter if you compress the tiff file though. The largest difference you'll see is in gradients, for example an evening sky going from somewhat blue-ish to something red-ish. Using only 8 bits will introduce nasty color steps and make your precious dslr look like a mobile phone.

Note that 16bit files need more memory when editing and more hard disk space, but in the year 2015 of our lord (other year number for followers of other lords) this shouldn't matter unless you're stacking a lot of layers in Photoshop.

To the OP, I wouldn't worry about the "wasting" of two bits. With 8-bit, your obliterating a lot of tonality and color fidelity in your images, and IMO that is the real waste. Storage space is dirt cheap. Really. I remember the days when ONE MEGABYTE hit a dollar-per-meg. Today, we can have gigabytes-per-dollar.


I routinely save my astronomy images in 64-bit IEEE float. Sometimes it isn't about the storage space, it's about the value of precision for everything that happens inbetween storing it once and storing it again: the processing. With 8 bit data, you have more limited numeric space within which to store the results of various processing algorithms. That means you will ultimately start accumulating an error, which with more and more edits, results in undesired artifacts appearing. The error rate for processing with 16-bit data is far lower. It is even lower with 32-bit integer data (which you can do with TIFF), and by the time you get to 32-bit float (i.e. full precision floating point HDR TIFF), for most normal photography you don't have anything to worry about.


I use 64-bit float for my astrophotography because I often have to perform iterative processes to the data in order to extract the detail I want revealed. I can run 100, 200, 500, sometimes even 1000 iterations of an algorithm on my data (and I usually run many algorithms to reduce noise, improve SNR, stretch, enhance, etc.), so keeping the data in as high a precision format as possible significantly limits the chance for that residual error from revealing itself as artifacts in the image.

 

[monkey44]

Thanks for all the replies -- it now seems to make sense to save ONLY the original RAW files.

Once I process it the first time (for a client, or myself) -- it's not often we need to return and grab it again, unless to fulfill another print order - or send electronic copy to a client ... So, just saving the RAW file will use one file space in the storage bin. Altho, I like to have a copy of the processed files as well, just for reference. I've always believed .jpeg degrades over time and over 'times opened' ... maybe that's not true anymore, and saving a processed .jpeg will save file space too.

 

[mackguyver]

Thanks for all the replies -- it now seems to make sense to save ONLY the original RAW files.

Once I process it the first time (for a client, or myself) -- it's not often we need to return and grab it again, unless to fulfill another print order - or send electronic copy to a client ... So, just saving the RAW file will use one file space in the storage bin. Altho, I like to have a copy of the processed files as well, just for reference. I've always believed .jpeg degrades over time and over 'times opened' ... maybe that's not true anymore, and saving a processed .jpeg will save file space too.

JPEGs don't degrade when you open them, only when you process and then re-save them. Personally, I save the RAW file and a processed TIFF or layered PSD file. Sometimes I do some pretty tricky post-processing and when I've misplaced the processed file, it can be a real PITA to try to re-process the shot to look the same for a print order. I also save the XML sidecar, but I've had issues with that in the past when I user upgrade applications, at least in terms of preserving the adjustments.

The only time I use JPEGs are for my website. I create minimum compression JPEG but it is deleted as soon as it is uploaded (using PhotoMechanic).

 

[Marsu42]

JPEGs don't degrade when you open them

Which is unfortunate because it de-values the shots. If they'd wear down like a vinyl record or a paper print, there's simply more to it than just digital data floating around the net.