Why Working In 16bit RAW Colour Makes Sense
Yes, I agree with your assessment, which is why I have at some length explained exactly what is the real worth of working in 16bit.
As you rightly say, simple copying of slides is one thing, then taking it further on in terms of colour and tonal correction, combined with retouching, is quite another journey altogether. And although I recommend VueScan for its ability to get the best possible quality 16bit RAW scan out any one individual model or make of scanner, it's colour correction tools are not exactly a one size fits all solution for everyone. Achieving top quality digital copies from analog film stock is both a science and a black art. The science, well anyone can learn that, but the black art part is a hard earned perishable skill that not everybody aspires to. So the choice of software really does help in the development of the latter.
As by way of example, my current digital camera work involves the extensive use of seven different software packages, and frankly, where do I start in explaining it all? Well, I can't really, as it has taken me 25 years of experience (a lot of it totally mind numbing!) to get to where I'm at now. So it is only natural for a graphics pundit to recommend to others what they use themselves, and out of respect for the intelligence of the reader, I'm not going to do that.
The real secret is to get the very best possible 16bit RAW archival quality file out of your scanner, and then afterwards you can spend as much time as you like figuring out what to do with it.
Apart from drum scanners, most film and desktop scanners work with an array of RGB light sensors combined with a mirror and lens system arranged in a line, that moves backwards and forwards over the film area powered by a motorised worm drive in a stepping motion - pixel by pixel. What VueScan (or any other good scanning software) will do is either move the sensor array forward pixel by pixel, and at each pixel point make multiple samples (just like CD player does when multi sampling) then summing them together in one file - or - making multiple passes, and merging all of the files together as one. The former is better, as the worm drive only moves once. The latter is perhaps not as good because the worm drive has to operate in such a way that perfect pixel registration is not always possible, and will tend to give a slightly softer scan.
But why multi-scan? The answer is:- if the scanner light source is too bright, it floods the light wells of the sensor, and bleeds into the surrounding pixels - not good. Too dark, and the result is a noisy, grainy scan. So there is a compromise at work here. In theory, all scanners should operate in 16bit - they don't, and the reason is that the 1st bit is too noisy in the shadow areas, and the 16th bit contains so much spurious highlight information, that it becomes almost impossible to either process efficiently or compress sufficiently as JPEGs or TIFs. What the scanner software engine does is top and tail the RAW scan down to 14bits of information operating within a 16bit colour space. A good scanner will by default operate in 14bit, a cheap but good film scanner will operate in 12bit, and a cheap but not so good desktop scanner will try and get away with just 10bits. Multi-pass scanning facilitates the summing of random noise into coherent information that considerably improves shadow detail and highlight tonality - it just takes a bit longer, and is to be recommended when working with cheap or low specification scanners.
These discrepancies in bit depth will have a decisive effect upon the amount of post production colour and tonal correction possible before the image begins to fall apart due to digital artefacts - often making an image look crunchy and posterised. A really excellent 16bit RAW scan can have the living daylights mashed out of it, and the end result will still look good.
So to summarise:- get the very best RAW scans possible from your scanner first time once only (with the least amount of physical damage to the film!) and then afterwards, edit away to your hearts content as many times as you like.
ToS
Yes, I agree with your assessment, which is why I have at some length explained exactly what is the real worth of working in 16bit.
As you rightly say, simple copying of slides is one thing, then taking it further on in terms of colour and tonal correction, combined with retouching, is quite another journey altogether. And although I recommend VueScan for its ability to get the best possible quality 16bit RAW scan out any one individual model or make of scanner, it's colour correction tools are not exactly a one size fits all solution for everyone. Achieving top quality digital copies from analog film stock is both a science and a black art. The science, well anyone can learn that, but the black art part is a hard earned perishable skill that not everybody aspires to. So the choice of software really does help in the development of the latter.
As by way of example, my current digital camera work involves the extensive use of seven different software packages, and frankly, where do I start in explaining it all? Well, I can't really, as it has taken me 25 years of experience (a lot of it totally mind numbing!) to get to where I'm at now. So it is only natural for a graphics pundit to recommend to others what they use themselves, and out of respect for the intelligence of the reader, I'm not going to do that.
The real secret is to get the very best possible 16bit RAW archival quality file out of your scanner, and then afterwards you can spend as much time as you like figuring out what to do with it.
Apart from drum scanners, most film and desktop scanners work with an array of RGB light sensors combined with a mirror and lens system arranged in a line, that moves backwards and forwards over the film area powered by a motorised worm drive in a stepping motion - pixel by pixel. What VueScan (or any other good scanning software) will do is either move the sensor array forward pixel by pixel, and at each pixel point make multiple samples (just like CD player does when multi sampling) then summing them together in one file - or - making multiple passes, and merging all of the files together as one. The former is better, as the worm drive only moves once. The latter is perhaps not as good because the worm drive has to operate in such a way that perfect pixel registration is not always possible, and will tend to give a slightly softer scan.
But why multi-scan? The answer is:- if the scanner light source is too bright, it floods the light wells of the sensor, and bleeds into the surrounding pixels - not good. Too dark, and the result is a noisy, grainy scan. So there is a compromise at work here. In theory, all scanners should operate in 16bit - they don't, and the reason is that the 1st bit is too noisy in the shadow areas, and the 16th bit contains so much spurious highlight information, that it becomes almost impossible to either process efficiently or compress sufficiently as JPEGs or TIFs. What the scanner software engine does is top and tail the RAW scan down to 14bits of information operating within a 16bit colour space. A good scanner will by default operate in 14bit, a cheap but good film scanner will operate in 12bit, and a cheap but not so good desktop scanner will try and get away with just 10bits. Multi-pass scanning facilitates the summing of random noise into coherent information that considerably improves shadow detail and highlight tonality - it just takes a bit longer, and is to be recommended when working with cheap or low specification scanners.
These discrepancies in bit depth will have a decisive effect upon the amount of post production colour and tonal correction possible before the image begins to fall apart due to digital artefacts - often making an image look crunchy and posterised. A really excellent 16bit RAW scan can have the living daylights mashed out of it, and the end result will still look good.
So to summarise:- get the very best RAW scans possible from your scanner first time once only (with the least amount of physical damage to the film!) and then afterwards, edit away to your hearts content as many times as you like.
ToS
Last edited:
Colour Space Theory In Post Production Work After Scanning In RAW
When scanning in 16bit RAW, you have the option to choose a colour space in which to save the file. This is not exactly true, by definition a RAW file is exactly that, a file filled purely with RAW data. Except that a RAW file, as a container, actually has a file header filled with what is known as 'metadata'. This is descriptive information that tells software further along the post production workflow what to do with the RAW file when opened, and in this case, RAW scanning data is consequently shoehorned into a colourspace - of which there are quite a few.
So let's stick to the two which are important to us, namely 'sRGB' and 'Adobe RGB-98'.
sRGB as a colour space, was invented by Microsoft way back in the beginning of the World Wide Web, and describes all the possible colours to be found on a web page as an 8bit hexadecimal number encoded in HTML, and is what lossy compressible JPEG's are made of. Most inkjet printers are RGB devices which convert RGB images (whatever their colour space) to CMYK on the fly using an sRGB colour conversion engine, for the principal reason that all the colours of CMYK lie inside the sRGB colour space, and this is the common ground upon which most manufacturers work together with as a colour space or language. It has a lot in common with MIDI code as found in music technology.
Adobe RGB-98 was invented in (yes!) 1998, and describes most of the colours to be found in most analog colour films, and is significantly larger in terms of colour gamut than sRGB. And as such is often best for use when opening up a RAW file to convert to Adobe RGB-98 for post production work. In 16bit A-RGB'98 you can do extreme corrections with excellent results. After that, a final conversion to a small, heavily compressed 8bit JPEG, and it will still look reasonable.
There are lots of other colour spaces out there, but for now, these two will do.
And there is a lot of image manipulation software out there too! Most are a pig to learn, all have their own positive attributes and failings, and often take a long time to learn. But fear not, I worked with 8bit files in the same version of Photoshop for 15 years with great results before finally making an upgrade in my workflow. If you are going to get serious about retouching, then get a Wacom stylus and pressure tablet - very gestural and a lot more intuitive than a mouse, with a far less risk of RSI as well.
You don't have to work with Photoshop. An amateur version of Projects HDR can yield remarkable results with RAW files. You can assign a colour space, work with auto generated pseudo 16bit bracketed exposures, do extensive colour and tonal correction work, and the retouching brushes are a doddle.
Anyways, hope what I have written over the last few weeks will be of further use to you all.
Good luck! ToS
When scanning in 16bit RAW, you have the option to choose a colour space in which to save the file. This is not exactly true, by definition a RAW file is exactly that, a file filled purely with RAW data. Except that a RAW file, as a container, actually has a file header filled with what is known as 'metadata'. This is descriptive information that tells software further along the post production workflow what to do with the RAW file when opened, and in this case, RAW scanning data is consequently shoehorned into a colourspace - of which there are quite a few.
So let's stick to the two which are important to us, namely 'sRGB' and 'Adobe RGB-98'.
sRGB as a colour space, was invented by Microsoft way back in the beginning of the World Wide Web, and describes all the possible colours to be found on a web page as an 8bit hexadecimal number encoded in HTML, and is what lossy compressible JPEG's are made of. Most inkjet printers are RGB devices which convert RGB images (whatever their colour space) to CMYK on the fly using an sRGB colour conversion engine, for the principal reason that all the colours of CMYK lie inside the sRGB colour space, and this is the common ground upon which most manufacturers work together with as a colour space or language. It has a lot in common with MIDI code as found in music technology.
Adobe RGB-98 was invented in (yes!) 1998, and describes most of the colours to be found in most analog colour films, and is significantly larger in terms of colour gamut than sRGB. And as such is often best for use when opening up a RAW file to convert to Adobe RGB-98 for post production work. In 16bit A-RGB'98 you can do extreme corrections with excellent results. After that, a final conversion to a small, heavily compressed 8bit JPEG, and it will still look reasonable.
There are lots of other colour spaces out there, but for now, these two will do.
And there is a lot of image manipulation software out there too! Most are a pig to learn, all have their own positive attributes and failings, and often take a long time to learn. But fear not, I worked with 8bit files in the same version of Photoshop for 15 years with great results before finally making an upgrade in my workflow. If you are going to get serious about retouching, then get a Wacom stylus and pressure tablet - very gestural and a lot more intuitive than a mouse, with a far less risk of RSI as well.
You don't have to work with Photoshop. An amateur version of Projects HDR can yield remarkable results with RAW files. You can assign a colour space, work with auto generated pseudo 16bit bracketed exposures, do extensive colour and tonal correction work, and the retouching brushes are a doddle.
Anyways, hope what I have written over the last few weeks will be of further use to you all.
Good luck! ToS
You have probably already checked out the following link? The guy upsets a lot of shutterbugs but I found the site a treasure trove when switching from analog to digital.
https://kenrockwell.com/tech/raw.htm
https://kenrockwell.com/tech/raw.htm
I like Ken Rockwell, says a lot of sense ......... and very good at setting up straw man arguments to be knocked down by his brand of good sense - beware the guru with a wife and kids to support. The article you refer to was written back in the Stone Age nearly 10 years ago, and the world had got a lot faster since then. If in doubt, shoot JPEG + RAW and be done with it. Or import into Adobe Lightroomand export as a future proof composite Adobe DNG file. Or should I say Adobe's idea of a future proof file format. Who knows?
I have not quite got around to deciding on a monthly payment plan yet.
I did use his suggested settings on used Nikon D40 and the results were excellent, even close up.
I had the images printed on vinyl, to be used on an A frame advertising board, 2 per side, so the images were a good size.
I have just looked up the saved images and the L Basic JPEGs were only 2.6 Meg, OK, there was a fair bit of blue sky.
I found the following to be a very interesting read.
https://kenrockwell.com/tech/mpmyth.htm
I did use his suggested settings on used Nikon D40 and the results were excellent, even close up.
I had the images printed on vinyl, to be used on an A frame advertising board, 2 per side, so the images were a good size.
I have just looked up the saved images and the L Basic JPEGs were only 2.6 Meg, OK, there was a fair bit of blue sky.
I found the following to be a very interesting read.
https://kenrockwell.com/tech/mpmyth.htm
Pixel Count Musings
Rockwell is right, after 6 megapixels anything more is for most purposes merely something to brag about.
I worked with two 7 megapixel Nikon D70cameras for nearly ten years, and only changed over to working with two 16 megapixel Fuji XE2s about three years ago. This was after months of research to find the right instrument to work with in my current creative practice.
To be honest, there are times when the extra pixel count can be a real pain, in that it all needs vastly more computing power and storage space. That means an awful amount of time sitting in front of a computer to get the kind of results I am after. But when generating panoramic pictures up to 3 gigapixels in size, I have to accept this bottleneck until I can afford an infinitely more powerful computer. The reality is I will be printing out pictures in tiled sections anything up to multiplex cinema screen size, and it all takes time. I have still got the Nikon's as a fallback, and with an adaptor the Fuji's take all my Nikon lenses.
The reason I shoot in RAW is that it enables me to change my mind about what it was I actually saw! In olden times, I would stick to one emulsion so as to learn how to see colour and light as per the film stock. Of course, that has now all changed, and the less experienced often get confused as to when to stop in the post production process. I like the Nikon colour - very accurate, like the old Kodak film stocks, but I also like the funky Fuji virtual film stocks which I can dial in and out like crazy - if I want to.
I am of the belief that all digital cameras are good, it is just a question of using them properly. Some of my best work has been done on a 15 years old 2 megapixel point and shoot Canon A40. It is noisy but beautifully noisy, and the colour at ISO 1600 is so sweet and soft. I have deliberately shot pictures at high JPEG compression to get exaggerated artefacts because of the way it enhances certain pictures.
A Nikon D40 is plenty enough to work with. If you can learn to shoot pictures just by manipulating the camera controls alone without having to tweak afterwards, then it will take you a long way. Two pieces of advice, shoot in RAW + JPEG and read the free manual. This may sound crazy, but I take the manuals everywhere I go! Sometimes for fun, I will sit down and read a manual just to see if I really know a camera, and it can be quite humbling. Work with JPEG and use RAW for when you might change your mind about what you remember.
Basically, what Rockwell is saying about pixel count and working with JPEG is more or less right. We use our eyes to look at pictures, and our ability to recognise pleasing colour and tonal harmonies is built into us. Photography is a bit like audio, in that it is best not to get too hung up about kit. If you are getting good results with the Nikon software and Irfan View, then stick with it, as quite often an upgrade can lead to too many choices which can really get in the way of your creativity. Less is often more!
ToS
Rockwell is right, after 6 megapixels anything more is for most purposes merely something to brag about.
I worked with two 7 megapixel Nikon D70cameras for nearly ten years, and only changed over to working with two 16 megapixel Fuji XE2s about three years ago. This was after months of research to find the right instrument to work with in my current creative practice.
To be honest, there are times when the extra pixel count can be a real pain, in that it all needs vastly more computing power and storage space. That means an awful amount of time sitting in front of a computer to get the kind of results I am after. But when generating panoramic pictures up to 3 gigapixels in size, I have to accept this bottleneck until I can afford an infinitely more powerful computer. The reality is I will be printing out pictures in tiled sections anything up to multiplex cinema screen size, and it all takes time. I have still got the Nikon's as a fallback, and with an adaptor the Fuji's take all my Nikon lenses.
The reason I shoot in RAW is that it enables me to change my mind about what it was I actually saw! In olden times, I would stick to one emulsion so as to learn how to see colour and light as per the film stock. Of course, that has now all changed, and the less experienced often get confused as to when to stop in the post production process. I like the Nikon colour - very accurate, like the old Kodak film stocks, but I also like the funky Fuji virtual film stocks which I can dial in and out like crazy - if I want to.
I am of the belief that all digital cameras are good, it is just a question of using them properly. Some of my best work has been done on a 15 years old 2 megapixel point and shoot Canon A40. It is noisy but beautifully noisy, and the colour at ISO 1600 is so sweet and soft. I have deliberately shot pictures at high JPEG compression to get exaggerated artefacts because of the way it enhances certain pictures.
A Nikon D40 is plenty enough to work with. If you can learn to shoot pictures just by manipulating the camera controls alone without having to tweak afterwards, then it will take you a long way. Two pieces of advice, shoot in RAW + JPEG and read the free manual. This may sound crazy, but I take the manuals everywhere I go! Sometimes for fun, I will sit down and read a manual just to see if I really know a camera, and it can be quite humbling. Work with JPEG and use RAW for when you might change your mind about what you remember.
Basically, what Rockwell is saying about pixel count and working with JPEG is more or less right. We use our eyes to look at pictures, and our ability to recognise pleasing colour and tonal harmonies is built into us. Photography is a bit like audio, in that it is best not to get too hung up about kit. If you are getting good results with the Nikon software and Irfan View, then stick with it, as quite often an upgrade can lead to too many choices which can really get in the way of your creativity. Less is often more!
ToS
Thanks for the reply.
We all look after our eyes and get them checked regularly and wear corrective lenses when necessary. All things being equal, normal people have no problem in finding a compressed image more than acceptable.
I find it strange that to most forum members here, compression is a dirty word when it comes to audio, as is modern equipment.
I updated my D40 to a D3200 a couple of years ago but dialed back the 24 meg.
Last year I purchased an "as new" Canon Powershot SD600 for $5 in a thrift store and is the camera I seem reach for most.
We all look after our eyes and get them checked regularly and wear corrective lenses when necessary. All things being equal, normal people have no problem in finding a compressed image more than acceptable.
I find it strange that to most forum members here, compression is a dirty word when it comes to audio, as is modern equipment.
I updated my D40 to a D3200 a couple of years ago but dialed back the 24 meg.
Last year I purchased an "as new" Canon Powershot SD600 for $5 in a thrift store and is the camera I seem reach for most.
You are welcome.
I think it is because the eye can discern around about 16 million colours, and is easily fooled by illusions. Vision is not static but 'liquid'. Sound is a different matter. From what I understand, the ear can discern something like 180,000 different actual sounds, and is therefore much more critical in its assessment of audio quality. Good cinema is actually a sound medium with visuals tacked on to keep the minds eye happy. Did I really say that?
ToS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.