RAW/TIFF Printing

[tclune]

I've been reading this thread & do not know what a 'JPEG artifact' is?

thanx ,
Diane

Structures can be introduced into an image due to the lossy compression routine used. These created structures are called "artifacts," The one for JPEG that everybody knows on sight is small squares that appear if you compress an image too much. The reason these squares appear is that JPEG divides your image into 8x8 pixel regions, and applies the compression to those. If the compression is too much, you start seeing the boundaries of the 8x8 pixel units.

Other JPEG artifacts include "posterization," in which a gradually-changing region of solid color breaks up into discrete regions -- it looks like a contour map of the original region. This is caused by over-aggressive quantization (lumping similar colors together into a single value), and can happen with even very mildly compressed JPEG when dealing with the right kind of image. You will often see it in photos that include large expanses of clear blue sky. There is a related but distinct phenomenon, in which colors start looking washed out. Your nice bright photo starts looking like a water color. This is often the first thing that people notice when JPEG is pushed too far. FWIW

[Glenn NK]

Thanks Tom, I did not know that (about the 8 x 8 pixel square).

Glenn

[AdelaideImage]

I support this point "You might be missing something in terms of image quality when converting to jpg but you won't be missing anything at all in the printed image." Check detail tiff printing and conversion info and template.

[Andrew1]

I agree with Tom's explanation of data manipulation and the inherent errors it creates. It doesn't matter if it's light or sound, digitization means rounding off analog information which means differences (errors)(aka noise) in the relayed information. If your own application can handle that, fine. We all live with those shortcomings almost every day. Doesn't mean it isn't there. Some we pay no mind to and others bother the hell out of us. (well, me anyway) As for printing I will go with some advice given to me by a printer when I converted over to digital photography. 8X10 or less use jpeg because it just doesn't matter to any (at that time) normally used photographic printer. Anything bigger use the more detailed TIFF. You may or may not be able to see the difference on any one particular photo or size but then again, you may. If you can both handle the translation then why not provide the most information you can? It's no big deal, and for some, does make a difference.

[Manfred M]

Another old thread comes to life...

Andrew - when we print, even on a very high end photo printer can only do so much. The printer manufacturers are usually pretty secretive about actual perforrmance and HP does claim to do around 1.2 distinct colours with their technlogy in their marketing literature; http://h10025.www1.hp.com/ewfrf/wc/d...002_USEN&lc=en

I would assume the competition would be in a similar range, although I've also found claims and calculations that indicate that high-end photo printers can only reproduce the equivalent of 6-bits per channel (256,000 distinct colours). After all there is only so much one can do with when mixing various CYMK variants. My Epson 3880 uses two flavours of black (photo and non-photo paper), two grays (light black and light, light black in Epson speak), two magentas (regular and light), two cyans (regular and light) and yellow; i.e. a maximum of 8 cartridges at a time with varying droplet size.

The bottom line is that a professional photo colour printer can produce less than 1/10th of even the sRGB colour space (if you truly believe the marketing literature and ignore that the gamuts are not identical). Throwing out most of the colour data is going to happen when you feed the image to the printer driver; regardless if you feed in a jpg, TIFF, psd or any other file format.

[Mark von Kanel]

after reading all this my head hurts! so from it all i think my workflow which is as followings is ok

I shoot, in RAW sRGB & convert to DNG on import. i then edit in DNG sRGB, i then tend to print from this stage in lightroom, i am not actually sure if lightroom converts the image in any way before it hits my printer , ill have to find out!

how would shooting in adobe RGB help? or would that be pointless for my printing at home (epson 3880).

there has been lots of discussion about DNG format, Tom do you have any input on this? i like it because of the embedded edit info rather than a side car file.

If having printing done commertially i send tiff images but have noticed that the results are often varied! so a lab that accepts a tiff file doesnt mean a good lab! thats why i decided to do my own prints (well it alo means i get to buy more gadgets)

[Manfred M]

Mark - if you are shooting RAW, colour space settings are meaningless. This camera setting only applies to jpg / tiff formats that your camera outputs, not the camera RAW. I do shot a lot of jpeg files, so I do have my camera set to Abobe RGB for these; I've never used TIFF output on the D800.

You can assign in a colour space in Photoshop or Lightroom (Lightroom has sRGB, Adobe RGB and ProPhoto; listed from narrowest to widest gamut). For printing I let the driver software in Photoshop worry about the output conversions to my Epson 3880 and have never had any issues here. I don't tend to print with Lightroom as I it only lets me use the printer drives The only issue I've found with this workflow is that it doesn't seem to work for B&W prints; the Photoshop drivers seem to result in a slightly green cast. Here I chose to use the Epson rather than Photoshop profiles.

I tend to stick to RAW versus DNG for most things because DxO Optics Pro can't read DNG. On the other hand I will use DNG when profiling with my Color Passport because the X-Rite software won't read RAW...

[Andrew1]

To my thinking, the extra data available applicable to this discussion doesn't mean there are extra colours created in a printer. It does however provide for a more accurate rendition/selection within the colours available. A better choice.

The extra data is more relevant when we discuss protocols and how the analog information is digitized and again when translated into colour. Errors which occur normally in all quantizing logs manifest themselves as artifacts we normally see as colour shifts or noise. Will I see the difference from a high end printer? Possibly, but some of that smoothness is the result of dithering, not information. It would be akin to the garbage on my tv screen when it's being provided a 720 signal instead of 1080.

Similarly another example, while it may certainly be good enough for relevant applications, the translation of Adobe DNG cannot equal the originating data of Nikon or Canon. Data is transposed, changed and possibly lost. Is it important? It is to many. A recognized member and instructor at NAPP states he won't use DNG until his camera manufacturer uses it for their own raw software. He feels the translation has problems and "Just isn't there yet." He doesn't offer specifics and I don't care. He's an expert whose teachings and informed opinions I trust and can learn from.

[Manfred M]

Remco - a number of Nikon users have found (and made statements on this site) that Adobe RAW converters do NOT always reproduce the colours accurately and there is a preferred direction to Capture NX2 or ViewNX2. If the initial mapping from the RAW file is incorrect, then the child processes (i.e. .DNG conversion) could be assumed to be suspect too.

I was first introduced to this issue by a photography instructor at the local community college where I've taken a few courses. He was clearly able to demonstrate that Capture NX2 produced different results than ACR. I have found the same thing in testing.

All that being said; I regularly use three different RAW converters; and all have their advantages and disadvantages:

1. View NX2 - best colours but a fairly basic interface and usability. It does not have the lens correction tools that other RAW converters have. Price is right; it comes bundled with the camera.

2. DxO Optics Pro 8 - colours are okay and the lens correction is fantastic. Downside, expensive and slow. I use it whenever I plan to make a large print.

3. ACR (and Lightroom too) - Nicely integrated into Creative Suite and has a half-decent lens correction package. Worst out of the box colour accuracy of the bunch. Okay if I do a custom profile.

[Colin Southern]

HP does claim to do around 1.2 distinct colours with their technlogy

Yep - sounds about right. Epson rules!

[revi]

Remco - a number of Nikon users have found (and made statements on this site) that Adobe RAW converters do NOT always reproduce the colours accurately and there is a preferred direction to Capture NX2 or ViewNX2. If the initial mapping from the RAW file is incorrect, then the child processes (i.e. .DNG conversion) could be assumed to be suspect too.

I was first introduced to this issue by a photography instructor at the local community college where I've taken a few courses. He was clearly able to demonstrate that Capture NX2 produced different results than ACR. I have found the same thing in testing.

All that being said; I regularly use three different RAW converters; and all have their advantages and disadvantages:

1. View NX2 - best colours but a fairly basic interface and usability. It does not have the lens correction tools that other RAW converters have. Price is right; it comes bundled with the camera.

2. DxO Optics Pro 8 - colours are okay and the lens correction is fantastic. Downside, expensive and slow. I use it whenever I plan to make a large print.

3. ACR (and Lightroom too) - Nicely integrated into Creative Suite and has a half-decent lens correction package. Worst out of the box colour accuracy of the bunch. Okay if I do a custom profile.

You were fast That post you reply to was up for about 5 min, and I was asking about what was lost in translation to DNG.

My point was NOT that the raw converters from Adobe would give the same results as the Nikon converters. That is quite possible and easy to understand when you know that the manufacturers are rather closed about what they store exactly in their metadata and what they do in their converters (and that goes for all manufacturers).

I was specifically asking about what was lost in the translation to DNG. If an image from a Nikon camera gives the same results when converted from DNG or from the original RAW, where's the translation loss (using the same converter, ofc, the only variable should be the format translation)?

I always get suspicious when someone makes a bold statement and doesn't give facts to back it up (and the more so if the person is a teacher at a high level)...

[revi]

Yep - sounds about right. Epson rules!

The difference a small 'M' makes

[Mark von Kanel]

Mark - if you are shooting RAW, colour space settings are meaningless. This camera setting only applies to jpg

DOH! yes i knew this deep down honest! and i rarely shoot JPEG

You can assign in a colour space in Photoshop or Lightroom (Lightroom has sRGB, Adobe RGB and ProPhoto; listed from narrowest to widest gamut). For printing I let the driver software in Photoshop worry about the output conversions to my Epson 3880 and have never had any issues here. I don't tend to print with Lightroom as I it only lets me use the printer drives

Im not sure i understand you here Manfred by drives do you mean ICC Profiles? if so then maybe you havnt put your profiles in the correct folder because ive always been able to select any of the profiles that i have, in LR. If not can you please clarify what you mean?

The only issue I've found with this workflow is that it doesn't seem to work for B&W prints; the Photoshop drivers seem to result in a slightly green cast. Here I chose to use the Epson rather than Photoshop profiles.

Again the confusion over drivers/profiles i assume they are the same and eagerly await clarification! i dont get any cast using custom profiles for my paper/ink/printer combo printing from light room. i dont print from PS, is there a reason for me to try/do this rather than use LR?

I tend to stick to RAW versus DNG for most things because DxO Optics Pro can't read DNG. On the other hand I will use DNG when profiling with my Color Passport because the X-Rite software won't read RAW...

LOL manufactures love to make life difficult! ill stick with DNG as adobe does what i need it too and im certain that DxO will come into line in time under the inevitable pressure of the huge giant that is adobe.....

[Mark von Kanel]

The bottom line is that a professional photo colour printer can produce less than 1/10th of even the sRGB colour space (if you truly believe the marketing literature and ignore that the gamuts are not identical). Throwing out most of the colour data is going to happen when you feed the image to the printer driver; regardless if you feed in a jpg, TIFF, psd or any other file format.

So what is the point of using Adobe RGB when shooting (in JPEG ) and processing ?

[revi]

I must say I find it hard to believe that a good printer covers less than 1/10th of the sRGB colour space (leaving aside that YMCK is a different system)
What can happen is that the total volume covered in colour space is comparable to sRGB (or bigger), but that within that space the distance between
neighbouring colours that can be displayed is larger (leading to less available colours).

The point of using Adobe RGB is not to have more colours available (that's limited by the number of bits you use), but to have colours that are not
contained within the space covered by sRGB, like the very saturated colours you can find in flowers.

Otoh, if your printer cannot print colours outside the sRGB space, you'll have to be careful when editing in a larger space, and it might not be worth
the bother.

[Manfred M]

You were fast That post you reply to was up for about 5 min, and I was asking about what was lost in translation to DNG.

My point was NOT that the raw converters from Adobe would give the same results as the Nikon converters. That is quite possible and easy to understand when you know that the manufacturers are rather closed about what they store exactly in their metadata and what they do in their converters (and that goes for all manufacturers).

I was specifically asking about what was lost in the translation to DNG. If an image from a Nikon camera gives the same results when converted from DNG or from the original RAW, where's the translation loss (using the same converter, ofc, the only variable should be the format translation)?

I always get suspicious when someone makes a bold statement and doesn't give facts to back it up (and the more so if the person is a teacher at a high level)...

My initial reaction was that Adobe would be using a common engine to do the conversions, so I would expect both to be identical, so I did a bit of a test.

I took a shot and used ACR to bring the same image across as both a RAW and DNG file; both as 16-bit files. The DNG file is 45.9MB and the NEF file is 51.0MB; so we are obviously looking at some level of (lossless??) compression because the DNG has the metadata and an embedded thumbnail, so there is more data that has to be stored. I always assumed that the DNG was essentially just a wrapper for the RAW data. I used identical settings in ACR when importing the file out of the camera.

Next part of the test was to bring both files into Photoshop (CS6), place one on top of the other and apply a blending mode (subtract or difference) to see if there a difference between the two files. I would have expected a pure black image, but instead I got this (this is a resized, jpeg, but it is still a fairly good representation of what I see on my screen):





I played around with the nudge function to see if it is a registration issue, but no, Photoshop appears to have placed the two images on top of each other. The colour profile that I assigned is AdobeRGB. I personally think this might be a bug in Photoshop, because when I do a selection and to a straight copy and paste and do a manual alignment, I can generate a pure black image.

When I first saw this image, my initial reaction was that the DNG conversion does result in some quality loss. Now I am not so sure.

[Manfred M]

DOH! yes i knew this deep down honest! and i rarely shoot JPEG

Im not sure i understand you here Manfred by drives do you mean ICC Profiles? if so then maybe you havnt put your profiles in the correct folder because ive always been able to select any of the profiles that i have, in LR. If not can you please clarify what you mean?

You can tell Photoshop to embed the colour space you use in the file; in CS6 I have a choice of sRGB, AdobeRBG and ProPhoto. Lightroom does the same, buy does not support the ProPhoto option

Again the confusion over drivers/profiles i assume they are the same and eagerly await clarification! i dont get any cast using custom profiles for my paper/ink/printer combo printing from light room. i dont print from PS, is there a reason for me to try/do this rather than use LR?

Again, a Photoshop, rather than Lightroom variable. When I print out of Photoshop I can select whether I let Photoshop or the printer manage colours. In Lightroom, I don’t have this choice and it uses the printer’s colour management algorithms

LOL manufactures love to make life difficult! ill stick with DNG as adobe does what i need it too and im certain that DxO will come into line in time under the inevitable pressure of the huge giant that is adobe.....

There are a number of other Giants in the mix; Canon, Nikon, Sony, etc. If the file format were truly “open” as Adobe claims, there might be more incentive to use it.

Read the licence agreement: http://www.adobe.com/products/photos...gspecification

If I were one of these companies legal departments, I would advise my clients to stay as far away from the DNG format as I can. The licence conditions are totally tilted in Adobe’s favour.

[Colin Southern]

If I were one of these companies legal departments, I would advise my clients to stay as far away from the DNG format as I can. The licence conditions are totally tilted in Adobe’s favour.

Which parts of the agreement do you find tilted or unreasonable?

[Colin Southern]

My initial reaction was that Adobe would be using a common engine to do the conversions, so I would expect both to be identical, so I did a bit of a test.

I took a shot and used ACR to bring the same image across as both a RAW and DNG file; both as 16-bit files. The DNG file is 45.9MB and the NEF file is 51.0MB; so we are obviously looking at some level of (lossless??) compression because the DNG has the metadata and an embedded thumbnail, so there is more data that has to be stored. I always assumed that the DNG was essentially just a wrapper for the RAW data. I used identical settings in ACR when importing the file out of the camera.

Next part of the test was to bring both files into Photoshop (CS6), place one on top of the other and apply a blending mode (subtract or difference) to see if there a difference between the two files. I would have expected a pure black image, but instead I got this (this is a resized, jpeg, but it is still a fairly good representation of what I see on my screen):





I played around with the nudge function to see if it is a registration issue, but no, Photoshop appears to have placed the two images on top of each other. The colour profile that I assigned is AdobeRGB. I personally think this might be a bug in Photoshop, because when I do a selection and to a straight copy and paste and do a manual alignment, I can generate a pure black image.

When I first saw this image, my initial reaction was that the DNG conversion does result in some quality loss. Now I am not so sure.

I just did the same with this image as a CR2 and DNG ...



and as expected, got this ...



Happy to send you the source files if it helps with testing.

[Manfred M]

I just did the same with this image as a CR2 and DNG ...



and as expected, got this ...



Happy to send you the source files if it helps with testing.

You got what I had expected to get.

What I am getting is what doesn't make any sense.