[stevebrot]

If I understood (finally) correctly, JPEG was designed to web display.

JPEG was given to us complements of NASA and predates the Internet. That being said, it is very useful for publishing compressed continuous tone content to the Web and has been a mainstay for photos since the very beginning. JPEG has limitations, but is more than adequate for most of the devices and monitors on which the content is being viewed.

RAW is just another example of doublespeak, it creates an illusion of complete control for the camera user, but the reality is not so clear.

Perhaps not doublespeak, but there is a lot of truth in your statement. A few mornings ago, I was testing a new lens under light overcast on some flowers in my back yard. I was shooting DNG and the rear LCD showed some fairly pleasing results. When I imported into Lightroom, the colors were not quite right. Obviously, auto WB was having trouble and conventional wisdom is that since I was shooting DNG (no WB applied to the actual capture data) this should be easily managed in PP. Yes, Lightroom will apply the camera's best guess from the file metadata; no, making a suitable correction would not be easy. Even without having to work around imposed JPEG rendering, the hard facts regarding the sensor response* to each of the elements in the first scene (lavender-colored flowers with yellow and grass-green in the background) meant that a simple shift in color temp (WB) and tint would fall far short regardless of gamut width, calibrations, and sophistication of tools.

In short, more than one pigment is at work in the flower petals and even with a color-managed workflow at all steps, embedded profiles, and due diligence by me, some viewers of the flowers in the final JPEG will see lavender, some magenta, and some cornflower blue, depending on the JPEG rendering engine used by the viewer software (usually a browser) and the display being used.

I did my best, but there are physical limits imposed by the hardware that I cannot overcome. Perhaps Foveon or maybe Ektar 100 might do better. This is just one area where control is limited. The other is managing contrast with limited bit-depth in the shadows.

What color flowers do you see?


Tritelia laxa (Ithuriel's Spear), Pentax K-3, Lester A. Dine 105/2.8 Macro

FWIW, there is a second set of photos of another flower that I have never successfully gotten decent color rendering of using digital capture. I am thinking that adding a frame with gray card or colorchecker (for a custom import profile) might be a better approach.


Steve

* I lay the blame on narrow band-pass filters used by the Bayer array and/or limitations of Lightroom's image processor and import profiles. I do have similar issues with other tools, however.

[beholder3]

Interesting thing, even when exporting dng to tiff 16bits, I could never recover shadow as well as with dng. So I guess with TIFF the dynamic range is lost. I noticed however no banding with TIFF 16bits as opposed to 8bits JPEG.

TIFFs already get the Gammacurve baked in unless you actually create ones without gamma correction. That might be part of the reason.

[stevebrot]

TIFFs already get the Gammacurve baked in unless you actually create ones without gamma correction. That might be part of the reason.

Being an actual image format, everything is baked!


Steve

[MD Optofonik]

RAW is the general term for the "native" file format of a digital camera, JPEG is a set of standardized algorithms for compressing visual data and the sensor component in a digital camera has to process the data read off the sensor in order to output a file in any format, RAW or otherwise. Conversion between file formats can be done in the camera or after downloading image files from the camera, but any data not stored in the "native" file format can't be included in another format. Converting a RAW file to a JPEG file will almost always contain less data because most of the compression algorithms combine neighboring pixels into a single value and if the compressed file is expanded again, differences between neighboring pixels in the original file are lost. In order to quickly move data from the sensor to the camera's other processors and the camera's memory, minimal processing is applied to the RAW file, but without some processing, the analog signals produced by the sensor itself are meaningless.

Without JPEG algorithms, the Internet would probably still look like text-based bulletin boards and digital cameras would never have been marketed to consumers. The problem is that once an image file is converted to a JPEG format, so much information is lost that any subsequent processing is going to contain visible artifacts that can spoil the image. DNG files are standardized containers for the image data and metadata supplied by the camera, so that processing software can use the same code for files from different cameras, but it is the embedded programming of the camera that determines what data is saved in the original image file, regardless of the output file format.

RAW is just another example of doublespeak, it creates an illusion of complete control for the camera user, but the reality is not so clear.

I have always understood "RAW" to be synonymous with the term "raw data" in any other form.

[Apet-Sure]

What color flowers do you see?

Quite lovely flowers on my new IPS monitor. It may not be exactly what your eyes and brain saw, but I still think it's a good shot.


I have some flowers in my yard, one is red and the other yellow, that I just can't get a good capture of. I've used a K10D and now a K-5iis, and various lenses. Those particular reds and yellows are always blown out, regardless of how drastic the exposure compensation is. I've used the 'bright' and 'natural' settings on the K-5, same result. Sunny days, overcast days, same result. It just beats me. Admittedly I mostly shoot jpeg sRGB, and may have different results with AdobeRGB or RAW. Those two flowers are the only subjects I've ever had that trouble with. Other reddish or yellowish flowers come out fine. Aaahhhrggg! And yes, I do know I should be shooting RAW regularly. I'm currently test driving darktable occasionally. Have also tried RawTherapee briefly. On my budget, the free ones are the only option.

[RGlasel]

I have always understood "RAW" to be synonymous with the term "raw data" in any other form.

At the most basic level, information has to conform to a known structure or pattern, otherwise it can't be distinguished from noise (in the Shannon-Weaver model) or chaos or the empty void, so I don't like the term "raw data" either. I suppose we could draw the line at the point where individual human interpretation modifies the received data, but that doesn't work so well with digital cameras. Translating excited electrons at photosites to a structured file of binary bits that can be processed by electronic devices, requires a lot of programming, which is to say a lot of processing, and that processing is programmed by someone other than the camera user. RAW is more like being lightly grilled, compared to breaded and deep-fried.

[stevebrot]

so I don't like the term "raw data" either.

I would prefer it be called something like "capture data".


Steve

[biz-engineer]

Being an actual image format, everything is baked!

Yes, everything is baked. It is baked and spread over a 16bits scale for the TIFF16 (without compression, color channel sub-sampling) instead of baked over an 8bits scale and compression of the JPEG.
Applying curves without any global change of level works better on TIFF16 vs JPEG, but a global change of exposure or shadow pull on the TIFF16 isn't much better than doing the same thing on a JPEG.
I guess that's why Pentax did not bother to include a TIFF 16bits as optional image format in camera. I only see TIFF as an advantage when looking after the ultimate sharpness that JPEG compression may not have retained.

[stevebrot]

Applying curves without any global change of light level works between on TIFF16 vs JPEG, but a global change of exposure or significant shadow pull on the TIFF16 isn't much better than doing the same thing on a JPEG.

Yep...what the extra bit depth gets you is additional headroom and tonal options in the high values. The bottom several stops remain data-impoverished.


Steve

[kaseki]

I would choose cornflower blue as closest, running Firefox 67.0.3 under Linus Mint MATE 18.3 and viewed on a BenQ SW2700 operating sRGB color gamut. I'm unclear whether any color profile is actually operative, as the ColorHug I bought proved to be mis-calibrated and I didn't accept its results for embedding when DisplayCal provided a temporary example of what it would do to the image. I need to get back into this subject if I ever get time and acquire a proper spectrophotometer.

I think that to reproduce all colors as they would be seen by the human eye, one would need the camera focal plane array filter spectral functions x detector spectral responsivities to match the color responses of the human eye, and similarly, the monitor color emission spectra would have to do likewise. Color measurement and generation seems to have gone another way, probably for practical or realizability reasons. (See CIE 1931 color space - Wikipedia)

[MD Optofonik]

Welly, welly; this certainly veered off into interesting and enlightening territory beyond my initial query (which was answered quite well).

Thank you and, please, carry on.

[clackers]

My post processing technique is to first export the OOC JPEGs to nice big fat TIFF files using GIMP
]

You might as well leave that step out, GnD.

The damage has been done shooting in 8-bit JPG, you don't get any of that data back by a subsequent conversion to 16-bit TIFF.

[Rondec]

You might as well leave that step out, GnD.

The damage has been done shooting in 8-bit JPG, you don't get any of that data back by a subsequent conversion to 16-bit TIFF.

Right. It's like going to the store, buying a quart of milk and taking it home and pouring it into a gallon jug. The TIFF file can't have any extra data that the jpeg didn't have in it.

I will say that my biggest problem with out of camera jpegs is that if things like sharpness and saturation are set too high you get artifacts that are really hard to fix. Shooting with these sorts of settings on low gives you a file you can work with after the fact, but then you have to spend time post processing and you will have better results with the RAW file -- both better highlight and shadow recovery. And with storage as cheap as it is, it isn't costly to shoot RAW.

[beholder3]

The custom image mode "flat" is specifically there to help postprocessing from JPG.

I do prefer the in cam sharpening over LR sharpening, so the only two reasons for me to shoot raw are correcting WB and shadow lifting. Since I prefer shooting in natural sunlight the latter is the big factor.

[goatsNdonkey]

Right. It's like going to the store, buying a quart of milk and taking it home and pouring it into a gallon jug. The TIFF file can't have any extra data that the jpeg didn't have in it.

I will say that my biggest problem with out of camera jpegs is that if things like sharpness and saturation are set too high you get artifacts that are really hard to fix. Shooting with these sorts of settings on low gives you a file you can work with after the fact, but then you have to spend time post processing and you will have better results with the RAW file -- both better highlight and shadow recovery. And with storage as cheap as it is, it isn't costly to shoot RAW.

Perhaps neither of you know this, but were I to do my adjustments to a jpeg image, every time I saved it it would get littler and littler and lose progressively more data from the original jpeg version. That is the data loss I am avoiding by exporting the original jpeg to a large uncompressed file format. OF COURSE, I don't gain back any data lost in the in-camera jpeg creation. But the OP's question had to do with a camera that doesn't shoot a RAW image like I can, one that only shoots jpegs.