Monochrome camera vs converting from color?

[Tim Lookingbill]

There was no adjustments applied. All as shot.

I'm talking about the before and after histograms for the yellow and magenta flowers. What did you do to get the after? There's two versions of the same image. What did you do to get the second one?

BTW how do you uninstall FastRawViewer from Mac OS 10.6.8. I just tossed the app from the app folder including preferences (.plist) in my user Library and LlibRaw folder in Application Support. Is there more files to uninstall?

[Iliah]

I'm talking about the before and after histograms for the yellow and magenta flowers. What did you do to get the after? There's two versions of the same image. What did you do to get the second one?

There is no before and after. One is embedded/external JPEG as the camera recorded it, the other is raw as the camera recorded it.

how do you uninstall FastRawViewer

As any other application, drag to the Trash and empty the Trash.

[petermfiore]

“There are two ways to be fooled. One is to believe what isn't true; the other is to refuse to believe what is true.” -Søren Kierkegaard

Art is not about truth...Art is a designed illusion.

Peter

[digitaldog]

Art is not about truth...Art is a designed illusion.

Peter

Agreed. Much of this discussion has nothing do do with art.

[Slobodan Blagojevic]

What do you think about this, Slobodan?...

I agree 

[Slobodan Blagojevic]

... In mentioning the Ansel Adams quote, I was thinking more about darkroom work than using filters in the field.  The fiddling with sliders and days at the computer seemed analogous to everything Adams did with his knowledge of photographic chemistry, dodging and burning, and masking (not to mention his knowledge of whatever I've left out here).

I agree, Jeffrey, and I got what you meant the first time. That is why I said if that's one's approach too, then Monochrom is not for them. I admit that my next sentence was logically incoherent with the former, so I crossed it now in the original post.

[Doug Gray]

I've just finished experimenting with Epson's ABW mode on my 9800 with a Win 10 system and have the following results.

1. As long as R==G==B, it makes no difference what colorspace the source image is in. For example, if the original image is in ProPhoto RGB, converting it to sRGB or anything else produces exactly the same prints except for extremely subtle stepping in 8 bit mode. Even there, it's a tiny effect as the L* changes at each step are only about 0.4 dE76.  I can't see them but if I photograph the print and bump up the contrast enough the tiny dE steps can be brought out. But selecting dither (see comment #4) eliminates even that.

2. The tone curve is significantly smoother than either driving directly to device RGB or through the highest quality profile in color mode. The reason appears to be use of K, LK, and LKK only inks (absent injecting a color tone) while the color mode will add in the M, LM, Y, C, and LC. even on a neutral tone curve.

3. The a* and b* values over the tone curve are also very smooth and don't exhibit the marked perturbations that occur printing in color mode. Likely cause is same as in #1.

4. Higher effective bit resolution with 8 bit drivers can be had by selecting the option to dither when converting to 8 bits in the Photoshop Settings main dialog. When this is selected gradients in 16 bit tiff files will be rendered smoothly. This happens automatically behind the scene even though the user isn't directly converting to 8 bits.

[Jim Kasson]

I've just finished experimenting with Epson's ABW mode on my 9800 with a Win 10 system and have the following results.

1. As long as R==G==B, it makes no difference what colorspace the source image is in. For example, if the original image is in ProPhoto RGB, converting it to sRGB or anything else produces exactly the same prints except for extremely subtle stepping in 8 bit mode. Even there, it's a tiny effect as the L* changes at each step are only about 0.4 dE76.  I can't see them but if I photograph the print and bump up the contrast enough the tiny dE steps can be brought out. But selecting dither (see comment #4) eliminates even that.

2. The tone curve is significantly smoother than either driving directly to device RGB or through the highest quality profile in color mode. The reason appears to be use of K, LK, and LKK only inks (absent injecting a color tone) while the color mode will add in the M, LM, Y, C, and LC. even on a neutral tone curve.

3. The a* and b* values over the tone curve are also very smooth and don't exhibit the marked perturbations that occur printing in color mode. Likely cause is same as in #1.

4. Higher effective bit resolution with 8 bit drivers can be had by selecting the option to dither when converting to 8 bits in the Photoshop Settings main dialog. When this is selected gradients in 16 bit tiff files will be rendered smoothly. This happens automatically behind the scene even though the user isn't directly converting to 8 bits.

Doug, I agree with everything you say, except about the absence of non-K process inks in ABW mode. Unless something has changed since I put 3880 prints under a (literal) microscope, there is some use of those inks in that mode, even with no tint applied.

http://blog.kasson.com/technical/450/

compare to color mode:

http://blog.kasson.com/technical/inkjet-printing-on-epson-part-2/

Jim

[Alan Goldhammer]

Jim is correct about the use of color inks in the Epson ABW mode.  However, a lot less is used than in the color mode though I don't know if anyone has quantified this.  You also know that this is the case as Epson have a tone wheel that can be used to create sepia prints.  I did some work with Mark at Aardenburg printing out a variety of B/W test patterns using various settings of the tone wheel with my 3880 and Hahnmuhle Photo Rag Ultra Smooth.  The results are up on the Aardenburg website.  You can also see the difference on the review of the Epson 3800 here:  http://www.outbackphoto.com/printinginsights/pi045/essay.html#20070123

[Doug Gray]

Doug, I agree with everything you say, except about the absence of non-K process inks in ABW mode. Unless something has changed since I put 3880 prints under a (literal) microscope, there is some use of those inks in that mode, even with no tint applied.

http://blog.kasson.com/technical/450/

compare to color mode:

http://blog.kasson.com/technical/inkjet-printing-on-epson-part-2/

Jim

Nice work Jim. Quite different and more precise from what I recall using a 30x inspection-scope. I was specifically looking at the region where the b* anomaly was occurring ( b* from -1 to -3 with L* increasing from 88 to 92) along  R=G=B 240 region. I did not look at abrupt, large changes such as the horizontal/vertical stripes you were investigating.

In color mode the density of MCY was very high while in ABW all I saw were the K's but I didn't investigate other regions, beyond short looks.

I don't have a scope with your resolution let alone camera attachment but I do have a macro lens that I've used to examine patterning. That might work here.

Additional:
I've macro photo'd some patches around RGB=240, greatly increased contrast and boosted saturation even more. In spite of some camera motion which I need to work on , the dots are now quite visible with the Y and LC, LM  easily visible from the boosts. There are colored dots in the ABW patches v color mode though at considerably reduced frequency. Not sure why I didn't see them previously. Perhaps the limits of the 30x scope and yellowish light. More to follow.

[Doug Gray]

Jim is correct about the use of color inks in the Epson ABW mode.  However, a lot less is used than in the color mode though I don't know if anyone has quantified this.  You also know that this is the case as Epson have a tone wheel that can be used to create sepia prints.  I did some work with Mark at Aardenburg printing out a variety of B/W test patterns using various settings of the tone wheel with my 3880 and Hahnmuhle Photo Rag Ultra Smooth.  The results are up on the Aardenburg website.  You can also see the difference on the review of the Epson 3800 here:  http://www.outbackphoto.com/printinginsights/pi045/essay.html#20070123

Thanks Alan,

Looking at the plots, the 3800 appears to have the same breakpoints at low and high L*. The steps are rather large so I can't say precisely but they are certainly consistent. Also, the large shift in b* around L*=90, which is occurring is my main motivation for looking at the ABW mode which does not exhibit this. This large shift isn't fully overcome by the limited resolution of color profile LUTs.

I'm running a set of tests from RGB:224 to 255 in steps of 0.5 in these different modes to get a bit more info in the region of interest. May also take some macro shots to get an idea of what the distribution frequency differences in the inks are.

[Deardorff]

On seeing color and shooting B&W.

I do have a number of friends with major color blindness who shoot B&W film. Color is a mystery to them and color gradation/balance not something they work with. A monochrome body would work well for them and a Fuji X-Pro2M would be most welcome. (that after I asked five of them)

As for how some of us see the world. With B&W film in the camera my mindset is way different from Chrome film. I look at things differently when shooting B&W. That further parsed if I am shooting for Pt/Pd prints or silver darkroom prints. Tonal relations, contrast, sharpness and light all play a part. I still find it difficult to think in B&W when shooting color. Probably just me but there is a difference. I know my best B&W work is when I am shooting B&W rather than color and converting. I shoot less, have a much higher incidence of keepers and find my 'vision' works a bit better without the mental distraction of "color or B&W".

Still hoping for an X-pro2M so I can take advantage of the Fuji glass, rangefinder style feel and have an experience closer to film as I work.

[Rob C]

On seeing color and shooting B&W.

I do have a number of friends with major color blindness who shoot B&W film. Color is a mystery to them and color gradation/balance not something they work with. A monochrome body would work well for them and a Fuji X-Pro2M would be most welcome. (that after I asked five of them)

As for how some of us see the world. With B&W film in the camera my mindset is way different from Chrome film. I look at things differently when shooting B&W. That further parsed if I am shooting for Pt/Pd prints or silver darkroom prints. Tonal relations, contrast, sharpness and light all play a part. I still find it difficult to think in B&W when shooting color. Probably just me but there is a difference. I know my best B&W work is when I am shooting B&W rather than color and converting. I shoot less, have a much higher incidence of keepers and find my 'vision' works a bit better without the mental distraction of "color or B&W".

Still hoping for an X-pro2M so I can take advantage of the Fuji glass, rangefinder style feel and have an experience closer to film as I work.

Now, if that camera were to be made, would it be priced higher or lower than its colour original? Leica had its own version of that pricing situation.

Rob

[Doug Gray]

I've attached macro photos of "neutral" patches having a L* of 87 on Costco Glossy (WP=L* 95). These are 3mm squares magnified. Contrast and saturation have been boosted for easier identification of the dot colors otherwise the dot colors are quite hard to discern. Note that the actual patch colors differ by only just over 2 dE with the predominant shift towards magenta for the one in Color mode. The percentage of LLK dots is quite high in the ABW patch with lower amounts of LC and LM and Y in comparison to the ABW mode patch.

I've also included graphs showing L*, a* and b* for Color and ABW modes for RGB steps from 224 to 255 in steps of .5 as measured with an Isis in M2 (uV Cut) mode.  The tone curves differ between ABW and Color with ABW RGB values of 230 and  Color RGB values of 242 producing L* of 87.

Note how the a*, and especially the b* values remain quite close to those of the paper white in the ABW mode while the a* and especially, b*, vary considerably. Since this occurs over a fairly short range of RGB values ICC profiles can only partially compensate for these perturbations.

[digitaldog]

You're not helping.

I'll attempt to help you, like Iliah, by suggesting you study exposure on the actual data: raw. Maybe a few of the 1000 you say you've captured. Hence my question about your awareness or lack thereof, of a raw histogram. Start here:

https://www.rawdigger.com


There's even a discount for the holidays!
Seems you're a tad confused between clipping, exposure and how/where that gets applied in raw versus JPEG (as evidence by your text: I'm talking about the before and after histograms for the yellow and magenta flowers. What did you do to get the after?).
Get a copy of RawDigger, attempt to use it properly, then you'll have a tool to perhaps understand how to expose your raw data! AND see how vastly different that is from a JPEG Histogram of the same capture!

[Doug Gray]

I'm referring to your demo image of the magenta flowers shot in overcast or shaded light in a concrete bird bath fixture you've used to explain optimal exposure. Why should flowers clip in low light? Can you explain how that happens on a sensor when the bird bath and surrounding greenery looks normally exposed.

Even with RAW with no clipping, sRGB, in comparison to say Adobe RGB, is easily clipped on the red channel and to a lesser degree, on the blue channel.  This occurs even though the chromaticities (xy cords) of the red and blue are the same.  It's due to the sRGB green channel being relatively unsaturated. Because of this the highly saturated sRGB red channel, even at 255, has  relatively low luminance because 255 on all channels must produce the white point and the red is much further from the white point than the green. Thus the red channel on sRGB can easily clip and is much more prone to it than Adobe RGB.

[digitaldog]

A picture is worth 1000's words (and in the old days, dollars)


sRGB from raw (ACR workflow options below image) and Histogram versus ProPhoto RGB and Histogram:

[Tim Lookingbill]

A picture is worth 1000's words (and in the old days, dollars)


sRGB from raw (ACR workflow options below image) and Histogram versus ProPhoto RGB and Histogram:

Yeah, I also get data clipping in the histogram when I switch from ProPhotoRGB to sRGB in ACR with some flat swaths of color like intense greens, yellows and oranges that don't have fine detail as in your example image do slightly dull. That's old news.

Again you're not helping, Andrew, by pointing out the obvious and you keep deflecting from my original point about detail not seen in a flower in the Raw sensor data that has to be wrangled with extreme draw down point curves in not only the midrange but also the highlights depending on the color of the flower no matter the overall luminance in the scene.

You can't (at least I can't) calibrate for optimum exposure for that kind of sensor behavior.

[digitaldog]

You can't (at least I can't) calibrate for optimum exposure for that kind of sensor behavior.

Begging the question again, have you ever tried viewing a raw Histogram when attempting to expose anything?

[Tim Lookingbill]

Even with RAW with no clipping, sRGB, in comparison to say Adobe RGB, is easily clipped on the red channel and to a lesser degree, on the blue channel.  This occurs even though the chromaticities (xy cords) of the red and blue are the same.  It's due to the sRGB green channel being relatively unsaturated. Because of this the highly saturated sRGB red channel, even at 255, has  relatively low luminance because 255 on all channels must produce the white point and the red is much further from the white point than the green. Thus the red channel on sRGB can easily clip and is much more prone to it than Adobe RGB.

That doesn't answer why I'm capturing flower detail that has to be edited to turn flat blobs of color into what I saw at the scene and I just guess at the "optimum" exposure I think will capture that detail.

Most flower images posted online from hobbyist photographers with $3000 and up camera gear don't suffer from sRGB gamut squeezing but just plain lack of ability and/or patience to edit their images by shooting Raw. And even when they do, they have no clue what flowers should look like after post processing.

Why can't camera manufacturers after years of sensor technology improvements fix this behavior capturing intensely colored objects and preserve fine detail?