Faithful Color Reproduction

BrWhatsit

Senior Member
Hello,

I've got a D40 that I have been using for about 7 years now. There are features I'd like to have from a more advanced camera, but that's not in the budget for the forseeable future, and this one serves for most things. But I'm having some trouble with color reproduction. I'm photographing hand-dyed yarns so that they can be sold online. Yarns with lots of different colors in them are fine; they require only minimal color correction in post-processing. But for yarns that are more monochromatic, I'm not getting accurate color. For instance, on a skein of yarn that is mostly blue with some purple areas, everything in the photo looks blue (what was purple looks dark blue) and it's not even a faithful representation of the actual blue. Similar results with a skein in the red-orange-yellow range; everything looks redder than it actually is. I've done a custom white balance and tried several different aperture settings. Am I missing something? Anyone else have this problem? How did you solve it?

Thank you very much,
Will
 

WayneF

Senior Member
A calibrated monitor is always a good thing, but what lighting source are you using? Custom white balance (from a known white or gray card) ought to do it, but fluorescent and CFL lights are not a continuous spectrum, and may never come out perfect in some cases. If so, try again with either sunlight or real incandescent lights or flash, to see if situation changes.
 

BrWhatsit

Senior Member
Thanks very much for your suggestions. I have tried various monitor calibration programs, which helped a little, some of the time. But the colors don't appear correctly even in the camera's display screen. I expect to have to do some minor tweaking in Photoshop, but I'm getting images that are way off, having to spend an unreasonable amount of time on image editing, and still not getting things to look right.

I'm using daylight CFL bulbs. I'll try shooting outside; I can at least learn if that works. Long-term, though, I'll need something that will give me consistent results regardless of the time of day or the weather. Could anybody recommend an artificial light source that gives results close to natural sunlight?

I'm shooting in jpeg, but I'm a little embarrassed to say I'm not really familiar with the picture control settings.
 

WayneF

Senior Member
Fluorescent lights are sort of a special problem, because they simply are not a continuous spectrum. The Sun is continuous, and incandescent lights are, and flash pretty much is - yet all of these have different types, and (except for the Sun), also age. Bottom line, we really never know what exact color our lights are. Saying, it's really not enough to simply set the camera to some one WB value. Instead, we have to correct the image to match the light we actually have. It is good to calibrate the monitor, but even if so, the image has to be corrected too.

The solution is to include a known neutral white card in a test picture (in the same lighting) and then use software that can easily correct it to be the correct known neutral color. If using Raw images/software, this is one click, to correct all of your pictures in that same light. Then we really don't worry about the color of the light, we simply just easily fix it. It is virtually trivial.

Adobe software (Elements, Photoshop) can also do this White Balance correction to JPG images, adequately, however this does not have the full range raw has. And of course Raw software does this even better. See White Balance Correction, with or without Raw

A problem like your yarn color needs more careful work than just some random snapshot of the dog. :)
 
Last edited:

Dyers

New member
I'm with the OP. Diagnosing a problem is sometimes more challenging than finding a fix. We've been chasing our tails about this for the better part of 2 years. When we read about white balance, and implemented it's use, there was a marked improvement in images taken under that specific light source (in this case, daylight CFL) when a wide variety of colors are present, but not when the colors present are near each other on the color wheel. We thought the white balance would solve the issues with the CFLs. Your explanations help better understand why it isn't the total solution. At first, our attempts to solve the issue were fairly logical, but as time passed without notable over-all improvement, and as logical solutions either escaped us, or failed, the attempts became more desperate.

Monitor calibration sounds right, like a correct part of the solution, but it left me wondering how one knows if progress has been made, or if the issue has just been exacerbated. We can spend countless hours trying to edit with the calibration results from one application, then calibrate using a different software only to find that the images we thought we were making look more true still look different. There are lots of pieces of software, and cool gadgets that each promise an exclusive solution to calibration, but I wonder if certainty can be found in using them; I'm beginning to wonder if monitor calibration isn't a little like the emperor's new clothes?

So, we've been given two likely means to solve our color issue. Using flash alone isn't likely to be helpful as the still life work we're doing with these yarns is rather close, and there isn't a white ceiling to bounce the flash. Years ago, I used what were marketed as "Grow-Lux" incandescent bulbs, but those colored the images towards blue. As someone suggested above, that's easy enough to correct in photo editing software (we currently use Photoshop CS3), and realize that will always be a part of the total solution. I have not tried other alternatives recently, and since the crack-down on incandescents, I don't even know what's still available. Light source is probably the single biggest readily affordable change we can make. What bulbs do you all use?

The second is raw. I know that BrWhatsit's D40 will shoot raw, that it's a lossless format, and not much else. That's an avenue I want to actively explore now, too. What might the work-flow of processing a raw image look like? Is Photoshop even the right software? Relatedly, if we shoot a bunch of images, offload them in jpg format which is lossy, edit, then save them as pngs, are those pngs still considered lossless?

Wayne; the article on color balance referenced in your post is very thorough, and is going to be a lot of help. One sitting with it was enough only to get a brief overview. It's not over my head, but it uses a language with which I'm not intimately familiar, so it takes longer to digest what I'm reading. The writer, Wayne Fulton (is that you?) talks about clicking on something white in the image that's in the same light as the subject (the white t-shirt vs. the white water in the rapids), and that's something I want to play with. A very low percentage of our images have any white; we usually shoot on a seamless background made from black Formica. Will it work to put something white just beside the subject that can be cropped out after color editing? Is a postage stamp sized white object big enough? Is a piece of printer photo paper worth trying, or would something spherical like a golf or ping-pong ball make better sense?

I've owned 5 cameras; an Argus C3 that had previously been my dad's, a mid-seventies Focal (K-Mart brand) 35mm SLR, an N90s, the only camera I will ever love, and two junky digital cameras I bought just for snap shots. The flash for his D40 is in need of repair, and the auto-focus isn't functioning (as I recall, the problem might be with the af lens, and not the body). I've been looking at other cameras, and settled on purchasing a D5300. I think this is something I need to do to 1. spend less time behind the camera, and 2. to spend less time in a photo editor. Your comments make me think the lighting, and raw issues are higher priorities that could use careful examination whatever camera we have; they're certainly more affordable. Are there any inherent problems with the D40 that suggest another camera is advisable sooner than later from your perspective?
 

WayneF

Senior Member
The second is raw. I know that BrWhatsit's D40 will shoot raw, that it's a lossless format, and not much else. That's an avenue I want to actively explore now, too. What might the work-flow of processing a raw image look like? Is Photoshop even the right software? Relatedly, if we shoot a bunch of images, offload them in jpg format which is lossy, edit, then save them as pngs, are those pngs still considered lossless?

I use Photoshop, but Lightroom would seem the good entry path to Raw.

PNG does use lossless compression (Zip method, royalty free), but it is not lossless editing. I am skeptical Raw software will write PNG. Plus PNG can be 8 or 16 bits, you would want 16 bits. And of course, it would already have White Balance (and gamma) done, so changes have to shift old changes back and forth, which is not good, and not lossless. Lossless Raw editing never shifts the data at all, except the one final last time.
So PNG would not be the same philosophy as Raw.

Frankly, seems unnecessary, your original Raw file is the best archive.

Wayne; the article on color balance referenced in your post is very thorough, and is going to be a lot of help. One sitting with it was enough only to get a brief overview. It's not over my head, but it uses a language with which I'm not intimately familiar, so it takes longer to digest what I'm reading. The writer, Wayne Fulton (is that you?) talks about clicking on something white in the image that's in the same light as the subject (the white t-shirt vs. the white water in the rapids), and that's something I want to play with. A very low percentage of our images have any white; we usually shoot on a seamless background made from black Formica. Will it work to put something white just beside the subject that can be cropped out after color editing? Is a postage stamp sized white object big enough? Is a piece of printer photo paper worth trying, or would something spherical like a golf or ping-pong ball make better sense?

Sure, just put a white card (Porta Brace or WhiBal are good) card in the scene, at the image edge to be easily cropped out, but in the same light. No problem. Maybe stamp size, certainly wallet size, it should be large enough to be able to click on it. Or what I do for studio sessions (many pictures, same lighting), with Raw, it can be in the first test picture, and same WB easily applied in one click to all images in the session (same lighting).

Cheap copy paper might be better than glossy photo paper, because the excessive brighteners might appear bluish (it is not a big problem, but it can be present). Or an envelope or a business card, etc. Nothing fancy. Anything is probably vastly better than nothing. Just in the same light.

If you have a few minutes, see the video at Why shoot Raw? It is about Raw, but frankly, it is mostly about white balance.

Many snapshot scenes do often have white things in them. They are not always perfect white, can be off color, but many white things are intended to look white, and many things are not bad. It is common to see postings of beginner images with terrible white balance, when simply clicking the wall white baseboard in the scene, or a piece of white paper in the scene, etc... will make it very vastly better color. Perhaps not 100% perfect, but generally overwhelmingly better than it was, and better than 90% :) I find picket fences, church steeples, envelopes, white dresses or T-shirts, to be vastly better than no try at all. Try it, and in the few cases not so, then click Undo, at least you tried, no harm done.

But something known to be a neutral color reference is best and reliable, and not hard to plan on.

Are there any inherent problems with the D40 that suggest another camera is advisable sooner than later from your perspective?

I'd say no problems at all. Newer cameras may have advantages, but in regard to white balance, raw is raw.
 
Last edited:

Dyers

New member
You have been an eye opener. I'm very grateful for all this help. I didn't know much about raw, and I'm beginning to learn how much I don't know. I like it that all of the original data collected by the camera is retained even after editing. I really like it that it's a tool that I already have on my machine. The bulk editing features seem amazing, and they are the basis of this question. In the images you edited simultaneously, your subject was not only sitting in the same lighting, but she was wearing the same color. Could you bulk edit as successfully if she were wearing different clothes in each image, or if different people were sitting in otherwise identical conditions? In our situation, we're photographing yarn which, while infinitely more patient than human subjects, are created in an infinitely wider variety of color. Is the bulk edit feature of raw likely to work for me, or are we better off by just editing the several views of each yarn together, but separate from any other colors in the queue to be corrected as you say?

After I asked about making a white card, I read what you said about brighteners that actually lean blue. Sensible about the copy paper alternative, but I'll try a Portabrace, $5.21 & free shipping at Amazon, but it's part of their Add-On line of low priced items that wouldn't otherwise be practical to ship, and you have to place a minimum order of $25, I think. I order from them weekly, so I just added it to my list.

When I open an image (jpeg) in camera raw, the white balance drop-down displays only 3 options: As Shot, Auto, & Custom. What do I do to get the other options I see in your video?

ETA: I forgot to ask about something you wrote in your last post.

PNG does use lossless compression (Zip method, royalty free), but it is not lossless editing. I am skeptical Raw software will write PNG. Plus PNG can be 8 or 16 bits, you would want 16 bits. And of course, it would already have White Balance (and gamma) done, so changes have to shift old changes back and forth, which is not good, and not lossless. Lossless Raw editing never shifts the data at all, except the one final last time.

How do you mean shifting?
 
Last edited:

WayneF

Senior Member
You have been an eye opener. I'm very grateful for all this help. I didn't know much about raw, and I'm beginning to learn how much I don't know. I like it that all of the original data collected by the camera is retained even after editing. I really like it that it's a tool that I already have on my machine. The bulk editing features seem amazing, and they are the basis of this question. In the images you edited simultaneously, your subject was not only sitting in the same lighting, but she was wearing the same color. Could you bulk edit as successfully if she were wearing different clothes in each image, or if different people were sitting in otherwise identical conditions? In our situation, we're photographing yarn which, while infinitely more patient than human subjects, are created in an infinitely wider variety of color. Is the bulk edit feature of raw likely to work for me, or are we better off by just editing the several views of each yarn together, but separate from any other colors in the queue to be corrected as you say?

If the goal is correct color, as opposed to tweaking to enhance colors, then no, it does not matter what color clothes or which people are sitting there. The WB goal is to make a white card actually come out white (don't position it so other colors can reflect onto the card - just the direct light only). When achieved, and the card is accurate, then all other colors will be shown as they are too, also accurate. You will have matched WB to the actual lights color.

Now, there are people that don't quite like neutral color, and prefer their portraits to be a little warm perhaps. But after correcting it, you can then slightly offset the WB a bit, to look as you prefer it (from this known base). Or not. To me, the goal is correct color.

Yes, if you don't change the lighting as you go, I think correcting the WB of the yarn will work fine. I think it is the only chance. :) OK, more crude tries (using only camera WB settings) can get halfway close, just never quite precise.

But if seeking most accurate color, I would not use fluorescent lighting, which may or may not be OK, but that is another issue. Incandescent, flash, or sunlight are fine. Don't mix them however, use one type of light source.

Note that exposure also can change shades of color. Overexposure reproduces the shade lighter, and underexposure makes it darker. You can adjust that in the Raw editor. And if you are printing it, that is another issue. And if shown on the internet, of course, you have no control about the calibration of a customers web monitor. Color is not the easiest thing. :) But accurate WB will at least let it start out right.

If using Adobe software, be sure your final JPG output uses sRGB profile, and NOT the Adobe RGB profile. The internet and and the photo printing places use sRGB.


After I asked about making a white card, I read what you said about brighteners that actually lean blue. Sensible about the copy paper alternative, but I'll try a Portabrace, $5.21 & free shipping at Amazon, but it's part of their Add-On line of low priced items that wouldn't otherwise be practical to ship, and you have to place a minimum order of $25, I think. I order from them weekly, so I just added it to my list.

When I open an image (jpeg) in camera raw, the white balance drop-down displays only 3 options: As Shot, Auto, & Custom. What do I do to get the other options I see in your video?

Is it Adobe Raw software? It sounds like you are using a JPG file instead of a Raw file. That's what they do.

Adobe Raw offers lossless editing of JPG too, however JPGs are already JPG, and they already have WB in them, so you cannot now choose original WB. It has been done. You can tweak it now. So now you see a + / - type of adjustment for JPG.

Whereas Raw files will show the normal WB menu, Daylight, incandescent, etc. They have no WB in them yet.


Lossless editing:

Adobe Raw does lossless edit on JPG too, but you need to know what that means. You edit it and change stuff, WB and color and contrast and exposure and crop it, and whatever. The original JPG data is NOT affected. Original file is always preserved. It saves a list of your edits (somewhere else in that JPG file), and when you output ANOTHER JPG, then it applies those changes to the original data, and you get a new corrected JPG. Data is shifted only that one time (plus the JPG the camera made).

If you show that original JPG file in other programs, they only see the original. You will see NO CHANGES. They don't know about this list of edits. You have to output another JPG for other uses, which will include your edits. NEVER EVER overwrite your original file, ever. But... no matter how often you go back in there and shift the edits around more, you are NOT shifting the original data back and forth. If you make more changes, you only change the list, not the data. You shift the original data only ONE TIME when you output the new JPG. Then, if you discover you want additional changes, you discard that last JPG (expendable), and go back to your lossless original master, change the change list, and output a new JPG, which is shifted only that one time. Never any shifting data back and forth, which can be destructive.

This list will be invisible to you, but it is how it works, and these settings will already be made in the raw tools. You can always reset them off, and have your original again. Raw files work that way too, lossless edits. It means the original data is always preserved. The output is shifted only one time.


ETA: I forgot to ask about something you wrote in your last post.

How do you mean shifting?

To change WB or exposure or contrast or anything affects tonal values. Changing tonal values is a shifting the RGB channels lower or higher (in the histogram). The way you change the color of anything involves shifting it in the histogram, from one value to another value.

For example, look again at White Balance Correction, with or without Raw

at the very bottom image there, at bottom of page. It is an animation showing white balance shifting channels, in this example, from incandescent to flash WB. Watch the RGB histogram channels shift. Red shifts one way, blue shifts the other. This is what WB does.

This is a very drastic change or shift. Not much else gets that drastic. WB is done in the camera to be JPG, while the data is still 12 bits. Or it is done after uploading Raw data into the computer, while still 12 bits. Raw has several important virtues, and this is one more. However, this much shift would be very hard on 8 bit JPG data, just couldn't do that extreme. Mild shifts work better on 8 bit JPG data.

But JPG already has the WB done on it, not necessarily precise, but milder adjustment is all JPG ought to need.

I'm going fast, I hope it makes sense.
 
Last edited:

Dyers

New member
If the goal is correct color, as opposed to tweaking to enhance colors, then no, it does not matter what color clothes or which people are sitting there. The WB goal is to make a white card actually come out white (don't position it so other colors can reflect onto the card - just the direct light only). When achieved, and the card is accurate, then all other colors will be shown as they are too, also accurate. You will have matched WB to the actual lights color.

The goal is definitely to correct the image to accurately represent the colors we worked for.

Yes, if you don't change the lighting as you go, I think correcting the WB of the yarn will work fine. I think it is the only chance. OK, more crude tries (using only camera WB settings) can get halfway close, just never quite precise.

Immediately after reading about white balance, then shooting using a custom white balance setting on his D40, we noticed much improvement in some images, but others that seemed unaffected. At first, we looked for relationships in the colors that were difficult. We theorized that certain colors were more difficult to get right; yellow is more difficult for some reason, but I don't know if that's a WB issue or not. Eventually, we found that many reds, oranges & especially yellow solids, and some variegated yarns using analogous colors, color wheel neighbors as it were, were the difficult skeins to bring to accurately represent the colors in the yarns. I'm anxious to try this using raw, a white card, and the additional smarts we've picked up from you. Waiting for a battery.

But if seeking most accurate color, I would not use fluorescent lighting, which may or may not be OK, but that is another issue. Incandescent, flash, or sunlight are fine. Don't mix them however, use one type of light source.

Incandescent, then, seems the more practical for us, but with time & available cash at a premium, this is a variable I'd like to check off the list without doing a lot of experimenting to find it. Should I look be looking at wattage, at color, at (do they even still use) lumens? If so what sorts of values am I after?

Is it Adobe Raw software? It sounds like you are using a JPG file instead of a Raw file. That's what they do.

Yes, Adobe Camera Raw. I didn't know I even had it until your video showed me how to access it through Bridge. And it was a jpg I tried; I haven't made a raw file yet to play around with.

Lossless editing:

Remember how Edith Bunker's facial expressions changed as she came to an understanding something? I caught mine doing that as I read your lossless editing explanation, including the answer to my question about shifting.

I'm going fast, I hope it makes sense.

Not too fast; it is making sense. I'm only mildly frustrated that I can't try this stuff out until the battery arrives; hopefully, that'll be Saturday.
 

Dyers

New member
I calibrate my computer monitor. These types of devices generally allow more accurate color reproduction so you should be able to view any differences between the real colors vs. the colors in your photos. From there you can tweak your colors using some type of editing software.

Hark, when you mentioned this Spyder4pro calibration system, it was the first I'd heard of it, I think. And Wayne, in your excellent article about WB correction, you said

It is a good plan to start with a monitor calibration system, to be sure you see things right. I use an old DataColor Spyder 2 Express (most basic version), which seems more than enough. Its sensor reads known colors and intensities on the monitor screen, and adjusts video response so that they display as expected.

The Spyder4 can be had for $79 incl. shipping. The knowledgebase at datacolor.com indicates Windows 8.1 compatibility. After 10 years with XP, I moved to 8.1 about 6 months ago.

The Spyder2 is still available in used condition, and I found 3 at or below $35 incl. shipping. However, the datacolor.com has 2 drivers that look like they might be suitable for this device, but neither of them have been updated since 2006, and 2008 respectively, and no mention of a Windows OS beyond Vista is made. If I can use the Spyder2 on my 8.1 (64 bit) box, I think I'll pick one up. If I have to get the Spyder4, I'll probably hold off for now. So, two questions:

1. Do you know if the older Spyder2 drivers can be used in successfully in 8.1, or if the Spyder4 drivers will function in a Spyder2 device?

2. If not, do you know of a free or low priced, entry level software-only solution for monitor calibration that will improve my confidence in the color reproduction on my monitor?
 

PaulPosition

Senior Member
Software only probably won't boost your confidence as it's relying entirely on you eyeballing the correct values. Well, that is, unless you think you're God-given gift to light spectrum analysis. I, for one, don't have this type of self-confidence. ;)

Oh, thinking about that bit where you see too little or too much variant on similar colors : jpegs, unlike raw, are affected by a myriad of contrast, saturation, sharpening and similar in-camera effects (picture control, advanced -_-lighting, etc.) that may play Havok with color rendering.
 
Last edited:

WayneF

Senior Member
Immediately after reading about white balance, then shooting using a custom white balance setting on his D40, we noticed much improvement in some images, but others that seemed unaffected. At first, we looked for relationships in the colors that were difficult. We theorized that certain colors were more difficult to get right; yellow is more difficult for some reason, but I don't know if that's a WB issue or not. Eventually, we found that many reds, oranges & especially yellow solids, and some variegated yarns using analogous colors, color wheel neighbors as it were, were the difficult skeins to bring to accurately represent the colors in the yarns. I'm anxious to try this using raw, a white card, and the additional smarts we've picked up from you. Waiting for a battery.

Fluorescent is tricky, not a full spectrum, which means some colors are simply missing. This makes illuminating those subject colors be a real problem. :) There is a CRI (Color Rendering Index) number on some fluorescent bulb packages, good ones are around CRI 90-92, which means the color spectrum is 90% as good as incandescent (many fluorescent are much worse). So then, most colors are OK, but there can be exceptions. The white card can match colors, but it cannot fill in missing colors. So if fluorescent, specifically look for high CRI lamps. If you have fluorescent bulbs where your wife chooses her clothing colors, she absolutely wants high CRI bulbs there. :)

But incandescent is better color, and no problem, but there are cool, warm, daylight type of bulbs, they should all match (wattage and age/hours too). And it is just me, but my notion is that the daylight type using phosphorus coatings are also not as fully complete spectrum. Nothing wrong with regular incandescent, if using the white card to match them.

Color can always be a problem. And exposure can shift colors a bit. Yellow and red are bright, esp after shifted to daylight or flash WB, and it is not uncommon than one of the three channels is clipping somewhat. This can change things, the color we see if clipped. Just saying, pay attention to the THREE CHANNEL RGB histogram in the camera (NOT the one single gray histogram) as you setup your exposure, and don't clip any channel on the right. Not saying it should be intentional, but a slight underexposure seems no big deal in Raw, we just boost it back up as desired. We can do that to all in one click too. But we can never recover clipped data.

EDIT: Raw files have no WB in them, so this shift is not yet done, but will be done later. But the camera LCD and histogram are showing a JPG which is embedded in the NEF file, which has the camera WB settings in it, so it is showing that final WB. So, the camera WB setting, which does not affect the Raw file, but should be approximately correct value anyway, for judging exposure on the camera histogram. And so you see a better LCD picture. And, Adobe Raw will try to use that WB setting in its AS SHOT result. You can do better than AS SHOT however, later.

Incandescent, then, seems the more practical for us, but with time & available cash at a premium, this is a variable I'd like to check off the list without doing a lot of experimenting to find it. Should I look be looking at wattage, at color, at (do they even still use) lumens? If so what sorts of values am I after?

Portraits of humans with incandescent lights are problematic. Not the color, but because the lights are fairly dim, and we end up with the lens wide open at 1/30 second, trying to make it work. (modern ISO helps today). But we'd like a shutter speed of say 1/100 second to stop motion of the subject squirming, and maybe we want f/11 for depth of field, esp on your yarn. Flash is greatly more practical to do that, brighter for human subjects.

But the yarn is inanimate, and any slow shutter speed, even one full second, is no big deal at all (assuming camera is fixed on a tripod). So brightest lights are unnecessary, optional. A bigger concern would be the "lighting". A light bulb is a small harsh light (dark sharp shadows), but large lights (like an umbrella, maybe 3 feet diameter) casts nearly shadowless soft light. This could be any white reflector, reflecting a big surface lighted area back to the subject (from many directions, simply because it is large, and near).

The problem with incandescents today is finding them to buy. Some CFL lamps are real problems, color wise. Amazon still has some various incandescents, and also I think 150 watt size may still be manufactured.

There are special photoflood incandescent bulbs for photography, high watts and short life, bright and usually 3400K for special older indoor film, which was important in 1960, but really no longer important for digital WB since we can match most situations now. But if you did want high power, this is one way. (high power is hot and uncomfortable of course).

But I think a longer shutter speed is of no concern in your use.
 
Last edited:

WayneF

Senior Member
1. Do you know if the older Spyder2 drivers can be used in successfully in 8.1, or if the Spyder4 drivers will function in a Spyder2 device?

The Spyder2 driver version 2.3.6 works fine in 64 bit Win7. I don't know about Win8, but I thought the drivers were the same?
 

Dyers

New member
Software only probably won't boost your confidence as it's relying entirely on you eyeballing the correct values. Well, that is, unless you think you're God-given gift to light spectrum analysis. I, for one, don't have this type of self-confidence. ;)

I think you're right about a software-only solution not boosting my confidence. The attempts I've made at using them left me rather jaded about monitor calibration. This thing seems like it might be worth trying, and it has a lot to do with the recommendations you all have made.

Oh, thinking about that bit where you see too little or too much variant on similar colors : jpegs, unlike raw, are affected by a myriad of contrast, saturation, sharpening and similar in-camera effects (picture control, advanced -_-lighting, etc.) that may play Havok with color rendering.

My N90s had auto settings, but with the exception of auto focus, I didn't ever learn to use them. Honestly, I don't know why raw never caught my attention before, but it seems like it's going to be one of the tools we'll be using to get the accurate color we need without so much tedious time spent trying to manually correct color with the tools you just mentioned.
 

Dyers

New member
Okay; that helps a lot. I think you've given me enough information to be able to find the bulbs I need; lower power to reduce heat, and to avoid the more expensive fixtures one needs to use with higher wattage lamps. Not going to concern myself with color temp. ratings, nor with designations such as "daylight" because I can correct whatever color the bulb introduces. Thank you very much.

You mention color clipping. I've heard audiophiles talk about clipping for 35 or more years, and while I have a vague idea of what that sounds like, I wouldn't want to have to try to explain it to anyone with the little bit of understanding I have. Used in this context, I'm kind of clueless. The first time I read about it's use in this way was in something you wrote, I believe in the article you referenced early on. What happens with color clipping; what would I be likely to see?

We sometimes need to photograph rather small, very reflective parts for sewing machines. With these, the color isn't as important as with the yarn, but we were still having lighting issues with them; intense hot spots, strong shadows, and harsh reflections. We built 3 frames of wood over which we stretched several layers of cheese cloth. These are only about 300mm square, but are fairly deep, about 75mm, and this allows them to stand in most any configuration to diffuse the light source(s). The yarn isn't reflective, and possibly due to the use of multiple light sources, and/or the black background, there isn't much shadow to contend with. What I do get are some hazy reflection of the light that bounces off the yarn on the black Formica we use as a seamless background, and I like that a lot. We do use a tripod for all of this work, and also the camera's self timer to avoid any camera movement when the shutter is released. An aperture of f11 is typical, as are very slow shutter speeds.

The Spyder2 driver version 2.3.6 works fine in 64 bit Win7. I don't know about Win8, but I thought the drivers were the same?

Which drivers do you suspect are the same; for Win7 & Win8, or for Spyder2 & Spyder4?

This has been enlightening, and immensely interesting. I didn't anticipate so much detail, so much individual attention in your responses, and I am very grateful for it. I want to get to work and put some of it to the test. I'll write more after trying out some of the yarns that have been more challenging so far.
 

WayneF

Senior Member
You mention color clipping. I've heard audiophiles talk about clipping for 35 or more years, and while I have a vague idea of what that sounds like, I wouldn't want to have to try to explain it to anyone with the little bit of understanding I have. Used in this context, I'm kind of clueless. The first time I read about it's use in this way was in something you wrote, I believe in the article you referenced early on. What happens with color clipping; what would I be likely to see?

Clipping is the single most important factor about monitoring exposure (because it cannot be corrected).
Clipping just means overexposure, but in a specific most-harmful way:

Adding additional exposure (in camera, or with ACR Exposure slider) moves the histogram data right (brighter).

When the right end of the data reaches the right end of the histogram, it can go no farther, there is nothing beyond 255 (8 bit numbers can only range from 0 to 255... 8 bits can count no higher, 256 requires a 9 bit number, which JPG does not have). So the right end just hangs there clipped at 255. Additional exposure just piles up more pixels, clipped at this right end at 255, and we see a tall spike develop in the histogram data right on the 255 end. That spike denotes clipping (data piled up at the 255 end).

We can have overexposure (say of dark colors) that does not reach 255, and this can be backed off and corrected. No actual problem, at least not in Raw. In fact, some intentionally do this, and call it Expose To The Right, as a way to reduce digital noise. When we shift the data back down, we shift the noise down too.

But if clipped at 255, it cannot be corrected or restored, that data is gone (clipped, converted to 255 when it should be higher). Mild cases probably only clip one of the RGB channels, but this still changes color, the colors are not what they should be.

You can see this happen in ACR, in a test, just boost exposure too much with the Exposure slider, more than enough to reach the right end.

(I should shout IMPORTANT here... :) ) When there is actual concern, in ACR, you can see which pixels are clipped or clipping, by holding the keyboard ALT key, and touching and holding the ACR Exposure slider with the mouse, and the screen image turns black, except for the pixels that are clipped. As you move the slider, you see more pixels be clipped. Adobe Levels works the same way. Now you know what is clipped, or what you are clipping.

If this test case clipping is something like a human face portrait, probably the face suffers early clipping. The skin color becomes white and pasty color, with no detail in it.

But there are don't care cases when a little clipping may not always matter (in unimportant image areas, maybe a glare on a table surface), and benign clipping is one way to increase brightness and/or contrast (there are better ways).


We cannot restore clipped data in the original image (if it is clipped and gone), so it is important to monitor this histogram in the camera, at the scene at the time, when it can still be corrected and repeated. It merely takes a glance. The camera shows one single gray histogram, which is useless to show clipping. It also shows three RGB channels, which is where it's at. If any of the three RGB channels show clipping (the spike touching the right end), back off on camera exposure. Red flowers in sunlight typically clip a little in the red channel.

Here is an example of that: Two types of Histograms


We sometimes need to photograph rather small, very reflective parts for sewing machines. With these, the color isn't as important as with the yarn, but we were still having lighting issues with them; intense hot spots, strong shadows, and harsh reflections. We built 3 frames of wood over which we stretched several layers of cheese cloth. These are only about 300mm square, but are fairly deep, about 75mm, and this allows them to stand in most any configuration to diffuse the light source(s). The yarn isn't reflective, and possibly due to the use of multiple light sources, and/or the black background, there isn't much shadow to contend with. What I do get are some hazy reflection of the light that bounces off the yarn on the black Formica we use as a seamless background, and I like that a lot. We do use a tripod for all of this work, and also the camera's self timer to avoid any camera movement when the shutter is released. An aperture of f11 is typical, as are very slow shutter speeds.

Your diffusion screens sound like a good thing.

Which drivers do you suspect are the same; for Win7 & Win8, or for Spyder2 & Spyder4?

I guessed maybe Win7 and Win8 can use the same driver? I don't know about Spyder4, but would not expect that compatibility. I don't know the difference, but the cost of the new one is not extreme. I just use the old one because it performs very well.

IMO, there is not a big difference, but there is a difference, and calibrated is better. LCD monitors are normally too bright for photos, and calibrating can bring that back down, so the images you adjust will look better elsewhere. I also set the cameras LCD Brightness to -1, to make it better match the computer monitor.
 

Dyers

New member
Your comments on anything you see here that could be improved are welcome.

RAW
light source: (two) 150w Soft White Standard Incandescent Lamps in Aluminum Clamp Fixtures rated to 150w
@ 40cm to 45cm from the subject.
skein2raw.jpg skein3raw.jpg skein4raw.jpg

After WB Tool
skein2wb.jpg skein3wb.jpg skein4wb.jpg

After RAW Correction
skein4raw_correct.jpg skein3raw_correct.jpg skein2raw_correct.jpg

The images in the After RAW Correction row are as close as I can come to accurate, and they're remarkably close. I could bring all the images (3 different shots each of all 3 skeins - this is only 1 of each) into white balance at once, and that only took a few seconds. I could then correct each group of 3 images for each yarn for color accuracy, though I cropped and retouched each image with the Retouch Tool individually. Getting from the middle row to the last row took about 70 minutes, though much of that time was spent experimenting with the various tabs full of tools. I think that will be much reduced when I have a better idea of what each does, and when to use them. This red-orange is the most accurate I've ever come with a shade in it's range. It's still a tad yellower than the yarn itself which is slightly rustier; the green sliders had no effect, and that's all that would have been required to tone down the orange ever so slightly. I chose the two greens because I had been unable to correct their color in a way that distinguished the two greens. The yellows in the 3rd image are right-on. I was unable to get this yellow & this green using Photoshop tools when the skein was photographed under daylight CFLs.

Peace,

Dave
 

WayneF

Senior Member
Your comments on anything you see here that could be improved are welcome.

RAW
light source: (two) 150w Soft White Standard Incandescent Lamps in Aluminum Clamp Fixtures rated to 150w
@ 40cm to 45cm from the subject.


After WB Tool
View attachment 98584


Thanks for including all the data. Makes understanding easy.

I fear you are seriously clipping (overexposing) the white card, to make it unusable. See the histogram? That bright spot at far right is the white card. Clipping it like this removes the color cast you are trying to see. See it in AS SHOT histograms... the red and green components of it are clipped there. That clipping corrupts the color of the card.

I suppose your lights are high in front, making the white card the nearest thing to the lights, and it is already the brightest thing. I would turn the camera horizontal, and put the white card a little behind the yarn (farther from lights), off to one side to allow cropping it out (a little below that bright spot maybe).

If it has to be this much closer than the yarn, then that might be the one place a gray card has advantage (a WhiBal card is good). Gray is harder to clip (its only advantage, white is likely more accurate, and more direct to the point of White Balance).




But the picture seems unpleasantly dark, unlike any catalog I ever saw. Catalogs are bright. Feels like straining to see the yarn here, in the dark. Needs more exposure (but watch out for the white card). Could simply crank exposure up here in Raw.

Except the red yarn is probably bright enough (holding ALT key while increasing exposure shows its highlights start to clip real soon, at least on top. Maybe the lighting could be made more even? Or maybe the dark background is not the best choice. It does hide the shadows cast by the lights, but maybe lighter gray, or even white? I would experiment with it once.

And notice that each color of yarn changes the shutter speed of your camera exposure. Yet the lights are the same, and the setup is the same, exposure would be expected to be the same. That is what reflective light meters do (which is not a plus). But there is no reason the same one correct exposure would not be correct for all. You could use Manual camera exposure, which after you experimented to get it right, which would stay the same for all.

Don't overexpose, esp including the white card, but slight underexposure is no big deal, since Raw exposure can simply boost it back up. That boost is about the same effect as boosting ISO.


Your ACR screen shows color profile is Adobe RGB. But all computer monitors and web screens (and online printing places, and your default printer driver) standardize on sRGB. This is a biggie. Just click that line (below the ACR picture) that says Adobe RGB, and change it to sRGB. This one click affects all future JPG outputs from ACR. You will also want to change it in Photoshop menu: Edit - Color Settings (so it won't fuss about the difference).


I was unable to get this yellow & this green using Photoshop tools when the skein was photographed under daylight CFLs.

Right, seriously avoid CFL when color is a concern. :)
 
Last edited:
Top