Certain colours flourescing in clothing - how to correct?

[gavrd1]

I have recently come across an issue in some clothing photography which was photographed under bounced speedlight flash lighting of 4 different garments in one frame.

Exposure was slightly under at 1/3 stop (to be safe) and white balance was correct (a custom one was taken) so colours all ended up correct except for one which was a pink pair of jeans. Even though all the other garments came out correct the pink jeans ended up being very bright and a little bit more orange looking.

I shot the images in raw and used lightroom 4 to edit them. When I adjusted the exposure to bring it to the correct brightness, all fine detail (i.e the texture) in the jeans disappeared and was very bright and washed out.

However, upon looking at the histograms, there was not a single colour channel that had blown out of range. Therefore I can only think there has been a reaction in the clothing, possibly the dye, which has caused this problem?

Any thoughts how to correct this other than just selectively trying to alter pink garment via hue/saturation? Although this does not solve the problem that once at the correct exposure the detail is all blown out.

Thanks in advance

[PhotomanJohn]

It is hard to visualize what is happening especially the loss of detail with only an increase of 1/3 stop. We could be of more help if you would post the photo in its original form and also with the adjustments you made.

John

[IzzieK]

First of all, welcome to CiC. It will be nice if you can fill in your details in your profile so that we will know how to address you and where in the world you are situated. Thank you...

[Manfred M]

Unfortunately, I suspect that you have run into metamerism; where different garments will show up having different shades, depending on the light source(s). If you are looking for a fix; I'm afraid I don't know of one.

In a past career, I worked for a clothing manufacturer; and colour match quality was measured using a MacBeth colour booth based on a standard 5500K "daylight" setting. This unfortunately did not reflect the lighting conditions at different retailers, and we would get returns because of "colour match" issues.

https://en.wikipedia.org/wiki/Metamerism_%28color%29

[gavrd1]

Thanks for the help everyone.

You can find the original image with the exposure untouched (a jpeg created directly from the raw in lightroom) at:

https://www.dropbox.com/s/8aztit7bscq37dy/clothes-0.jpg

And a copy of the same image with a +0.36 exposure adjustment can be found at:

https://www.dropbox.com/s/44gbdttbu5...hes%2B0.36.jpg

The first image with the exposure correction applied is the correct exposure and tone for all the other garments yet the pinky/orange becomes very bright when it should be deeper rather than bright.

When zoomed in on that image you will also see that virtually all detail has washed out too. If you look at the image with no correction you will see that there is more detail, but look closely at that too and you will see that is even starting to disappear even though it is underexposed slightly.

The pinky/orange jeans are the same material as the blue ones which has displayed correctly.

Any further thoughts anyone?

Thanks

PS. Download the images first as dropbox tends to compress them when displayed within the browser!

[Mike Buckley]

Is it possible that the malcontent color of the pink jeans is not within the color space being used? I literally ask that question, rather than pose it as a likely explanation of the cause of the problem.

[Manfred M]

Gav - Your white balance looks to be bang on; but with a flash, remember that you are looking at daylight colour balance (5500K), and the lights that you are looking at the garments under are likely not. For instance, fluorescent lights tend to be deficient in the red part of the spectrum (especially the cool white type commonly used in commercial settings).

If this is the case, I would expect a far more vibrant red under flash, and that seems to match what you are seeing. You are seeing that the reds are"deeper" which I suspect means darker under the working lights, which tends to support my suggestion that the ambient light source is red deficient.

The other problem with fluorescents is that they have different phosphor coatings to create a "more natural" light. These tend to cover certain wavelengths (as opposed to a smoother curve found with other light sources).

I will always include a colour swatch in my setup shots when photographing products, to ensure I am working from a good base. I use the X-rite Colorchecker Passport and will do a custom profile to work with. That eliminates any doubts about the accuracy of the colour reproduction. This won't solve the problem for you, but will at least narrow dow

http://xritephoto.com/ph_product_ove...ction=overview

The other issue could be your computer screen. If it has not been calibrated and profiled, the colour reproduction may well be off.

[gavrd1]

I did initially wonder that, although the images are used only for a website so I tend to shoot in srgb as opposed to the wider adobe 1998 colour space. With hindsight it may be easier to shoot these work in adobe 1998 and convert back after if needed!

I have shot hundreds of garments in the same way over the years for the company and only ever once had a similar problem before. They were a couple of normal tshirts where the colours reacted in a similar way and this ended up being the light source. As soon as I photographed them in natural daylight instead of flash or tungsten light the colours came out perfect so they were not related to colour space.

I just wondered if the same thing may be happening here.

[Manfred M]

If you are shooting RAW, then colour space makes not difference; as they do not apply to them.

Your work colour space in post-processing, on the other hand By default I use ProPhoto as my working colour space, but anything displayed on the web has be be converted to sRGB, as this is all that is the colour space supported by web browsers.

I would suggest you try shooting this under different lighting conditions; I suspect that this might be the "fix".

[Inkanyezi]

Metamerism is one factor, the other is contrast. The light is very soft, as well as frontal, which does not bring forth much texture. The exposure has put the colour in the shoulder of the tonal curve, where less detail is perceived, and all those factors cooperate in making the pants flat, image-wise.

To bring forth structure, the texture of the textile, a directional lighting works better. Also, to increase contrast in order to perceive more texture, the tone of the item must fall within a part of the tonal curve with more contrast, preferably not at the shoulder - i.e. keeping exposure a tad lower.

One problem source is the speedlights. There is no need for them with a static object, and probably the image or the pants themselves, will never be looked at in the light of a speedlite. When I take objects like these, I use regular household lamps, the ones people mostly use. That caters for a lot of the metameric differences that may occur. I also use directional lighting always, when structure should be retained. If you still want it soft, a honeycomb on the softbox might be needed to preserve structure. Regular household bulbs, colour 827 or 830 (2700 or 3000 K, Ra 80+) for Northern Europe, or perhaps daylight lamps for regions where people mostly use them, are in most cases good enough for shooting clothes.

Pink should not be so much of a problem; more difficult with violet, dark greenish and brown tones.

[PhotomanJohn]

Gav - I have one more comment and that is I have had a heck of a time with speedlights on some materials after they have been ironed. The light areas in your second image look like reflections and are not evident in areas that are not flat and more evident in areas where the clothes were ironed over seams. It looks just like the problems I have had before. I would suggest that you try not placing the hot iron directly on the material when pressing clothing for photos. Use a towel or something over the clothes so the surface doesn't get flattened out and become glossy.

John

[gavrd1]

Thanks for all the help, everything said does make perfect sense.

I shall be nearby the venue again within the next few days so I shall see if I can reshoot them and try a few different scenarios - one with ambient room light, one with speedlights show more directional as opposed to softer bounce and finally then shoot them in pure daylight.

I'll make sure they're not ironed or positioned perfectly flat and see what happens!

[Inkanyezi]

I often shoot products in window light, daylight that is directional. It is a bit trickier than lamps, as it is constantly changing, but provided it is shot RAW and there is a WB target somewhere in the shot, it shouldn't be a problem. Metameric differences though make the image different from one that is shot in lämp light.

[pnodrog]

You could try photographing using only tungsten lights and then correcting for colour temperature. If it is fluorescence it will be far more apparent under high colour temperature lighting and especially if any UV is present. Fluorescence is caused by an atom absorbing energy from radiation at a short wavelength that temporarily bumps an electron into a higher orbit and when it returns to its normal orbit emits light at a specific lower wavelength. If the illumination light contains no wavelengths sufficiently short(energetic) enough fluorescence will not occur.

[Loose Canon]

Hi Gav,

I took the liberty of loading both of your photos Gav. And I realize they are the .jpgs, but I found that in ACR when the shot named “clothes-0” exposure was adjusted to +.36 the pink jeans were blown way out. With no adjustments the shot named “clothes+.36” was starting to show signs of clipping in the pink jeans in ACR.

Since you have done a ton of these you would know better than me and I’m probably way off here but I would have to agree that a major part of this is the frontal/flat lighting angle. But maybe not so much the speedlight itself. It looks like it is washing out the entire set to me. Adjusting the tonal range brings back a bit of contrast and color to my eye.

This is “clothes-0”, before and after, with no exposure compensation, but the black point set and a tad of positive clarity in ACR. No further adjustments, not as much loss of detail, and quick and easy to batch process if you have a ton of these. I was very fairly conservative with blacks @ -36 and clarity @+16. Its subtle as seen here but was a pretty nice difference as seen in ACR. I might even consider a tad of a curves adjustment as well on top of that or you could push it until it looked right. A correct white balance alone does not necessarily guarantee accurate color. There are other factors involved as well. I was a bit surprised that since you are doing this much of these you are not using custom color profiles.

If these are new they probably have not been laundered, but I read the other day that some laundry detergents have a phosphorous ingredient that cause clothes to fluoresce (so your whites can be whiter, colors brighter!). Pretty big difference that can be seen under UV lighting (black lights).

Anyway Gav, just a Theory!

[RustBeltRaw]

Clipping thresholds vary program-to-program. The current version of ACR has a lower threshold for blown RGB highlights than the highlights warnings in my cameras, for instance. The difference is slight, but it's there, and my camera does not provide single-channel clipping warnings (grr!).

I think there are two problems at work, both of which have been sussed out by my erudite colleagues. I suspect Terry is correct in suggesting a blown red channel, Urban's harder-light idea is also excellent. Lighting something flat and matte with a big source, like a ceiling-bounced hot shoe flash, will yield an extremely flash quality of light, much like what you're struggling with.

If you have a pair of hot shoe flashes handy, I'd try placing them about 5 feet away and a foot or so off the floor, aimed somewhere between horizontal and right at the jeans, at about 10 o'clock and 2 o'clock with the jeans arranged at the center of the clock. Then add a ceiling-bounced light or a bounce board for fill light. That should bring out some texture. But bear in mind, I have little experience in shooting cloth. That just seems like a good starting point to me.

David Hobby, as usual, has a Strobist post on lighting for detail which could be helpful.

[t6b9p]

If the illumination light contains no wavelengths sufficiently short(energetic) enough fluorescence will not occur.

Not quite true - fluorescence can be caused by any wavelength e.g. blue light can result in red fluorescence in porphyrins and highlighter pens fluoresce strongly to blue light.

There have been some excellent suggestions but as a last resort you may want to consider this.
Fluorescence is an emission process. If your fabric is really fluorescing, and it is fluorescing strongly, then detail can be lost because you are now photographing a subject emitting light as well as reflecting light. Unfortunately, in the process any texture emphasized by shadows may become subdued by the fluorescence emission. I have experienced this with some opaque materials but it is typically more an issue with transparent fluorescing subjects.

[pnodrog]

Not quite true - fluorescence can be caused by any wavelength e.g. blue light can result in red fluorescence in porphyrins and highlighter pens fluoresce strongly to blue light.

The statement was absolutely true in that the light being emitted will be at a lower wavelength than the light being absorbed. The statement I made never excluded blue light from being able to result in fluorescence.The lower the colour temperature of the illuminating light the less likely fluorescence will be observed in the region of human response and the lower the frequency of the light being emitted will be. Overall we agree but be very careful to read a statement before you say they are not quite true. It may have left room for misinterpretation but to say it is not quite true is a bit like saying quantum theory is not quite true.

Of course it is further complicated in that the camera's sensor is more sensitive to red and IR than we are. Lowering the colour temperature of the illumination was merely a suggested technique that may work.

[Inkanyezi]

If you want a soft directional light from a camera-mounted speedlite, the easiest method is to use a reflector for bouncing. I.e. you get for example a white cardboard sheet of reasonable size from a stationery or artist's supplies shop, and someone holds this as the virtual light source from where the light should be coming, and then you turn the flash head toward that white sheet. Without an assistant, a white door might serve, or you hang a white sheet on the wall or on a door. The reflector will serve as a "window".

You will probably have to shade the strobe head, so that no direct light from it will hit the subject. You can easily make a shade for the flash, in the same fashion as a lens shade, but folded from black cardboard and attached to the flash head with a rubber band or adhesive tape.

[t6b9p]

Paul

I understand what you are trying to say and overall we do agree however your statement basically states that fluorescence cannot occur if the wavelength isn't short enough. My point is that it doesn't matter the wavelength, all wavelengths are short enough, as any wavelength (UV, blue, green, red, IR) can cause fluorescence providing an excitation center exists for that wavelength.

The statement was absolutely true in that the light being emitted will be at a lower wavelength than the light being absorbed.

By lower you mean longer.

This is true for most cases of luminescence however (although not a possibility in the OP situation) this statement neglects the rarer upconversion process.