Which techniques do you use to find the right brightness setting for a photo for web?

Question

I'm taking photographs of paintings on a white wall, and I am post-producing them for use on a website.

Very often I'm asking myself "Is this the right brightness? Too dark? Too light?".

I know that nothing replaces:

• experience (know how your own monitor renders well known photos that are well balanced)

• having a calibrated monitor (I already did this)

• testing the photo on many devices (I do test on various PC laptops, iPad, Macbook, smartphones, etc.)

What other techniques (possibly as objective as possible) help you to determine if a photo is too dark, too light?

(I sometimes give a look to the histogram, mostly to know if nothing is "burnt" in highlights, but I haven't found a reproducible method that works 100% to determine is a photo is too light or too dark)

NB: the main problem is that most devices (smartphones, laptops) have a variable screen brightness setting (I often change it accordingly to the room light, and, in the case of the phone, it can even change brightness automatically), so at the end you never know if the brightness condition of the device is good to evaluate the brightness of a photo.

Example here: how to determine which one of these 4 variations would be best adapted for most users?

Answer 1

When you photograph any well (evenly) lit object against a white wall, you diminish the contrast by introducing flare from the reflection of light from the wall into the lens.

Ideally, you would use a neutral (18%) grey b/g to minimize the amount of flare to preserve the contrast. Even better would be a matte black background for the same reason. Set the exposure from a reading from an 18% grey card at the surface of the artwork.

You would then close crop the image and add the white b/g in post for the Web display.

After you get a balanced (check the histogram) image, use that one.

Lighting is better if you use two balanced sources on either side of the artwork at 45° to the original so as to minimize reflections from the flat artwork surface.

When the image looks okay on your calibrated screen, you're good to go. Each user will have to adjust their screen for their conditions, good or bad — You can't take responsibility for remote locations.

Definition: A well-exposed photographic image has a full range of tones and shows detail in the darkest and lightest areas of the image (specular highlights excepted).

Recommendation: A good (not perfect) photographic colour reference is an X-Rite (brand) ColorCheker™ which is made from stable pigments (24 of them) free from fluorescent, ultra-violet, and metameric visual contaminants so they photograph as they appear visually. They're EXPENSIVE but for a good reason. Tip: include one in every photo to ensure consistency.

Answer 2

Here are two photos of the same painting:

I have seen in museums how dark old paintings can be, so I guess the leftmost version lies a little less. The rightmost is sweetened. It's adjusted for full brightness range coverage and subjectively good looking colorfulness. The contrast increase causes increase of color differences and saturation which has been compensated by reducing the saturation and apparently by recoloring the blue cloth. Quite far from the truth, I guess.

You must decide how much you want to sweeten those images which look a little dull on the screen. Ask the payer of the bill.

If you want to stay as truthful as possible (all real world colors are not possible in RGB screens) take two photos of every painting with exactly same light and camera exposure (use manual mode!). The first image has the painting as is and the other has a partial reference overlay image which has at least a piece of black, a piece of white and a piece of 50% grey. It preferably also has some painted colors. It can contain a small painting.

You adjust the photo in your computer so that the reference looks as truthful as possible without causing too much unwanted effects to the painting. In Photoshop there should be no difficulties to apply the same adjustments to the referenceless photo. Learn to use Camera Raw, adjustment layers and smart filters.

Be sure the reference doesn't cover the brightest nor the darkest area of the painting.

How to prepare the lights and set properly the camera is off-topic here, but at least have some lights, use a DSLR, have a sturdy tripod under it and shoot RAWs, not JPGs. Make all possible to keep glaring reflections away. I have used polarizing filters and used TFT displays as a source of polarized light. The resulted long exposure was possible due the sturdiness of my tripod and timed camera triggering.

If your photos must contain a piece of gallery wall, you can insert it afterwards to keep it consistent.

There can be some reason why you cannot use extra lights. In that case place a BIG grey paper in front of the painting to get a map how the light is distributed. You can compensate the uneven light with it in Photoshop. You can also fix the white balance with it in Camera RAW. Now all your examples look too yellow. In addition they have much camera noise which appears as random colorfulness on the wall.

Here's a totally subjective attempt to fix your leftmost example. It seemed to be the least overexposed version in the bright reflection area on the wall:

Without a reference the result cannot be reliable, it's a guess.

User Stan suggested to use a color reference card as a reliable reference, He hinted that camera can distort colors due those wavelengths which are beyond the visible range but still seen by the camera. The problem is significant at least for those take photos of old paintings because some traditional color materials can reflect well invisible radiation. Color reference card can help to reveal the problem, but IR and UV filtering cures it.

Museums do not allow flashes because the send UV radiation. It detoriates materials, they do not care do your photos get the colors right. Led lights make UV internally but convert it to visible range with fluorescent materials. I have seen writings that they can still leak some UV like fluorescent lamps. One example: https://www.premierltg.com/do-led-lights-produce-uv-led-tanning-beds/. Unfortunately I cannot say does the leaked UV be a cause for wrong photo colors or a reason to kick you out if you try to use LED lights when you take your photos.

Answer 3

It is a really nice question, but it is a bit broad. Let me mention some elements that could be considered.

I am not sure it is initially the responsibility of a web designer to do this. It depends on other factors like the type of client. If the client needs some specific characteristics, for example, a photographer's website, and an art gallery are two examples of where the designer should relly on the treatment of the original image.

What is the correct exposition of the photos? On the case you are posting, I am not sure if you are starting with an underexposed image and blowing it up from there, again, it is a photography technique thing, not initially a designer's thing. This would need a correct exposition setting, including white balance but the mainly gray point. One device that will help you is using an incident light meter, but also the camera must be well known for the photographer, including the histogram.

Another thing to consider is the overall tone of the background, and not only on the photo but on the website itself, is it dark? is it light?

And another thing is the mood, is it a happy site, for wedding photography or a moody dark site for Halloween costumes?

But in general terms, you should not crop either the black point or the white point, and this information is mainly on the histogram, in the example you posted, image number 4 looks that you chopped the white, so, opinion-based number 3 looks in general ok.

This could be the main factor, the white point, and the dark point, but this only works if you actually have one and the other.

It becomes more complicated when adjusting the middle curves, adding gamma to a photo, for example, to pick detail on the shadows. This is relative and when achieved will give you a style, and this is not something that can be objectively measured.