One observation I have made during my years of red light therapy usage is a dramatic difference between making contact with the LED diode with the skin versus using it some distance away.
The optical properties of the skin are a critical aspect of red light therapy that is often talked about. We know that below 600nm then hemoglobin, melanin, and other chromophores in the skin will absorb the light before making much penetration. Above 980nm is where water absorption will take over. So between 600 to 900nm is usually considered the ideal "optical window." However this "window" of the skin offers more reflectance than we think!
So this persistent question I always had is about the skin reflectance! It is obvious by observation that the skin reflects red light, the fact that any selfie made under red light shows the skin reflecting a lot of light!
How much light is reflected? Is it significant? How come nobody talks about it?
Unfortunately the red light therapy industry has gone down the wrong path. A very popular company heavily marketed their "clinical power LED" panels and became the standard for the industry. Unfortunately the high power of the LEDs requires users to stand at least 6 inches to avoid tissue heating, biphasic dose, and nnEMFs. At this distance the photon beam will be highly divergent and allow for reflection losses from the skin.
How many high-powered panels like this are used in Red Light Therapy, LLLT, and Photobiomodulation studies on Pubmed? None. So why is everyone so excited to use it? Marketing.
In reality, most studies are using lasers, LED arrays, and LED or laser clusters and mostly placing them in contact with the skin. In one review on the complexity of photobiomodulation parameters, it was simply put as "The first issue to be addressed is light reflection from the surface of the skin, which can be minimized if the optical probe is held in firm contact with the skin."
Our cells beneath the skin are VERY responsive to red and near-infrared light. It only takes 1 to 4 Joules/cm^2 to reach a stimulatory dose. After 4 Joules in most cases the benefits start to decline and eventually reach an inhibitory region. This is well defined as the biphasic dose response curve. Another study makes a specific note about their treatment protocol "placed in contact with the skin and firm pressure applied." 
We can not assume that skin reflectance is insignificant, given how many studies use the lasers or LEDs directly on the skin! The review article also includes that the "Application Technique" (contact or non-contact) should always be reported in studies because it is just as important as other parameters such as wavelength, intensity, and dosage. 
Since we evolved under the sun which emits anywhere from 20 to 40 mW/cm^2 in the Red and Near-Infrared range, it is likely a natural protection from overdose is having higher skin reflectance in this range.
Using a LED or laser in a non-contact method some distance away from the diode will suffer from light losses from skin reflectance. Ultimately it is difficult to fully determine the amount of reflectance due to variations in skin types, skin thickness, and other factors. One researcher says "the number of photons that actually penetrate into the tissue to arrive at the pathological lesion is highly variable."
Some researchers assume white skin will reflect around 90 to 95% of red and near-infrared light! This is very significant amount! Another research paper cites that only 10-12% of light irradiance is absorbed in skin models. 
Researchers assume that the intensities used to reach deeper depths of such as 100mW/cm^2 to 300mW/cm^2 come with the drawbacks of heating the surface tissue and skin.  So essentially to get a proper dose to deeper tissue, you must overdose the skin. This seems like a very inefficient way of doing things that if done improperly would negate the positive effects of red light therapy.
So lower wattage, lower intensity, and smaller more convenient products must be developed that offer more precise dosages, delivered in contact with the skin. This is why we often see LED clusters, arrays, and lasers being used in contact with the skin when we read the actual studies in PubMed. Because they know they get the most precise dosage and eliminate the reflection variable.
Image made by GembaRed LLC adapted from diagrams in 
Although non-contact can certainly be effective and beneficial, we want to make sure people are aware of the actual science of red light therapy. One blog post in the "Rants" section of the Thor Laser blog is simply 3 sentences long and states "You must treat the skin directly, preferably in contact (or very close)."
Too often companies are cherry-picking the science to make their products sound good. At GembaRed we want people to embrace the complexity of red light therapy and choose products based on sound science. We hope this leads to more development of low-powered, efficient, affordable, and convenient products to be used in contact with the skin.
Barolet D, Christiaens F, Hamblin MR. Infrared and Skin: Friend or Foe. J Photochem Photobiol B. 2016.
 Scott Roberts. LED THERAPY
 ASTM Solar Irradiance Spectrum
https://rredc.nrel.gov/solar//spectra/am1.5/ASTMG173/ASTMG173.html?fbclid=IwAR160C9tsMFBrFuoKmDeYlwmzYEIQ3ZnzxIxH_z_iCfmP9SUWVO3GicBdlA Toshio Ohshiro
The Proximal Priority Theory: An Updated Technique in Low Level Laser Therapy with an 830 nm GaAlAs Laser