Hey everyone, Grzegorz Baran here. Any photogrammetry reconstruction directly depends on the quality of images we use. With better images we get better reconstruction results. A good capture can save us a lot of time and effort while a bad one might result with a reconstruction failure and make entire effort useless. No matter if we shot with a mobile, a drone or a DSLR camera, our main goal should be always to get the best possible images with the main focus on pure surface information. Outdoor scanning is directly dependent on weather and lighting conditions. The better they are, the better images we can capture. Unfortunately, even if in theory we can pick the day to proceed with the capture, we can’t ever fully control the environment and the weather. That’s why to get better images we need to find ways to mitigate at least some of these environmental variables. One of the most challenging outdoor conditions we have to deal with on a daily basis is the light It is never ever fully predictable as it is always in a constant change. Luckily, we have a whole range of tools that we can use in response to it and mitigate to some degree the negative impact it might have. For example, to compensate for a slower shutter speed in low light conditions we can use a monopod or a tripod and stabilise the camera. On the other hand, to cope with sunny weather and avoid direct shadows we can scan in a shade cast by large natural structures, but also cast our own shadow using mobile light reflector, an umbrella or even a tent. But probably the most efficient and the most predictable tool which guarantees predictable and consistent light for outdoor close surface scanning is the flash light. It is basically our own, mobile source of light. strong enough and correctly you're so it can override even a direct sun and direct shadows Also, when aligned with the lens direction it doesn’t cast shadows visible for the camera anymore. The benefits of the flash are tremendous. By illuminating the surface with the flash in poor lighting conditions we can reduce the camera exposure time to the level we are able to stabilise the camera even with bare hands. By illuminating the surface in a sunny day, we can even over bright the sun and make direct shadows invisible. At least in theory as in practice, this approach isn’t perfect and as each, apart from some advantages, it also has its disadvantages. And this is something I would like to present you in this video: the outdoor surface capture with the ring flash-based photogrammetry setup. I guess some of you has seen this weird looking tool used for photogrammetry scanning. This is a ring flash light with the camera mounted on it. In standard outdoor lighting conditions, any environmental light casts shadows independently to the camera. Because the ring flash illuminates the light from the camera point of view, every shadow is always cast always behind the subject from the camera point of view, therefore invisible for the camera. This way we can quite efficiently capture shadow-less images. Shadows are the result of existing lighting and they change following any light changes. They get sharper when cast by the direct light and soft when comes from the ambient light. Shadows change the position and angle when the position of the source light changes. Because every modern 3D lighting system simulates the shadow behaviour, to get proper light response from materials we create, we need to make sure that there is no any shadow information left on the material itself. So, any surface we capture with material in mind, should be captured with as little shadow information embedded as possible. If for any reason the data we captured still contains shadow information, we need to remove it from the albedo before finalising the PBR material The shadow removal process is called albedo delighting. The stronger shadows surface contains, the more destructive for the albedo the delighting process is. There are two types of shadows we have to deal with: ambient occlusion shadows and direct shadows. Ambient occlusion shadows, called also soft shadows, are usually cast by indirect or ambient lighting. These are surface shadow types we usually see during an overcast day or when the surface is hidden in a shadow already. These shadows adds subtle depth to any surface we see, emphasize any crevices and can be often interpreted as a dirt. The most exposed surface parts get the least of shadow while surfaces hidden deeper in crevices gets the most of it and look darker. Since ambient occlusion shadows are usually very subtle, they are also quite easy to remove during the photo editing stage or further de-lighting processing. It gets much harder when we deal with direct shadows. These are shadows cast by a direct sun or any strong point source of light which results with a big contrast between illuminated and non-illuminated areas. Direct shadows split any surface into two totally different tonal levels and if the difference is significant, we might not be able to fit the image into available dynamic range camera offers and lose some surface data by under or overexposing it. And even if we manage to fit the dynamic range and capture all the surface data with direct shadows included, it is still very hard to remove them from the image without losing the original data. So, the best way really to deal with direct shadows is to avoid them at all cost. It’s not a big deal when we look on a single material with shadows embedded in it as our mind will interpret it probably as caused by an additional light source, but as soon we add more materials to the scene where each has a shadow embedded facing in a different direction, our mind won’t find any rational explanation anymore end the image will look fake for us afterwards. Any material which contains a direct shadow in the albedo looks bad as any simulation of light response looks incorrect and introduces fake light behaviour. This is why when we plan to capture any surface with the creation of PBR material in mind, we need to be sure there is no any strong shadow information in the albedo left. Planning a capture trip during an overcast and cloudy days is a very good solution. This way we get enough light to take images with ambient light only, and usually without any direct sun light we would struggle with. Unfortunately, clouds often reduce the amount of light for the camera which has to be compensated somehow, usually with slower shutter speed. Longer exposure affects sharpness of each image and if camera isn’t stable enough, we can get a lot of blurred images. This is why it’s a good idea to compensate lack of light by stabilising the camera with a tripod or a monopod. Of course, there are more ways to reduce exposure time to manageable values. For example, we can widen the aperture at the cost of shallower depth of field. We might also increase the ISO, but this way we always introduce some unwanted and totally random ISO noise to each image. As you can see on this shot, the higher ISO, the more noise we get. That level of noise is based on original images but I increased the contrast so the noise is visible on our screens. Since these images were taken in 6000x4000 px resolution, lets zoom into 1:1 aspect ratio so we can see what is the actual cost of high ISOs. This is why I recommend staying around ISO100 setting and never go above ISO400. Anyway, outdoor weather changes and even cloudy day doesn’t guarantee consistent level of lighting. Sometimes clouds get thinner, sometimes thicker. Sometimes the sun totally comes out for a while and hides behind another cloud a few minutes later. Moreover, cloudy weather is often accompanied by a rainfall, so by scheduling any capture sessions during the cloudy season, we simply risk getting wet but we also risk that any surface we wanted to capture gets wet and therefore becomes useless for scanning. Anyway, scanning during any cloudy day is so far the best and safest option and usually guarantees very good results. It doesn’t mean that scanning in a dry and sunny day is a bad idea. As long we have a good plan and tools to deal with the sun light it’s also totally doable. Also, since we can’t control the weather, and sometimes scanning trips takes longer than a day or two, we always need a plan B and know how to deal with different lighting conditions. So, our main goal is not to wait for perfect weather conditions, but to use the equipment and tools available for us and mitigate the effects of weather and outdoor lighting the way that will enable us to do our job properly regardless the lighting conditions. And for this, there is another solution which comes in handy and which is the subject of this video. We can simply try to overwrite the existing light conditions by using our own artificial lighting system. If the artificial light is strong enough, we can overwrite any direct sun light, bring lighting consistency when the cloud covers changes and reduce exposure time during dark and cloudy weather. This way we can even scan at night in total darkness, which is an option I personally don’t recommend due to practical navigational but also safety reasons. And this is where the ring flash-based photogrammetry setup comes in handy. It allows us to capture any close surfaces in pretty tough weather and lighting conditions. With the flash we can scan during cloudy and windy days, but also when it is sunny and even at night. Unfortunately, even with the flash we still can’t scan wet surfaces. Of course, it’s not as easy as it might look like and there are a few limitations we need to deal with, so let’s jump into details These are the key pieces of this setup. A ring flash light, a camera, 2 polarisation filters and the flash transmitter with the receiver. The flash I use is the Pixapro’s RIKO400, which is the version of Godox AR400 also known as WISTO400. These are all the same ring flash lights just made under different brand and differs mostly with a label on it. This flash light is much cheaper to other competitive flashes of this type but it does basically the same job. It's very durable and reliable in practice. The battery, depending on the power setting lasts for about 400 shots on full power to about 800 when we set it to quarter of its power. The flash tube is fully replaceable if burns out at some point when overused so if this flash is well maintained, it should last for years. The camera I use is the Canon80D with original Canons 18-135mm lens. I am planning to upgrade it at some point when the shutter dies but I am still not sure what brand and camera type I should get. Regarding the lens, the one I use doesn’t creep down when facing to the ground as my Sigma 24-70mm one does and this is the main reason why I don’t use the sigma one which in my opinion with this focal range would be a better pick for this job. To allow for communication between the camera and the flash so we get the flash triggered each time the camera takes the picture I use Godox XT32C flash trigger and this small receiver we plug into side USB slot. The XT32C transmitter has to be mounted to the camera hot shoe. Now, if synchronised, we can control most of flash settings through the XT32C interface.
Both, the transmitter and receiver work in the same, 2.4GHz frequency. To deal with reflective surfaces and benefit even more from the artificial light, we also need 2 polarisation filters so we can cross polarise the light and get rid of any surface glares. In this case one filter has to be mounted directly to the lens, while the second one to the ring flash.
The lens I use doesn’t rotate the outer ring when auto zooming in and out so doesn’t change the polarisation angle after its set, which is an important feature if we are looking for a good lens with cross polarisation in mind. The one I use is the HOYA Circular Polariser in slim version. The glare to be cut by the polarisation filter on a lens has to be cast by polarised under the same angle light. And this is why we need the second, big polarisation filter we mount to the ring flash. For this purpose I use Scan Space polariser with magnetic mount. Next, I need to align both filters the way they cover the same polarisation angle.
Both were designed the same way, so the arrow mark on the lens filter has to be aligned handle from filter mounted to the flash. There are a few more things I use on a daily basis even if they are not core part of this setup. A calibration stuff made of two folded rulers and the X-Rite color checker. A tripod with 90degree column mechanism, and a simple wired remote shutter release trigger which I found really handy as in this setup during any scanning we have quite limited access to the camera. And basically, that’s the basic setup. Might look a bit weird for those who haven’t seen it before and it is quite heavy and definitely not very handy to be held with bare hands. When I use it, I have noticed even more passer-by's asking to explain what I do than when I scan with just a camera or simply with the drone. It’s worth to know a few things to work with this setup efficiently. The flash, when properly used can be really powerful. It’s hard to see with a naked eye how strong the light it outputs really is, since the flash lasts maximum of 1/300th of a second. Also, this is the reason why we can’t see many of these on the video, it’s because the camera I use records only 30 frames per second. And even if would set it to 60 it’s still not quick enough to fully catch the flash in action. Luckily the camera can do it since we can set the shutter speed the way we want. Here is a quick comparison of different power settings with filters mounted on, and without any. The lower the flash power setting is, the shorter the flash duration is. Since the AR400 ring flash has no TTL option and doesn’t auto-adjust its output power to any distance changes, we need to be sure that the distance to the subject during the entire capture is pretty the same. Without TTL option the flash shots every time with constant, fixed flash power level we initially set. To make sure we fill the entire dynamic range efficiently, we need to manually adjust the camera settings until we get the best dynamic range coverage with the certain flash power and distance to the subject. It means that before I start any shooting, I take a few test shots with the color checker visible in each, and analyse the histogram. I usually set the flash power at 1/4th and adjust the camera setting until the histogram covers roughly this area. We just need to be sure it’s not too close to the left and the right part of the chart as this way we risk getting under or over exposed images and irreversible loss of some data. Basically any pixels touching left or right edge of the histogram get cut and are lost forever. A good and proper pixel distribution should look like this. They should be somewhere around safe, middle area of the dynamic range, with any spikes kept away from left and right sides of histogram chart. In this case, spike close to the right side is caused by bright white ruler reflecting the light from the flash, so it’s nothing to worry about, and even though, it is still slightly away from the border edge. Of course pixel distribution depends on the exact surface type and differs between each.
When we capture with cross polarisation setup we should try to stay slightly more away from the left edge representing dark values as cross polarisation loses some contrast details in darker areas what can affect the photogrammetry reconstruction. Basically, try to get most pixels somewhere in the middle and treat a histogram as a visualisation showing the way the data is stored in our dynamic range container because this is what it exactly does. The common camera setting I set when I capture with cross polarisation are:
The flash power set to 1/4th ISO to 200, aperture to 6.7 and focal length somewhere between cropped 24 to 35mm. I usually mount the setup on two extended tripods legs which gives me about 80 cm of the distance from the ring flash lamp to the ground. Of course, depending to the light conditions, I need to modify these values for certain cases. The rule of light fall-off, called an Inverse Square rule, says that the light loses 75% of its power if we double the distance to the subject. It means that any significant distance change is going to significantly affect brightness of images we capture. Since we setup the camera to fill dynamic range for certain distance only, to get consistent results we need to keep all shooting parameters constant during the entire capture, and the distance to the subject is a key here. Otherwise, we are going to end up with series of images where some are fine, but some are overexposed and some underexposed which makes further reconstruction process much harder and sometimes even impossible. Similar rule applies to the camera and the flash angle. If we tilt the camera so it doesn’t face perpendicularly to the surface anymore, the distance measured from one edge of the image to the surface will differ to the distance measured from the opposite edge of the image to the surface. Also, the closer we are to the surface the more significant the distance proportionally becomes.
It means that if we angle the camera this way, the top of the image captured under this angle is going to be much darker to its bottom part. If we angle the flash and the camera the way we double the distance between both frame edges, we can get even 75% of gradient across the image. To avoid this, we need to make sure that the camera and the flash, faces always as perpendicular to the surface as possible during entire capture. I have found that a tripod seems to be the best solution for ground captures as it guarantees constant distance and consistent angling during the capture. Of course, we can shoot holding the camera in bare hands, but when we do, we need to be sure it is away enough so we don’t capture our own shoes. This setup is usually over 3kg heavy and I can tell you that taking a few hundred of images while holding 3kg in front of you as steady and perpendicular to the ground surface as possible, trying to maintain the same height at all time isn’t an easy task. I tried a few times and I believe it’s a very bad and exhausting idea. Of course, sometimes it’s the only choice, but, if possible, I strongly recommend a tripod for this job. The tripod I usually use is a aluminium Manfrotto 190GO! with 90-degree arm. Since as said, with the flash all captures have to be perpendicular to the surface I use just 2 front legs. I use the 3rd one only to park the setup on the ground to mount or demount it or to hold the setup straight. Basically, I set the flash and camera angle the way I can’t see tripods legs anymore. I also use ruler line on the ground as a horizontal reference to make sure the camera isn’t tilted any way. In this position there is no easy access to the cameras shutter release button during the capture and it is where the wired shutter release remote comes in handy. I usually wrap it around the third leg so the cable accidentally doesn’t get into the camera view.
And that’s it. This is the full flash-based setup for surface capture. Of course, it’s not so easy and there are still some technical constrains we need to deal with. The AR400 ring flash has some limits. I usually take around 200 to 300 images to capture an average surface. The number of images needed depends on the surface complexity. While 180 images might be enough to capture 2 square meters of concrete surface, for 2 square meters of grass we might need even 400 or 500 images. The AR400 flash has an overheating protection counter built in which limits the number of images it can trigger in a single series. After the flash hits this number, it activates the protection and need 10 minutes break to cool down. This number of images in a series depends on the actual flash power. AR400 allows us to shot just 40 consecutive images with the flash when set to 100% of output power after each activates the over-heating protection and 10 minutes timer. When we set the flash output power to half, we can trigger the flash 100 times before it activates the overheating protection. 1/4th of power allows us to take 200 consecutive flashes before protection is being activated. Of course, the lower the power of the flash output, the more consecutive flashes we can trigger, but I would say that for materials capture, any power below 1/4th is useless and 1/4th of the flash power is the lowest choice which makes sense. There is a workaround for the over-heating protection. If we turn the flash off after the protection was activated, and turn it back on, the over-heating protection goes away. I don’t recommend it but I guess, when we shot during winter when its cold outside and the flash gets cold much quicker than when working in a summer heat, this workaround makes sense. But I would be very careful with this as the overheating protection was designed for a reason and when we overheat the flash it is going to break and won’t work for us anymore. So honestly, I would use our own judgement as we can clearly see how hot the flash is. In a summer heat when it gets hot much quicker, I wouldn’t even risk to rely on built-in overhitting protection counter and just wait when it gets too hot until it cools down a bit. So, I would recommend to use your own judgment and don’t get crazy as it might cost us a loss of a flash light. This workaround for over-heating protection I mentioned allows us to skip over-heating protection just 2 times after which we need to wait 10 minutes anyway. Also bear in mind, that when we use the flash, waiting 10 minutes isn’t really a big issue, since unlike when we scan without it when we need to hurry before the light changes, the flash gives us much more independency to any outdoor lighting changes, not mentioning that capturing with the flash is much quicker due to much shorter exposure times. Let’s jump into detailed camera and flash settings. I have found that the best setting for any standard ground surface capture with the cross polarisation, when we aim to capture around 200 images, is to mount the camera and the flash about 80cm away to the captured surface. This is the distance I get when I fully expand just 2 front legs of the tripod. For this distance I set the focal length to around 24-35mm on a cropped sensor camera, which would be an equivalent of about 35 to 50 mm for full frame one. I set the flash to 1/4th of output power. I set the shutter speed to 1/250 sec or faster, but not faster than an actual cameras curtain speed which depends on the camera, which without high-speed sync is usually about 1/300th to 1/350ts of sec. I set the aperture to f/6.7. and the ISO to 200. Of course, the camera setting differs depending on a few more factors and we have to change these to adopt to the lighting conditions and capture type. When we use cross polarisation, we lose some of the light power and we need to compensate it with the higher ISO. The result we capture with the full cross polarised setup gives us more consistent results to the one we would get without polarisation filters. Unfortunately, use of both filters costs us around 3ev’s of light, around 1.5 per filter, even if we use a good one. Also, full cross polarisation cuts some information in darker areas and oversaturates the surfaces. So sometimes, for non-reflective surfaces where the glare isn’t an issue, we can prioritise a flash power or number of consecutive shots we can capture over lack of glare. It is important to understand when we can benefit from cross polarisation and use it purely as a tool which it really is. Let’s make a quick experiment and grab a few examples of different substance types together: This is a quick test scene I built to test how cross polarisation affects different substance types. To avoid obvious light bounces and absorb excess of light I used a black fabric as a background. I started with the X-Rite color checker as a color reference in the middle and put set of different kind of props around to fill entire camera frame. We have 2 different pieces of burned wood. Two plastic caps, a black and a purple one. A piece of aluminium foil, an owl made of silver, a ring made of gold. A multitool made of stainless-steel metal with titanium handle. 3 pieces of coal A piece of red brick we can see on a standard brick wall. A plastic ruler I usually use as a scale reference for scanning. A piece of denim fabric. Two different types of sand. A standard, stainless-steel knife. And some more organic stuff like snail shell, a sea shell I found on the beach grain, different types of leaves, and apple two grass blades a feather a lump of earth a cone, some types of wood with different surface types a piece of standard white paper some sugar some salt a few different types of stones with different color, reflectance and surface roughness level and a stone gargoyle With this scene prepared I captured a few images just to see how these different substance types react to the flash light when its cross polarised and when it’s not.
This is the shot in ambient environment light without any direct shadows. This is what we get in standard lighting scenario without any flash but also without any direct light. As you can see, they look natural. Each substance contains some natural reflections and ambient shadows which for most of these substances can be removed or at least calmed down during the image postproduction and delighting stage. This would be the look of surfaces captured in ideal outdoor conditions which are not very common. Now let’s take a look what happens to these substances when we use the flash light to overwrite the exterior lighting and get rid of any direct shadows. As you can see all the shadow information is gone and while for some surfaces, we can still read the color information, some gets noticeable reflection glares. It means that the flash works fine for nonreflective surfaces like soil, some types of rough wood, some types of rough stones. For some like sand, sugar or salt even if they look kind of ok-ish, they get some noticeable micro glares which make them appear brighter to what they really are, but it’s nothing major from this distance. But any reflective surfaces like metals, plastic, coal or shiny stone, not mentioning organic props like leaves, grass blades or an apple, all of these get very strong surface response to the light and definitely cannot be captured this way. And for these shiny and reflective surfaces we definitely need to introduce a cross polarisation approach. This is the image taken with full cross polarisation. As you can see this is the best one regarding the color data from all. All glares are gone. The color gets slightly more saturation and we lost some light response in some darker areas. Unfortunately, stainless steel cannot be captured this way since it reflects all polarised light to the camera under the same angle, and all this light gets cut but the filter. The silver owl looks pretty good, but that's because the silver surface normally oxidizes over time. We can also use half polarisation and cross polarise the light under 45 degree angle. This way we can get some of both words. Get rid of shadows, remove some glares but also get some actual color information. Each of these approaches is just a tool in our workflow. If we have to decide which of these approaches is the best, I would say that each has its specific pros and cons Each works differently when deal with certain type of surfaces and we should aim for the one which gives us the best and the most accurate physically correct surface information for exact surface type in exact environmental conditions. But without any other constrains I would probably go with the full cross polarisation one as this one gives the most consistent and clean surface information free of shadows and glares. Especially that glare can be really tricky as every single surface reflects the light to some degree and gets micro glares. This is why some surfaces looks brighter when captured with the flash, even if we cant see any specular on them. Anyway, bear in mind that the color is just a frequency of light. The color itself isn’t a real thing at all and is very abstract. There is no point to deliver 100% correct color information as current rendering systems has no performance capacity to fully simulated the light response. We just need to be sure that the data is mostly clean from some external visual contamination like shadows, glares and reflections which should be simulated for certain material by 3D rendering systems later. So as long we know exactly what happens to each image, we can use the color checker and calibrate each photo at photo editing stage and adjust all values until each is correct. Each of these approaches is fine when properly applied and each need further color and luminance calibration. But cross polarised approach usually gives us the best quality data for photogrammetry reconstruction. Let’s take a look on 2 real life examples where I used the flash to capture the surface. The one with big scattered rocks is the first one. It was captured in a cloudy day which would be fine to capture this surface even without the flash at all. But due to surface depth and some specularity I decided that the cross-polarisation approach should give me the best result. Here are 3 comparison images. The first one is taken without any flash used and is lit only by environmental ambient light. As you can see, the surface has some ambient shadows, but rocks are pretty matt and don’t get any significant environmental reflections which would affect the reconstruction quality. The next one, is taken with a naked flash without cross-polarisation. This way we get rid of most ambient occlusion shadows but cast quite noticeable glare on dark polished stones. Bright stones, also reflects some flash light back to the camera but not in form of shaped glare but rather by causing overall decrease in saturation values and increase in luminance. The color isn’t an issue as can be tweaked easily using values captured with the color checker but the glare noticeable on a few dark rocks, which in this case are more reflective to others, might have some negative impact on a final result. It’s still nothing scarry as during the reconstruction the photogrammetry software gets an average pixel color value from a few different overlapping images taken under different angle what to some degree flattens the glare In result we might get some level of noise and inaccuracy during the reconstruction, but as long the glare isn’t really significant, we should get rid of it anyway. In cases where glare can affect the quality of the final reconstruction the 3rd approach comes in handy. This one is taken with the flash and two polarisation filters set for full cross polarisation. This way the glare is gone as are all ambient shadows. Since we get less flash light bounced, even from matte surfaces, the cross-polarisation results with slightly higher level of saturation, more contrast but unfortunately proportionally less tone values in darker areas. The cross-polarisation approach is a bit different one when compared to two previous approaches and with this one we need to pay more attention to darker values when setting the camera. If possible, we should set the camera the way to push pixel distribution slightly higher in brighter range and keep them away from very dark tonal values. With the cross polarisation it is really easy to get under or low exposed images and risk the detail lose for example on these dark stones. Here is a comparison of 3 approaches together before and after color calibration. Technically the only big problem in this certain capture in these outdoor lighting conditions is the ambient shadow caused by the depth of the rocks. But I would say that all of these 3 approaches in these exact lighting conditions should do fine for photogrammetry reconstruction. Each just need slightly different photo editing treatment. The last, fully cross polarised approach brings us very clean and well-lit surface without any environmental influence. So, no shadows, no bounced light and what’s the most important, no glare. So why I wouldn’t always choose the fully cross polarised approach? As you can see this surface was captured in perfect lighting conditions. To give you the context of lighting these are the light conditions without any flash used. It's quite dark and the camera needs really solid stabilisation and long exposure time to take useful image. And this is the shot taken with the flash set to 1/4th of its power but without any cross-polarisation filter which would block the light when mounted. With this power we fully overwrite any existing light and we don’t have any problems regarding the stabilisation as the shutter speed can be set to really fast values. Technically it would do the job even in a direct sun but would give us glare we want to avoid. To compare, this is the shot of the flash set to the same 1/4th of its output power but with the polarisation filter mounted on. So basically, the full cross polarisation setup with the flash set to 1/4th of its power is great in given conditions but isn’t strong enough to fully overwrite any stronger outdoor lighting from such distance. Anyway, I set two filters ON to go with full cross polarisation, set the flash to quarter of its power, extended two front legs so the flash was around 80 cm to the surface. For surface with such level of height complexity I set the focal length to 24mm.
ISO to 200. Aperture to 6.7 and shutter speed to 1/250 of sec and I took 161 perpendicular to the surface images with this setting. Next, I calibrated the color and luminance for captured images using color checker values as a color and luminance reference, tweaked some values to get rid of lens distortions, and I used these images to generate the final geometry I could turn into clean PBR environment material. And here it is. Let’s pick another, but this time really extreme example. A cracked soil surface, which was totally exposed to a very strong, direct midday sun. To make the case even worse, I decided to approach this surface from the incorrect direction and cast my own and equipment’s shadow on the surface itself. First, I took a few shots to setup the camera properly. This is the shot of the surface without the flash to compare the difference. Of course, these are direct shadows on the level which would totally mess with the reconstruction Since there was no way to overwrite the sun this way on this flash power and since the earth is a nonreflective type of surface I decided to took the polarisation filter from the flash off and repeated the shot. As you can see the increase in the actual light power is incredible. I adjusted the camera so the image isn’t overexposed, and this is the result. Shadows are barely visible, but we can still notice them. Since I couldn’t just increase the flash output power as it would affect the number of images I can take, I applied the inverse square rule and decreased the distance to the surface about 25-30%. This way I significantly increased the amount of light from the flash hitting the subject surface. And this another image made to compare the difference without the flash, and the image with the Flash. As you can see now shadows are really barely noticeable and these were settings I decided to proceed with: 24mm of focal length, ISO100, aperture of f/11, and the shutter speed set to 1/250 sec I believe the only workable way which left to fully remove the sunlight is to decrease the distance to the surface even more but it would force us to take more images to cover the same area. But since the photogrammetry software always takes average values of the same point from the corresponding images, even if there is some shadow left, as long its mobile, it would be averaged during the reconstruction and won’t appear in final result. Finally, I took 197 images to capture this surface. And this is the final material which comes out of this scan. It looks totally fine and I can call this scan as a success. Of course, capturing this surface with the cross polarisation would give probably even better and more accurate result, especially especially as if we zoom in to this surface really close to see it in a micro scale, we can see it still has some reflective micro elements which probably bounces back with micro glare, but similar to many other surfaces we consider as a nonreflective, these micro glares get scattered when seen from the distance and they don’t make any difference. In this case the top priority is definitely on getting rid of any direct sunlight by overwriting it with flash and it worked pretty well in my opinion. Nothing is perfect and each approach has some pros and cons and this one isn’t an exception. The biggest problem with this ‘close to the surface’ approach is the spine pain. The gear with the flash light can be quite heavy and lifting it up hundreds or sometimes even thousands of times so close to the ground can be very exhausting and these types of captures when we bend over it don’t sound like a great and fun option in long run. And this is the part I want to improve and make the gear somehow lighter and easier to move from one spot to another. My first idea was to replace the tripod with a monopod. Unfortunately, the monopod lacks of the extended arm which would hold the flash and the camera perpendicular to the surface in front of it. So, to mount the camera away from the monopod’s column I purchased a separate boom arm and mounted it on a monopod. It seemed to work and I gave it a try but it failed in practice. I have found that the monopod with an arm isn’t much lighter, but is totally unstable instead. It helped me to realise that a tripod can be parked on a ground what makes entire mounting and demounting quite easy and safe for the gear while monopod simply can’t. With the monopod it’s not as easy as everything has to be held all the time up with one hand which turns setting the gear into a nightmare. By putting it on the ground or leaning against something the gear gets dirty and since its unstable it might get damaged easily. Also moving with the monopod with the heavy flash and the camera mounted to it, isn’t easier at all. Two front legs of tripod actually help to avoid any side tilting. I have also realised that the weight of the equipment is also problematic when I deal with vertical surfaces. Monopod just doesn’t work like it used to when it was used to hold just the camera. And the actual weight is the key factor I didn’t consider before. So, I dropped that idea and decided to use the tripod even for vertical surfaces. When the access to the subject gets really tricky, I just use bare hands to hold the camera. In long term, to save on weight I am considering to get a lighter, carbon tripod instead. I still didn’t find a good solution for low level ground scanning useful to overwrite the direct sun, but I have a few ideas in mind and I am testing these. Of course, I will be happy to share the best findings with you when I find any. And I guess that’s it for this video,
I really hope you found something interesting and useful in here. If you did, please leave the thumbs up, share it and drop a comment. If you want to see even more content like this one, please subscribe to my channel as it helps a lot and motivates to move forward. Big thanks to all of you who did it already. Stay safe and till the next one. Bye !!!! :D
