Public Lab Research note


Grassroots Underwater Photography?!!

by ajawitz | July 25, 2014 16:47 25 Jul 16:47 | #10977 | #10977

What I want to do

I want to examine potential low-cost, underwater imaging techniques by adapting approaches originally developed for aerial mapping. Especially regarding new possibilities opened up by the Mobius Actioncam,

Background

In the process of selecting materials for a waterproof KAP rig, I came across some of those cheap floating LED "disco balls" like the one pictured below-

Floating_Light_Show.jpg

Of course, its impossible for me to see something like this without trying to figure out how I can hack it to do something else... Seeing as I don't own a pool, it would be of little use in its intended purpose, but I DO live on the coast and I WAS looking for something that could both protect a kite camera from crashing while also protecting it from water damage should it fall into the ocean...
Ultimately, the shape of the LED disco ball didn't lend itself to easy harnessing on a kite line, but it DID open up the possibility for exploring a whole new world- underwater!!! Considering our overarching project goal is to find better ways of tracking a species of non-native crab, an underwater capability seems especially fitting.

My attempt

In its stock form the LED Float looks like this-

StockFloat.jpg

The LEDs and Battery are housed in the lid while a heavy circular weight is embedded in a plastic tube that sticks out of the bottom.

Stockinterior.jpg

If the yellow weight is removed (carefully without breaking the plastic) it leaves a perfect space to fit a camera lens.

RemovalBallast.jpg

.The key enabler for this project was the ability to add an extension cable to the lens module on the Mobius Cam. In this case I used a standard lens though a wide angle might work as well.

Mobius_Extension.jpg

I took some extra waterproofing measures by adding silicon sealant around the edges and by wrapping the main camera electronics with a dry bag. After adding some heavy bolts for ballast and foam pellets for cushioning, the end result looked like this-

CamFloat1.jpg

The top lid has a little hole through which a line can be attached.

Ring.jpg

The only thing left to do was to awkwardly throw it in the water!

Awkward.jpg

Results

Floating.jpg

While some moisture did leak in, it wasn't enough to affect the camera and further investigation revealed a loose corner of the dry bag had been caught in the threads of the lid and created a small gap.

The camera results were far better than anything I had imagined! I set it to video mode for the first time but I will try using timelapse too. Some photos are blurry but I can't tell if it has to do with the transparent plastic on the float or if its just a simple focus issue.-

vlcsnap-2014-07-24-21h27m01s0.png

vlcsnap-2014-07-24-21h28m32s120.png

vlcsnap-2014-07-24-21h25m50s51.png

vlcsnap-2014-07-24-21h24m46s184.png

vlcsnap-2014-07-24-21h24m30s51.png

The full video feed and selected screen stills can be found at https://picasaweb.google.com/107186102357963872918/UnderwaterMobiusCam1stTest?authuser=0&feat=directlink

Questions and next steps

The implications of this successful experiment are potentially huge! So many grassroots mapping projects are already focused on coastal regions or marshlands and such techniques would allow them to use some of the same equipment to view and collect data underwater! The initial premise can be improved on in immeasurable ways. The most obvious for me would be to test how the Infragram Point and Shoot version of the Mobius Cam performs underwater! It could also be adapted as a great platform for other PLOTS projects like the RIFFLE by adding Arduino sensors or even a Raspberry Pi. Another immediate possibility is the fact that the stock LEDs that come with the floating light show can easily be swapped out for IR or smarter RGB LEDS like the Neopixel for various nighttime experiments. However, I would want to confirm that such experiments would not have a destructive impact on wildlife first. I know some fishermen use IR cams as fishfinders so it must be possible, but its not something I would do without significant input from others. In theory, I suppose one could even develop the float for my original intended purpose as a hybrid KAP/Underwater rig and actually crash the kite into the water on purpose! That leaves serious practical issues like reeling in a wet kite of course, but it would be a good way to make a deep cast... Immediate improvements I will be making will be improvements to the ballasting system to make the bottom more stable, further waterproofing measures and I might even try using a fishing pole to cast the float farther than I can physically throw.

Why I'm interested

Part of a long term effort to develop citizen-centric research techniques to track impact of an invasive crab species in coastal Maine- http://publiclab.org/notes/code4maine/05-08-2014/mobilizing-data-to-combat-the-evil-green-crab

Began as a standard aerial mapping project using KAP techniques- http://publiclab.org/notes/code4maine/06-25-2014/kap-test-for-invasives-monitoring-project

Developed further into a design/build process for incorporating the new Mobius Actioncam into a water-resistant Kite Aerial Photography rig- http://publiclab.org/notes/code4maine/07-09-2014/kap-rig-for-mobius-action-cam

Version 2= Photo Fishing!

In both attempts to create a waterproof housing for the Mobius Cam, the most challenging part was accommodating the lens. In the initial underwater design, I was able to use the bottom of the clear plastic stem beneath the floating light platform to fit the Mobius lens. However, as some commenters pointed out, the transparent plastic of the LED rig really wasn't meant to be an optical lens, and it didn't take long for it to get so scratched up it wouldn't work. So back to the design board! So if my first design began life as a Kite rig, then my new design, rendered below, bears more of a resemblance to the sport of fishing... Only we're fishing for photos in this case...

Photo_Fishing_2.png

The idea is to use the floats as one would use a bobber on a fishing line, whereas the camera could be attached akin to a "hook"to sink below at varying depths.

Photo_Fishing_1.png

As of this writing, I was able to find at least one U.S.-based retailer of specially designed underwater cases for the Mobiuscam- http://www.eyeofmineactioncameras.com/Mobius_Actioncam_Waterproof_Housing_p/emuwo.htm. The $30 price tag is more than reasonable considering the range of alternatives. That would leave plenty of room in the floating rig to add sensors, transmitters etc... Pictured below with a GPS antenna.

GPS_Antenna.png

Questions and next steps

As of this writing, the new design should be ready for testing within the next week as I'm still waiting for the new camera case to be delivered. In the meantime, I've been experimenting with different sensor combos to embed in the float. Many questions remain concerning the overall usefullness of underwater photography. Does it complement grassroots mapping? If so, how should the imagery best be utilized? Could underwater views be added as a layer on Mapknitter? If so, how should they be georeferenced? The photo below shows one of many GPS configurations I've been experimenting with-

GPSEnclsoure.jpg

The photo shows a solar powered LED Light Float with factory electronics removed and replaced by various components designed for wearable applications. All of my experiments are based around adafruit's excellent Ultimate GPS Module which includes an RTC, built-in datalogging and an antenna attachment.

Many RC/UAV pilots have reported significant interference when using the Mobius ActionCam with a GPS. However, such issues would be highly unlikely if the camera was submerged while the GPS was isolated in its own enclosure on the surface.


11 Comments

So cool!!!!

Reply to this comment...


Really interesting! Eustatic (Scott Eustis) and I talked about trying to figure this out years ago to do mapping of grass beds off Florida.

Reply to this comment...


Thats funny because I stumbled into this project while trying to adapt Scott's/Eustatic's floating KAP rig to the Mobius- http://publiclab.org/notes/code4maine/07-09-2014/kap-rig-for-mobius-action-cam :)

Reply to this comment...


Now I'm really curious to try this with Infragram! Warren- Am I correct in guessing that the Infrablue pack of IR filters will not work with the Mobius and that a Red filter is required instead? If so, is there a particular filter that is recommended? It doesn't look like the filter replacement process is exactly easy, but this is another reason why the removable lens modules are such a huge advantage since they only cost $30 each. Also, have the Infragram modifications been tested with the wide-angle lens or just with the standard?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


We'll soon be shipping red filters (Rosco #17 "Fire") in the filter kits too, but just need to update the documentation. For conversion, definitely check out @cfastie's http://publiclab.org/notes/cfastie/04-22-2014/mobius-ir-conversion

Reply to this comment...


That is correct that a red filter seems to work much better than a blue filter in a Mobius. The conversion Jeff cited is just for removing the IR block filter. Adding an internal filter requires another very easy process that Mathew illustrates here. I have worked only with the standard lens. The Rosco red filter and Wratten 25A filter have not been carefully compared but appear to produce similar results. You can order a little piece of Wratten 25A here.

I think both the underwater and aerial cases you are making deserve to have optical windows. Clear glass camera filters (Skylight 1A or UV) are very cheap and could be glued/siliconed into a hole in your cases. Then the camera can shoot through the glass.

Maybe this is worth a try for the Mobius:

waterproof_case.PNG

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


I'd definitely like to learn more about optical windows. I was actually preparing for something very similar to what you suggest since I didn't expect the images to come out as well as they did. I did some research and found the case made by https://joovuu.com/gb/home/103-mobius-waterproof-case.html. and was considering an alternate build akin to your suggestion that would use a larger version of the LED float that came in the same package. Seen here next to the float I ended up using-

LArgeandSmall.jpg

But the case would have to ship from the UK and the pictures came out far better than I thought they would. Plus, I only just now figured out how to change the focus on the lens and its already had a big impact on some tests I made. I'm also a little concerned that theres a lot that could go wrong any time you cut holes into something what was a watertight enclosure and doing so might negate any advantages the float initially had. I do agree with you about the optical lens though and I've also been looking into using the larger device as a sensor platform since it can pack a lot more electronics-

Sensor_float.jpg

Pictured inside is a travel router configured to serve as an OpenWRT mesh node along with an Arduino Leonardo, a ds1820 temp sensor and a typicla pH probe.

I also just noticed some even cheaper devices that are supposed to float in a pool and automatically release chlorine. I didnt see if it released on a set timer or if it makes its calculations based on water quality measurements. If it turns out to be the latter, it could be the perfect floating sensor platform!

Reply to this comment...


Neat stuff!

Reply to this comment...


waay coool

Reply to this comment...


This is super cool! I'd consider taking photos without any filter in the camera-- the #19 Fire will probably not be helpful. UV is the light that penetrates deepest into water, so filtration in the blue/UV region is going to make the image dark. That's why so many sea creatures can see into the UV.

Expensive option: Quartz optics would be ideal, there is a RPi camera clone that can take replaceable lenses: http://publiclab.org/notes/noctividus/06-25-2014/raspberry-pi-ndvi UV lens manufacturer that makes compatible lenses: http://universeoptics.com/new.html

Reply to this comment...


Yeah, I figured underwater light plays by a whole new set of rules. So much remains to be learned that it could easily get overwhelming. Underwater photography is probably best viewed as something closer to a microscope than a macro-mapping tool. So there are going to be a lot of variables. For example, the addition of artificial light is never considered in aerial photography, for obvious reasons. But how would it effect imagery underwater? What about IR LEDs? Considering our overall project goal is to track invasive Green Crabs, would it be possible to find a way to make light sensitive to the chitin in crab shells? Or perhaps we could actually develop a seabed mapping tool with an ultrasonic sensor? Initially, I figured the underwater pics could be added as a clickable layer on Mapknitter. Depending on how many images we end up generating however, such data would be really tough to work with without some form of geotagging. My most recent edits show how I've been trying to incorporate a GPS module into the rig, but I still need to find a way to match the individual photos with the logged geocoordinates. Luckily, the Adafruit GPS module includes an embedded RTC and its own datalogger. So the key is going to be setting some kind of reference point between the Mobius timelapse settings and the GPS logger.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Login to comment.