All of my 3D images can be viewed in different 3D formats in my 3D photos page here :


In August 2010 I started taking 3d photos using the cha cha method with one camera (Minolta dimage 7i).

The cha cha method only works when there is no movement in the scene and it also gets inconsistent results as the distance between L/R images is slightly different each time.


Weeks later I made this first rig with two very cheap canon A3350s and metal brackets I had lying around. There are two magnets holding it in the open position.

It was intended to have adjustable lens centre width, but there was nothing keeping the front of the cameras aligned and pointing forwards.


I made this second rig in 2014 using two Canon S95s, stereodatamaker software to syncronise them and, a trigger switch made for me by Frans van de Kamp.

It was still very cheap and worked ok for over three years, but the hinge had some play in it so the cameras could move up and down at the ends.

You can see on the right side a digital level held on with sugru which was good while it worked, but it was easy to leave it switched on which drained the batteries.

Like the previous rig there was still nothing stopping the cameras from rotating around the mounting bolts and it's important in 3d photography for the cameras to be equally aligned.


I thought in March 2018 I could make a very good stereo rig with my 3d printer.

It was maybe a couple of days of designing on the computer and I ended up sticking with the third design after some slight changes.

So this was now my third rig.


I spent ages filing the two faces that meet together when it is opened up fully as these faces determine the alignment of the cameras.

Because of the rippled finish of the 3d print, these faces were nowhere near as flat as they should be.

And the round parts of the hinge were also badly rippled stopping it from opening easily so I had to file and sand those parts too.

A metal file takes a long time to remove plastic and, to avoid getting plastic shavings in the camera lenses, I had to take the cameras out.

I would then file some plastic off, then put the cameras back in with the bolts and see if they lined up properly.

You can see the two magnets that I also used in the older rig which keeps it in the open position.

Although I carefully measured the magnets with digital calipers, it is normal for 3d printers to print with slightly different dimensions so they didn't fit into their holes properly.

Also, after I had filed down these surfaces flat, the magnets then stuck out about 1mm so I had to try and remove plastic from inside the holes.

I realised after spending ages trying to make them fit that I would be quicker making the walls in that part of the rig 1mm thicker in the design and printing two more new parts.

But even in the 3rd rig I still had to spend ages trimming off tiny amounts of plastic inside the holes for the magnets with a chisel till I got them to fit flush with the surface.

While I was doing that I gave myself a couple of finger cuts with the chisel.


You can see how well the lenses line up in this photo.

And it's now not possible for them to rotate out of position as they did regularly in the last rig.

It's good that I can try this rig out and make any changes needed easily and it only costs about 50p for the plastic to make another one.



NEW RIG 2020

In November 2020 I decided to upgrade the now ageing Canon S95s to two Canon S110s.

The S110 is only slightly different in dimensions to the S95 so I was able to use my current rig and make slight adjustments with a file.

After spending maybe a few days updating the stereodatamaker software on both S110s and getting it all working fine I bought two Sony RX100M1s as an impulse buy thinking I could use those in a rig.

But, before they were delivered, I found out the RX100M1s can't be syncronised as they need a micro usb multiport so I had to buy two RX100M2s.

With the RX100M2s you don't need any software to syncronise the cameras, you just need a standard trigger cable in each camera with the wires joined together.

Sony makes a single cable to join the multiports together but it is very expensive at 50GBP. I got two multiport to 2.5mm trigger cables for 11.88GBP total and cut off the 2.5mm connectors.

With my cables there was a red, white and black wire inside. I joined those together and now, when I press the shutter button on any camera, it will focus and trigger the other camera at pretty much exactly the same time.

I took some test shots of cars passing close by at about 60mph and they were synced perfectly.

Another easy sync option would be to fit a wireless remote receiver into each hot shoe and trigger them both with one wireless transmitter, but even the smallest receivers I could find were too bulky for this compact rig.


This is the previous S95/S110 rig which had ridges on the front at the bottom which added a massive amount of strength to the whole part.


But the RX100M2 has a tilting screen which moves over the bottom edge of the camera where that ridge would normally be.


So I started with the 3d cad model of my last rig, made adjustments to fit the size of the RX100M2s, and had to remove those ridges.

But, without those ridges, the part had much less strength and the flat part that bolts to the bottom of the camera was flexing quite a bit.

The Sonys are heavier than the Canons and the weight of the camera was bending the bottom part down making the rig useless as both cameras have to be aligned with each other.


There was no room on the bottom to add ridges and I ended up having to add an extra part on the top (with a ridge) that completely covers the flash.

The camera slides tightly into that part and doesn't allow the bottom part to flex. It also makes it impossible for the cameras to rotate around the mounting holes.

I never use the flash for 3d images so it's ok to cover it and, the hot shoe is still accessible so I can add a flash later if needed.

(Before adding this part I checked how the flash works to make sure it didn't damage the flash mechanism.

The flash is spring loaded, so it is released from inside and it springs up, then you push it back down by hand after use).


In my first few test shots, one camera had much darker images than the other even though they were both at the same settings ie. aperture/iso etc.

I realised later that one camera had the flash enabled but it couldn't spring up. So the camera was adjusting the exposure time etc. expecting the extra light from the flash.

Like all my previous rigs, I have two round pieces of sugru that prevent the cameras bashing together when folded in half. They also stop the rig from closing too far and damaging the hinge.

The original micro usb connectors were quite long so I cut them in half (without cutting through the wires) and added sugru after bending the cable into the best position.

The image above shows the orientation I use when shooting. The right side cable is bent downwards and the sugru rounded off on top so my index finger can reach over to the shutter.


This is the bottom side with the large round knurled securing bolt which has a threaded tripod mount hole in the middle and, the large piece of sugru is also covering the joints of the three wires.


Even though the RX100M2s are about 8mm thicker than the S95s, I am still able to (just) fit this into my jacket pocket when folded in half.


The original magnets you can see further up the page were already weak when I started using them years ago and now they were not doing anything to hold the two sides together.

So I got neodymium magnets which are very strong and even at just 2mm thick these magnets were not easy to keep glued in place.

The strength of the magnets kept pulling one side or the other out, so I redesigned one part with the magnet hole on the inside. There's now two magnets inside on the left below and one glued in on the right.


One of the rig sides split apart in the middle when I first pushed the camera in as it was too tight a fit. The other side split apart in the middle when I dropped it, so both splits had to be repaired.

I used my soldering iron to melt/weld the joins in my 3d printed KX500 fairing and used the same technique here. I melted a thin groove in the split, filled that groove with new melted plastic, then smoothed it over.

Another technique I learned when making the fairing was to use hot water to bend the plastic. The front faces of the rig were not aligned causing one camera to point slightly up and the other slightly down.

I managed to get them much closer by running the hinge area under very hot tap water for around 10 seconds then pushing the part into shape and holding it for about a minute until it cooled a bit.

The cameras were still not aligned horizontally with the left camera tilted down a touch but I fixed that the same way.


I bought a Roland GX-24 vinyl cutter to make graphics for my KX500 restoration project and was surprised how small it cut this tiny CH sticker.

The GX-24 can cut vinyl up to 23 inches wide and this sticker is just 20mm long and 2.4mm high.


The S95/S110s make very nice quality 3d images that look very good even when viewed close up on my 48 inch 4k monitor so I didn't really need to upgrade.

But I did some test comparison shots and there is a noticeable difference in quality on the RX100 which has twice the resolution and a sensor almost twice the size.

The S95 is 10mp 1/1.7" (actual sensor size 7.53mm x 5.64mm), the S110 is 12mp 1/1.7", and the RX100M2 is 20mp 1" (actual sensor size 13.2mm x 8.8mm).

The S95 and RX100M2 both have a 35mm equivalent of 28mm at wide angle which looks fairly natural in 3D.


One difference that might be a disadvantage is the distance between the lens centres. My S95/110 rig had a centre distance of 85mm.

The RX100 has a larger diameter lens so the centre gap is 100mm which is quite a bit more than the considered ideal 65mm for 3d stereo rigs.

A gap larger than 65mm can cause objects to appear smaller than they are which is sometimes an effect that I like, but after some test shots the gap seems to be ok.


Another problem that couldn't be fixed is the battery door. In my previous rig I could fully open both battery doors to change the batteries and sd cards without taking the cameras out of the rig.

But the RX100M2 battery door hinge is actually on the far side of the tripod mount hole making it impossible to open the door even a small amount with anything screwed into the tripod mount.

So I have to remove the cameras to change the batteries, but photos are transferred easily using a usb cable so I don't need to access the sd card. The same usb cable charges the battery.

This shows the location of the battery door hinge.


The Sonys are much more expensive than the Canons, but I was lucky to get two of them on ebay auction for around 100GBP each.

I also got the S110s and RX100M1s cheap so, after selling them (for what they were worth) and the old S95s, the upgrade ended up costing pretty much zero.



COLOUR ISSUES.

It took a while to notice that all the red/blue anaglyphs with the RX100M2 had lots of green in them.

I first thought it was normal for a dubious anaglyph to add a green tint, but I compared anaglyphs of photos that were taken in the same spot with the S95s and the S95 colours were much better with no extra green anywhere.


I always add vibrance to my anaglyphs to compensate for the colours being dulled by wearing the coloured lenses, but I also think vibrance makes them look much better overall.

Adding vibrance to the RX100M2 red/blue anaglyph images changes anything grey or brown (paths, roads, concrete, walls and wood) into green. Even though the sxs image colours are not green.

I found out after lots of experimenting that a blue cooling filter (80) removes the green, but also makes any blue like sky and water far too blue as the blue has already had vibrance added to it.

I tried changing the white balance in the camera to be more blue, but it did the same as adding a blue filter to everything making the sky/water too blue after adding vibrance.

So I created a photoshop image processor script that adds vibrance, then selects a colour range to only include brownish greyish colours, then applies the blue filter to just those colours leaving the sky/water unaffected.

You can see the difference in the amount of green below (wear red/blue glasses). The top image is just added vibrance and the other with the cool blue added to just brown/grey colours.

In the top image, every part of the tree and all of the stones and bricks are green even though there are lots of different coloured stones.


I found out that canon cameras in general have very different colours to sony cameras (adding vibrance exaggerates the difference) and I also noticed the grass in S95 anaglyphs looks very bright and vibrant green.

So I added a third step to the photoshop script that selects general grass colours and saturates them. This ends up with similar results to the S95.

So my process with the RX100M2 is :

1. Adjust the stereo window, save as SXS tiff. 2. Run the photoshop script, save as tiff. 3. Stereophotomaker to add logo to images, save as SXS jpg. 4. Stereophotomaker multi conversion, save all as red/blue anaglyph jpg.



All of my 3D images can be viewed in different 3D formats on my website here :





Click on the donate button to give Craig Hood some money for no reason at all.