Gopro Hero 4 for UV photography/video?

[enricosavazzi]

It is the same sensor size as in the Hero 4, formally 1/2.3".

 

I have seen some sample pictures shot with D-mount lenses on the Pentax Q that show substantial vignetting, others that show acceptable or no vignetting. As a whole, D-mount legacy lenses (and some C-mount legacy lenses) were designed for use on 8mm, Super-8 etc. movie cameras. This makes the Hero 4 and Pentax Q sensor a bit too large for most of these lenses (the image area on Super-8 is larger than on 8mm film, but still a little too small). One needs luck to find a D-mount lens that provides adequate coverage, but it is not impossible.

 

There were also some lens designs originally developed for 16 mm and then made in D-mount barrels for use on high-quality 8 mm cameras, so one may get lucky for several different reasons. Very low focal lengths are the biggest problem, but not entirely hopeless.

 

Prices may change, so it never hurts to ask again. Unless the price goes up, like in the CoastalOpt 60 mm. It went up by more than $ 1,000 since I bought mine a couple of years ago.

[OlDoinyo]

FWIW, my father used C-mount lenses on his 16mm Bolex movie camera, so at least some of those should be capable of covering more than 8mm.

 

Edit: Would the C-mount Kern Switar 10/1.6 be a serious candidate, or is that too long?

[enricosavazzi]

That is correct, the large majority of the legacy C-mount lenses are for 16 mm film. C-mount RX lenses, however, are common but peculiar beasts. They are designed to correct the spherical and chromatic aberrations introduced by a 9.5 mm thick fixed beam splitter in the Bolex Reflex. At least the fastest and shortest FLs among the RX lenses do not perform well with only air between the rear of the lens and the film/sensor. This can be corrected by adding a 9.5 mm thick optical glass plate (UV-transparent in our case) at the rear of the lens, or a sandwich of a UV-pass filter and glass plate of equivalent total thickness.The exact thickness probably is not very critical, +- 1 mm might still work well enough. Then you also need to add a 2-3 mm shim to the lens to restore infinity focus, yet another gotcha.

 

I don't know enough about the prices of RX versus non-RX lenses on eBay, but since the shortest RX lenses do not work well on digital cameras (unless you know how to make them work) we might have a price advantage here. Or maybe most sellers just don't know/don't care and try to sell everything at the highest possible prices, leaving the buyers to puzzle out why some lenses are fine and others with just two more letters engraved on the barrel are duds.

 

Other reflex cine cameras used a moving mirror or a fixed pellicle beam splitter, and used C-mount lenses designed accordingly. We can use these lenses without adding a thick glass plate at the rear of the lens.

 

The lenses for the 16 mm format I tried so far all cover the Hero 4 sensor without any vignetting, which is expected because all variations of the 16 mm format are significantly larger.

[OlDoinyo]

Re the RX lenses: if the prism is part of the correction for spherical and chromatic aberrations, then I would imagine that for full correction it is critical that any replacement element match exactly in terms of thickness, refractive index, and dispersion. Not just any random piece of glass would do. Position in the optical train may also be critical.

 

OTOH, some claim that they just use the lenses air-gapped on digital gear and compensate for the aberration with some post-processing wizardry. Whether to take this seriously is a judgment call:

 

http://www.digitalbolex.com/forum/lenses-and-accessories/rx-lenses-will-work-with-the-d16/

 

And is this "DV" version of the lens eqivalent to "RX" or "non-RX?"

 

http://www.ebay.com/itm/SWITAR-DV-1-1-6-f-10mm-BOLEX-CAMERA-LENS-1-6-10mm-AR-CAPS-EXCELLENT-GLASS-SMOOTH-/151684800102?pt=LH_DefaultDomain_0&hash=item23511e5e66

 

Some sources imply that unless it says "RX" or H16" on the lens, it is not an RX. But you can't believe everything you read online!

[enricosavazzi]

Re the RX lenses: if the prism is part of the correction for spherical and chromatic aberrations, then I would imagine that for full correction it is critical that any replacement element match exactly in terms of thickness, refractive index, and dispersion. Not just any random piece of glass would do. Position in the optical train may also be critical.

 

[...]

 

And is this "DV" version of the lens eqivalent to "RX" or "non-RX?"

 

http://www.ebay.com/...=item23511e5e66

 

Some sources imply that unless it says "RX" or H16" on the lens, it is not an RX. But you can't believe everything you read online!

It is not a clear-cut go/no go with respect to the thickness and refraction index of the beam splitter. There is some tolerance. For example, by stopping down the 10 mm Switar to about f/4 or beyond, image quality should be restored even without any glass between lens and sensor (depending on the sensor resolution, desired image quality etc.). At f/2.8, a 6 mm UV-pass filter might already be enough correction. At f/1.6, the original 9.5 mm is most likely to work, but 10 mm might give results indistinguishable in practice. Same thing for the refraction index and dispersion of the glass. If I remember correctly, spherical aberration is proportional to the 6th power of lens size, so stopping down a little goes a long way to reduce it.

 

Position in the optical train is not a factor, it does not change the amount or type of aberration as long as the plate surfaces remain perpendicular to the lens axis.

 

Spherical aberration cannot be corrected in post-processing. Transversal chromatic aberration often can.

 

I don't know about the DV version. In general, I think these lens makers were trying hard to make lens types easily distinguished, to avoid "accidents" due to the mechanically compatible C and RX mounts. If it is engraved RX, then it is RX. If it is not engraved RX (and a greedy seller has not somehow altered the lens or mixed-and-matched parts of the barrel from different lenses), then it should be a non-RX. There might be slight mechanical differences in barrel length, but without direct access to a large collection of lenses to study, we are helpless.

[OlDoinyo]

I looked it up: "DV/AR" is actually an early run of the lenses later called "RX."

 

If you don't have it already, the Zeiss-Jena Tevidon 10/2 might also be worth a look--definitely not RX. I also note that there exists a Computar 8.5/1.3, but it is cheap and may be of dubious quality.

 

Some videographers have also used the Pentax 6/1.2; it is not natively a C-mount lens, but several have modded it to this mount.

 

Theia offers true rectilinear optics in C-mount all the way down to 1.28mm. They are said to transmit IR, but I don't know about UV, nor about the size of the image circle. Claimed FOV is 135 degrees--a true short wide angle.

[enricosavazzi]

Edit: Would the C-mount Kern Switar 10/1.6 be a serious candidate, or is that too long?

It is a serious candidate indeed. Of course it should be tested. Some of these Kern lenses were made in the '70s or '80s, so their coatings might be good enough to be bad news for UV photographers.

 

An interesting thing about the Kern Switar 10 mm f/1.6 (only the model with preset levers) is that there is a dedicated Aspheron front-mount accessory that converts it to a 5.5 mm lens. The Switar + Aspheron must be focused into the macro range of the lens to achieve infinity focus. I guess one could alternatively add shims to the lens mount. If (and it is a big "if") the Aspheron and the Switar 10 mm transmit UV, this might be one of the best solutions for high image quality. I was able to find this particular Aspheron (note that there are several Aspherons, made for different lenses) on eBay for a reasonable enough price, so I will report in due time.

 

In the mean time, I found this Som Berthiot Cinor 10mm f/1.9 quite usable in UV. Perhaps not as good as the Xenon 16 mm I tested above, but one cannot really complain. The problem with these legacy lenses is that each seems to use a unique filter size, so getting all the necessary step-up and step-down adapters takes time.

 

Edit - This is a non-RX version of the lens. There is also an RX version for Bolex Reflex.

 

[JCDowdy]

Another UKA quartz lens that might fit your sensor better than their, yet to be fully unveiled, S-mount collection is their 9mm C-mount.

[enricosavazzi]

Another good find, this time rather unexpected.

 

This is the Taylor, Taylor & Hobson Pelotal 6.5mm f/1.75 in D mount, probably made in the late 1950s. There are rather hard-to-find D to C adapters, which should rather be called D to CS because this is what works at infinity focus. This rather odd lens was made for the 8 mm format, which is smaller than the Gopro 4 sensor. I fully expected this lens to vignette on the Hero 4, but perhaps not by much. I have seen samples with other lenses for the 8 mm format on the Pentax Q, which has the same sensor size. Most of these lenses vignette quite badly.

 

 

Initial tests, however, showed only a very modest darkening of the corners and no true vignetting.

This lens has, at first sight, no front filter mount. However, at closer investigation, the front ring unscrews from the barrel (where the lens barrel shows a thin line). The ring that comes off has a drop-in mount for 25 mm unmounted filters, which is potentially usable for our work. Another oddity is that this lens has an aperture ring but no focus ring. The Backbone modification kit for the Hero 4 adds a squeaky but usable focusing thread, which comes in handy in this case.

 

 

After removing this ring, the lens produces no detectable vignetting. This picture is at f/1.75 without filters, so it shows mostly NIR and VIS.

The thread at the front of the barrel is 28mm-ish. None of my 28mm step-up filters fit, except an odd one made of plastic (and marked 28mm). I was forced to hold a Baader U by hand at the front of the lens, on a 38mm to 52mm step-up adapter. I focused on an external HD monitor connected to the Gopro 4.

All following tests were shot with Baader U and the camera on a tripod. It is impossible to know the exact exposure time, but it was always less than one second. ISO is also a great mistery with the Hero.

 

 

The view with Baader U. There is some vignetting because of the step-up adapter. I will probably epoxy a step-up adapter around the outside of the barrel, which would solve the problem. Because of the wide angle of view (equivalent to 37 mm FL on full frame) and recessed front element, a 1 1/4" Baader U is too small. One needs the 2" version.

 

 

1:1 center pixel crop, at f/1.75. Rather poor.

 

 

at f/2.8, much better. I would call this quite sharp for a Gopro, in fact.

 

 

at f/4, slightly (and questionably) sharper. From here onwards, it gets worse because of diffraction. This is quite natural considering the small sensor size.

 

As a whole, I think I am set with the Hero 4 at the wide end of the focal range, unless one of the cheap S-mount lenses for IR turns out to transmit also sufficient UV. I am not aware of any legacy lenses with significantly shorter FLs in C, CS and D mounts. There are of course plenty of modern fisheye lenses in C and CS mounts, but there is little hope that their multicoatings let in enough UV. There is a small hope left for S-mount lenses because these typically use simpler optical formulas and thinner elements.

[JCDowdy]

Looks like a luck find. You say, "The thread at the front of the barrel is 28mm-ish", but none of your 28mm step rings fit. Did you try a 1.25" filter cell directly on the barrel? The 1.25" filter is usually 28.4 -28.5mm x 0.5 - 0.6mm depending on what you read. A short 1.25" filter cell, like the Baader, might not vignette if it is directly on the lens without a step-up adapter. Does the plastic 28mm that fit vignette any?

[enricosavazzi]

Looks like a luck find. You say, "The thread at the front of the barrel is 28mm-ish", but none of your 28mm step rings fit. Did you try a 1.25" filter cell directly on the barrel? The 1.25" filter is usually 28.4 -28.5mm x 0.5 - 0.6mm depending on what you read. A short 1.25" filter cell, like the Baader, might not vignette if it is directly on the lens without a step-up adapter. Does the plastic 28mm that fit vignette any?

Good idea, thanks. I still have the original cells of a few Baader 1.25" filters, including the U. The cell screws into the lens filter mount about one turn before stopping (different pitch I guess). However, still no luck. There is heavy vignetting by the cell front edge.

There is vignetting also with the lone plastic step-up ring that fits the lens thread. Even the unmounted filter holder that came with the lens vignettes a little on the Hero 4.

 

It seems this combination of lens and sensor tolerates nothing screwed into its filter mount, so I still think I need to attach a wider adapter outside the barrel. A 34mm-to-something step-up ring is a bit too loose, but I ordered a 33-to-something that might just fit with a little epoxy.

 

Edit - The thread pitch on the lens filter mount seems to be about 0.75, so it is not one of the ordinarily used filter threads. On the other hand, this lens was made roughly at the time I started attending grammar school (for reference, I am now three years from retirement), so quite a few things have changed since then.

[JCDowdy]

.....Even the unmounted filter holder that came with the lens vignettes a little on the Hero 4. .....It seems this combination of lens and sensor tolerates nothing screwed into its filter mount........

 

That is to bad, I would be very tempted to drop in a 1" filter stack into the OEM filter holder.

 

Question: How much filter space is there behind the lens in your Back-Bone mount?

[enricosavazzi]

Question: How much filter space is there behind the lens in your Back-Bone mount?

Not much usable space with a D-mount lens. I tried today placing there a 6 mm thick 12.5 mm dia. UV-pass filter in unthreaded metal cell, and it does not fit. The problem is the narrow throat of the S mount at the rear of the CS mount. My train of thought was that this Pelotal lens might be designed for a reflex camera of some sort, and therefore the lens performance fully open might improve with an added glass thickness between lens and sensor, However, no luck. In any case, performance is already good without added glass already from f/2.8.

 

I tested since a Kern Switar 10 mm f/1.6 Reflex (RX), and it performs well from f/4 up without any added glass between sensor and lens. However, it does not focus at infinity (this is the same problem we get when we remove the built-in filter from a camera and don't replace it with a fused silica window).

 

The IR- and UV-cut filters that came with the Back-Bone kit are 15 mm dia. and less than 1 mm thick. The rear of the S mount can contain one or two of these.

 

There is a more recent version of the modification kit visible on the Back-Bone site that can mount one (perhaps two) of these filters in a plastic slider inserted from the side. My version can only mount these filter(s) semi-permanently.

[JCDowdy]

I was looking at the Back-Bone site and it appeared to me that one or two 15mm filters could be installed underneath the S-mount but they do not say how thick the filter(s) could be.

 

 

What I was trying to ask was how deeply into this space does your D-mount lens intrude and is sufficient space remaining in this gap to house a filter?

[enricosavazzi]

I was looking at the Back-Bone site and it appeared to me that one or two 15mm filters could be installed underneath the S-mount but they do not say how thick the filter(s) could be.

Less than 1 mm each.

What I was trying to ask was how deeply into this space does your D-mount lens intrude and is sufficient space remaining in this gap to house a filter?

It is difficult to measure accurately, but there seems to be 6 mm between the rear element of the Pelotal and the sensor window. So, with the two Back-Bone filters on the sensor (in practice, directly touching the sensor window and each other), roughly 4 mm spacing is left.

[JCDowdy]

That seems like enough room to accommodate appropriately sized filters. The diagram says that "Filter Spacer / Holder" will adapt a 10mm filter.

To bad you don't know some nice lady who might have some sufficiently large Baader U shards. ;)

[enricosavazzi]

I managed to modify a few of the lenses I found to be usable in UV to carry a Baader U filter. The first example is shot with the Pelotal 6.5 mm, stopped down at f/5.6. This is a moderate wideangle even on the Hero.

 

Somewhat cropped. Illumination source is a battery of four 9W fluorescent tubes for fingernail gel hardening (they carry a "365 nm" label, I don't know how reliable but they seem to emit a relatively broad spectrum including VIS blue).

 

There is an obvious diaphragm-shaped flare spot, caused by including the fluorescent tubes in the picture frame and not using a lens shade. Both are bad ideas, of course, but this is only a test.

 

The subject is a sectioned lily (?), completely yellow in VIS. The UV pattern is evident.

 

 

And this with Schneider Xenon 16 mm at f/2 and a 330WB80 small filter mounted at the rear of the lens. Not as sharp as the first example. I did try to focus with a field monitor that allows 1:1 pixel crop magnification, but it did not get any better than this. Probably the problem is caused by spherical aberration introduced by the thick filter between lens and sensor, combined with the almost fully open lens aperture.

 

[nfoto]

Interesting observations, thanks for posting these.

[JCDowdy]

Looks good, can't wait for a bee's-eye-view fly by video!

 

Also very curios to see how these C-mounts do on your µ4/3.

 

The 9W fluorescent nail polish lamps I have evaluated (click on Fig.2) are usually just a little longer wavelength peak ~370-375, similar to insect traps (aka Bug Zappers) the 365nm could also be a UV-A dermatological phototherapy lamp. What brand and model did you use? I might have an irradiance spectrum in my files.

[enricosavazzi]

Looks good, can't wait for a bee's-eye-view fly by video!

 

Also very curios to see how these C-mounts do on your µ4/3.

they all vignette, unfortunately. Some more than others, but none is acceptable as a Micro 4/3 lens.

 

When I have time, I am going to make a comparative test of some of these lenses. I will have to test them on Micro 4/3, since the Gopro Hero only has automatic exposure and would not allow me to compare the UV transmission in any meaningful way.

The 9W fluorescent nail polish lamps I have evaluated (click on Fig.2) are usually just a little longer wavelength peak ~370-375, similar to insect traps (aka Bug Zappers) the 365nm could also be a UV-A dermatological phototherapy lamp. What brand and model did you use? I might have an irradiance spectrum in my files.

They are no-brand from China, the cheapest I found. On the lamp it says only "UV 9W 365 nm CE RoHS". Ebay item n. 281419785210. I am now waiting for another set of four to complete my homemade lamp, which has 8 sockets. The armature surface is about 40 x 17 cm and at the back of the tubes I have an aluminium reflector plate with a random embossed pattern (designed as diffuser/reflector for home illumination armatures). This is a relatively large armature that should be sufficient for subjects like small to medium-sized potted plants. A possible upgrade path could be to decrease the spacing between tubes and use 12 or 16 tubes.

[JCDowdy]

They are no-brand from China, the cheapest I found. On the lamp it says only "UV 9W 365 nm CE RoHS". Ebay item n. 281419785210.

 

As I have more than a passing interest in these fingernail curing lamps, I have ordered some and will check them out.

[Andrea B.]

John, that was a very interesting paper.

[JCDowdy]

......no-brand from China, the cheapest I found. On the lamp it says only "UV 9W 365 nm CE RoHS". Ebay item n. 281419785210. I am now waiting for another set of four to complete my homemade lamp, which has 8 sockets. The armature surface is about 40 x 17 cm and at the back of the tubes I have an aluminium reflector plate with a random embossed pattern (designed as diffuser/reflector for home illumination armatures). This is a relatively large armature that should be sufficient for subjects like small to medium-sized potted plants. A possible upgrade path could be to decrease the spacing between tubes and use 12 or 16 tubes.

 

What ballast are you using?

I am on my second set of nonworking lamps none of which will ignite on any of my little magnetic ballasts.

The eBay seller is pleading ignorance, sincerely I fear, as to ballast compatibility!

[enricosavazzi]

eBay n. 161706722820. These are designed for different bulb sockets and have six output contacts for each pair of bulbs. I connected two of the 9W bulbs in series and connected the remaining two contacts of the bulbs to the outermost contacts of the ballast (i.e. output contacts 2,3, 4 and 5 of the ballast in each strip of six output contacts are not used). It was a gamble, but works fine. I did not try other ballasts except one built into an Ikea table lamp that accepts the 9W bulb (it also works fine).

[JCDowdy]

Ah, that explains why they won't light.

None of the ~dozen PL9 devices I have in the lab are equipped with electronic ballasts.

I will need to find a 120V electronic ballast.

Thanks