IR video problem solved?

[Alastair Traill]

Over the last few weeks I have been doing some infrared video recording of small nocturnal marsupials. Initially the results were acceptable but then started to deteriorate. At first I thought it was the illumination or my ability to focus. I then realized that the problem was related to the subject distance and that the greater the distance the worse the image. I had started with fairly close shots and then proceeded to more distant shots of a metre or two. I was using a vari-focal lens that was corrected for IR work. It does not have any focus, aperture or zoom markings, it just has arrows for guidance. In use you can set the focus ring and then adjust the zoom until the image is focused or set the zoom ring and then focus. The lack of a conventional infinity stop turned out to be a bit of a trap. Eventually I tried a prime 8 mm lens and found that when the lens was set to infinity my focus point was close to 30cm from the lens front. Using the lens formula X= f squared on d, I calculated that the back focus distance was 0.0085” too long. The apparent increase in back focal distance was caused by the necessary removal of the behind the lens IR cut glass filter for this IR illuminated project. Had I tried my 8mm lens first I would have realised what the problem was straight away. I was thrown by the lack of conventional infinity point of the vari-focal lens, despite the effective 0.0085” increase of back focal distance I could still find a focus point, however it was not the best focus the lens could achieve. I shortened the lens mount by the calculated amount which improved my results considerably. A side observation is that the IR corrected vari-focal lens produced a brighter image than the uncorrected prime lens.

The closer screen shot were taken before the lens mount shortening and the more distant shots are roughly paired with the sharper shots being post lens mount shortening. The animal is the Brush-tailed phascogale (or tuan) carnivorous marsupial with a couple of dubious claims to fame. Its conservation status listing is ‘vulnerable’ and the male is the largest species that dies after mating.

http://www.dvinfo.net/forum/images/a...attach/jpg.gif

[Bob Hart]

Alistair.


You are probably already aware of the issue of mixed IR and visible light sources. Depending upon the relative power of the IR and visible spectrum light sources, you may have soft and muddy results happening inconsistently and unpredictably. You could try killing off visible light sources with ND filters. Do not of course use IRND filters.

[Jim Michael]

Would a cutoff filter make more sense for removing visible spectrum?

[Alastair Traill]

Thanks Bob and thanks Jim,

I have not tried cutting out visible light. I have some clips taken at dusk and they do not seem to be any worse than those taken when it is dark. My IR source puts out a very faint pink glow but I am sure that is a not problem. When I try a 5 second exposure using a GH5 and this pink glow for illumination it does not produce an image.

My main concern is whether or not I have the optimum back focus distance. In film days it was possible to remove the lens and fit a steel plate at the film plane and use a mechanical way of measuring the back focal distance e.g. by means of a depth micrometer. If the distance was acceptable the steel plate would be replaced by a front surfaced mirror and an autocollimator used to check the image produced by the lens. If all was well the image would be in focus at the infinity mark on the lens. No such luck with a video camera. I have a collection of ‘C’ mount lenses ranging from vintage to current models and their back focal distances all differ as do the flange depths of the cameras I use them on. If you explore the lens formula that I mentioned you will notice that the shorter focal length the more precisely the flange depth has to be set. For example if the flange distance is 0.1 mm too long a 100mm lens would not focus past 100 m, a 20mm lens’s would not focus past 4m and a 5mm lens would not focus past 250mm. My IR corrected Vari-focal lens ranges from 4.5 to 10 mm.

It is a few years since I last had the chance to record the activities of this species. In that time the availability and suitability of IR illumination has improved greatly and the same applies for cameras. With the new gear my results are considerably improved however I would appreciate seeing frame grabs or clips from others involved with similar projects.

[Vishal Jadhav]

Hi,

Whats the camera you have used for the IR video, is it GH5 ?

[Alastair Traill]

Hi Vishal,

An essential feature of a camera used for Infrared (IR) work is either the absence of an IR cut filter or the ability to remove it. It is also desirable to have normal control of sensitivity, aperture and shutter speed. There are cameras that have controls to remove the IR cut filter but they always seem to then set the sensitivity to the maximum, fully open the aperture and make the shutter speed as slow as possible. The reason manufacturers do this is said to be to eliminate any possibility of using the camera in sunlight as certain clothing materials allow IR to pass. As a matter of curiosity I have tested my own clothing but have not found any that IR passes to a significant extent.

I do not use a GH5 as reversibly removing the IR filter is probably not practicable although I remember reading on this forum that a user has a converted GH4. I use GoPros that have been converted to ‘C’ mount by Backbone.ca. I have a converted 3, a 4 and a 5 – all are black models. The lens that has produced my best results is an IR corrected 4.5 – 10 mm Varifocal. As mentioned in my earlier posts I have had problems setting the back focal distance. My guess is that is now about right at least on my GoPro 4+ black. Since starting this thread I have changed my field monitor to a Ninja Inferno, partly to try higher frame rates but also because it has Pre-roll and an interesting focus assist settings where any sharp edges are coloured (your choice of colour) and the colour is removed from the rest of the frame. It works well. By using a recording video assist, remote focus on the GoPro and a motorized tripod head I can work at a distance from the camera and subject.

[Vishal Jadhav]

Thanks Alastair,

Gopro is something i have used but the utility you have put it to takes it to a all new level.

I am currently struggling to come up with a camera to film at night and that doesnt break the bank. However have not been able to make any videos except for my camera traps.

[Doug Jensen]

Have you looked at the Sony PXW-Z90? The 4K infrared capabilities are amazing if you use an external light. I use a $40 battery operated Polaroid light that lights things up like daytime out to about 20'. Very easy and you also have a great camcorder for regular shooting, too, at the flip of a switch. If I was really serious about nighttime shooting I'd try setting up three of the Polaroids in an area for traditional 3-point lighting. I'll bet it would look awesome for the type of subject matter in your first post. I think I just talked myself into giving it a try later this summer when I've got spare time.

[Alastair Traill]

Thanks Doug for your suggestion,

I have had a quick look at the PXW-Z90 specifications and User’s Manual. There is a warning not to use Nightshot in bright conditions as ‘doing so may damage the camera’. This suggests to me that the aperture is set to full, sensitivity is set to maximum and shutter speed is set to longest duration possible when Nightshot is selected.

I will be very interested to see what IR results you come up with your PXW-Z90 and polaroid illuminators.

I will probably continue with ‘C’ mount GoPros as I find them very useful for the plants, fungi and animals I now work with. I have spent a lot of time developing gadgets to assist including a mini fluid-damped boom set-up mostly used for low level work. The camera is mounted on a motorised slide for focusing. This has been very useful in the field and for bench top applications that include time-lapse clips.
For anyone interested there is a photo of my current IR illuminator, my GoPro set-up below and a couple of my mini boom.http://www.dvinfo.net/forum/images/attach/jpg.gif
http://www.dvinfo.net/forum/images/a...attach/jpg.gif

[Doug Jensen]

Yes, the Z90 must be set to Full Auto to use Nightshot, but the camera is smart enough to only add as much gain as is necessary to get a correct exposure after the aperture has been maxed out. Shutter speed is just the normal shutter speed, it does not go into a slow-shutter mode. Nighthshot on the Z90 is super simple to setup and it works very well.

What is your IR illuminator? It looks quite powerful. I could light up a whole forest with that thing and the Z90!

[Alastair Traill]

Thanks again Doug,

It sounds as though the Full Auto setting on the Z90 is what I would have expected for a Full Auto setting. Does this mean that all Nightshot does is remove the IR cut filter? With all this control is the warning not to use Nightshot in bright conditions unnecessary? The warning certainly put me off buying a Z90 for IR use.

A common IR clip for me includes a lot of black background and a relatively small lit area. How would the Z90 cope with that? The only auto exposure camera I have used for IR work would overexpose. Can the Z90 be set to expose for a user-selected portion of the frame?

I got away to a bad start with IR illumination. When I started there were only individual IR emitting diodes on offer, then small IR led panels became available. So after trying increasing numbers of diodes I bought a panel but it was not enough so to double the output I bought another, that was still not enough so to double the output again I had to buy 2 more. Then I discovered that for less money I could buy more powerful IR illuminators for outdoor security areas so I purchased four. To cut weight I removed the panels from their housings and set them up in parallel on the simple frame shown in the photo. I use a variable power supply that gives me wide control of the brightness. The supply is fitted with an ammeter. I find the ammeter useful for trouble-shooting as well as calculating power consumption that can be up to ~ 30 watts. I do not know how well the spectral output of the lamps matches the spectral sensitivity of the GoPro but it performs better than an Orion security camera.
For my current project the IR output is about right, I can stop down the lens from its maximum of f1.8. As mentioned earlier the lens has no graduations but I would like to think I am using f4 to f5.6.

[Bo Skelmose]

Hi
I use two converted GH4 and the IR filter have been changed with one of glass so the focus on the lens mach with IR light wave length. I also use some Yi4k+ and the M12 lens can just be turned into focus. The latest I have tried are the new AIDA 6g-200 cmount camera. I use them with RS485 and can control most settings from my PC. I am struggeling with the controls as the AIDA software have not been updated to the new models. I cannot always control the shutter/gain setting without it sometimes think it should do it automatic. But very happy for them. For tele work I use the panasonic AU-EVA1 that do have a switchable IR filter...

[Alastair Traill]

Hi Bo,
I guess that your converted GH4 cameras are now dedicated IR cameras and that it is not a simple matter to replace the IR cut filters. Is that correct? I will try and find if I can get plain glass ‘filters’ with the right combination of refractive index, thickness and diameter to use in my GoPros. Are you using IR corrected lenses? Could you post any frame grabs?

I am now finding that if all goes well I am getting results that please at least myself and they are certainly way ahead of what I could achieve a few years ago. However I have found another trap when using my IR corrected vari-focal lens. The lens has 3 control rings that all have the same feel and are difficult to distinguish in the dark. They are also close together which increases the difficulty. This means it is easy to turn the wrong ring. The problem can be prevented by tightening a small screw that prevents accidental rotation. However even when carefully tightening the screw on the focus ring or the focal length ring there is a high risk of changing the focus. The issue became apparent when I was checking the lens on my autocollimator. The autocollimator provides an illuminated graduated test image set at the focal point of a converging lens and so appears to be at infinity. It is a useful and sensitive device.

[Bob Hart]

Alistair.


In regard to IR illumination, security outfits used to sell small lamps with about 15 IR LED emitters inside of them. It worked well on a CCD B/W 1/3" CS-Mount security camera which had some IR sensitivity. It was good only out to about 2 metres for good contrast. I think they only sell combined camera/LED modules now.

There was a filmie in Kalgoorlie a few years back who I understand at some point worked on one of the earlier reality shows like Castaway or Survivor. For the night-vision stuff they were then doing I understand, maybe incorrectly that he made up a powerful panel of a hundred or so IR emitting LEDS.

Ebay has some smaller IR LED lamps, some like the one I mentioned and others built into garage or workshop lamp housings which are styled like the older 500watt halogen bulb lamps. If you can generate enough IR light, then you may be able to close the iris on your lenses and hopefully achieve a deeper depth-of-field. How that works with IR light I have never tested for.

For finding the best forward offset for your lenses, you might be able to use very thin neoprene "O" rings as compressable shims around the 0.7mm pitch threaded section for the C-Mount on back of the lens. Slide a fine piece of fishing line through the "O" ring as you stretch it onto the threaded shoulder on the C-Mount lens then screw the lens in against the "O" ring until you find the sweet position.

You may be able to use the fishing line to extract the "O" ring, leaving the lens in place and then use engineer's or mechanic's feeler gauges to measure the gap between back of lens and front of mount flange faces and cut a shim washer of required thickness. Alternatively, count the number of turns of unscrewing the lens until it pulls free, remove the "O" ring, then screw the lens back into the mount for the same number of turns, then measure the gap for making a metal shim washer.

Or you could take it easy and just use the thin "O"ring as is to hold the lens firm in the C-Mount threads forward of normal flange face contact. You would probably have to use a trace of silicone grease on the "O" ring so it does not grip and tear apart and may release more easily when being teased out of the gap with the fishing line. Chances are if the ring has to be stretched too wide to roll it onto the threads, it may extrude into the threads and make them tight. Keep the silicone off the lens glass because it is dreadful stuff to clean off lens coatings.

Just a few thoughts.

[Alastair Traill]

This thread resulted from focussing issues when using infrared on ‘C’ mount modified GoPros that I now have sorted at least when using the models specified below. With the onset of warmer weather my sugar glider and phascogale subjects are rising later and are less interested in my food offerings so it is becoming harder to obtain footage. The phascogales (tuans) have a conservation status of vulnerable and the unenviable characteristic of being the largest species that has an obligatory die-off for the males after mating. Below are some comments and pictures of equipment used that may be of interest.


Flange depth on Backbone GoPros: -
I now have a clearer idea about flange distance on my Backbone modified GoPros: -
1. The GoPro 5 black requires shims (0.010”) to achieve the correct CS flange distance for visible light use. To use for IR work the shims and IR cut filter have to be removed.
2. For the GoPro 4 black no shims are required for visible light but for IR use the IR cut filter has to be removed PLUS 0.010” has to be removed from the mount. I was unaware of the flange depth differences between the two models however my observations have been confirmed by Backbone. I found out by making myself a ‘dummy camera’ with the correct flange depth and by adjusting my lenses to match by machining material from their CS mount adaptors. The result is that all my lenses can now be set to infinity for infinity shots with some degree of success. I also use a grub screw to lock the now individualised CS adapters to the correct ‘C’ mount lens. An irritating feature of ‘C’ mount lenses using ‘CS’ adaptors is that the lens may become unscrewed from its mount when being removed from the camera, the grub screw prevents this.
3. Although I have collimated all my lenses that have infinity stops I still do not know how to ensure that my vari-focal lenses that are also IR corrected are set to their optimum. Any ideas would be welcome.

Ninja Inferno: -
During this project I purchased an Atomos Ninja Inferno. As mentioned before it has a very useful Pre-roll facility, Time lapse and a very good set of Focus Assist features etc. However it cannot be started and stopped with a simple remote control such as the Prolanc I used on my BMD VA 4k. The only remote that I am aware of performs extra roles, requires 2 batteries and is rather expensive if you only want the start/stop/ function. Without a remote I found it difficult to keep an eye on the Ninja screen, an eye on my subjects (that may only be a faint silhouette) AND keep a finger close to the touch screen start logo. It was also remarkably tiring to do so. As a workaround I made a ‘finger guide’ consisting of a piece of aluminium shaped to support a finger and mounted about 8mm from the screen. It has now evolved to the stage when the NI is mounted close to a platform thus providing support for a hand also.

Illumination: -
I am basically happy with my IR illuminator except that I get a hotspot if I work with the lights less than about 2 metres away. Voltage to the IR illuminators can be adjusted in 0.1volt steps and that give me very fine control. A trap I tend to fall into is that I set up my gear in the absence of the animals and adjust IR intensity using false colours on the Ninja Inferno. The problem is that the animals have white fur on their feet that reflect IR very strongly.

Tripod Extender: -
Another item used in this project that may be of interest is my ‘tripod extender’ developed primarily to get a camera very close to the ceiling of our verandah for the recording the local paper wasps nesting activity. Their breeding season has just started. Depending on the legs used it stands up to 7’ tall plus regular tripod, however ~ 7’ is a good maximum combination. The centre of the extender is a tube machined at both ends to accept a 4” ball or a 3” ball using an adapter. It will also accept a flat plate. I can mount a boom set-up using either end of this tube. If mounted on top I can attain heights of ~ 14’ if using my longest boom. It is a fairly simple and rugged device that could also be useful in a wet-land.

Cable tidy: -
A final item is finally a simple, cheap and practical way to keep cables tidy for storage. It consists of a short length of double sided Velcro (preferably white that can be written on) stapled around the cable so that it does not get lost (see pics).

I am having trouble loading pictures.

http://www.dvinfo.net/forum/attachme...attach/jpg.gif