5
\$\begingroup\$

My Problem ist actually simple: I am trying to take photos of flat surface items (large flags, large maps, antique rugs and so on) in a room that does not offer enough height.

I have been "researching" this matter for quite some time now but haven't found a solution for this yet. I am an absolute amateur in regards to photography, I am however quite capable in IT matters (programming, scripting).

The objects are too large or often too fragile to mount them on a wall, so taking the photo from the ceiling would be my approach. A "vertical setup" would be advantageous for the large quantity of items as they could just be laid out flat one piece at a time.

The measurements and forms vary but let's say that I need to capture about 5.50 x 4.50 meters from a maximum distance of 2.30 meters.

I do not need any depth/perspective. Just an orthogonal and planar surface image. Field of view, Zoom factor and focus may remain fixed as I would crop the items in Photoshop and due to the fixed nature of the room height.

Detail/Resolution need not be as high as in art photography. But color and form have to be as it were a single photo.

My ideal result would be something like what a flatbed scanner would create. Not mechanically of course but in terms of capturing.

I know about fisheye lenses, however I fear that correcting the distortion would not work on larger items or not even work at that distance and these dimensions. Details at the edge of the object cannot turn out to be blurry. Looking into panorama photography I've only found solutions for rotating cameras. I'd much more prefer a kind of matrix of multiple cameras. I reckon 2 or 3 cameras would suffice. Syncing multiple DSLRs or even GoPros seems to be a popular thing nowadays and there is lots of material to read, but everything i found concentrates on either 3D capturing, timelaps/bullet-effect or 360° videos. I have not been able to find a solution for an array of cameras just "scanning" or stitching together a surface. Except for satellite/drone image stitching which would be exactly what I wanted, but in one synced up shot and with all the automation which this fixed setup would theoretically allow.

My ideal setup would be as fixed as possible once calibrated. Actually the camera(s) would be dedicated and should not have to be touched at all and remain connected to a power source / cable or wifi at all time to avoid movement. I have looked into photo stitching with Hugin, but I am unsure how much automation would be possible. Also I am unsure which cameras (dslr or not) would be right (centralized trigger).

Let's assume that I would be willing to buy all equipment necessary as long as it is consumer grade price level. Lighting would be installed in the same step.

I have read through lots of answers here on stackexchange and elsewhere. There is one interesting comment here https://photo.stackexchange.com/a/57219 but the author (Unapiedra) does not describe the actual implementation of the solution sufficiently.

Thank you in advance for any helpful answers or pointers in the right direction.

Marc

\$\endgroup\$
4
  • \$\begingroup\$ In my opinion the implementation you linked is described pretty well. Probably you need to be more specific on what aditional detail you need. But that is diferent than ask someone to prepare a manual for you. ;o) \$\endgroup\$
    – Rafael
    Mar 26, 2016 at 22:12
  • \$\begingroup\$ Rafael, you are right. I wouldnt go so far as to say I need the step by step manual for the suggestion which was posted, but I'd really need details on what cameras and lenses would qualify for these dimensions and for the remote control. Also whether the automation would really have a chance of working as suggested in the answer. \$\endgroup\$
    – Marc
    Mar 26, 2016 at 22:18
  • \$\begingroup\$ as for Hugin, it is possible to create a stitching project and then use it repeatedly from command line on new incoming images - might work if your camera(s) keep the same view all the time \$\endgroup\$
    – szulat
    Mar 27, 2016 at 0:19
  • \$\begingroup\$ All provided answers are very good. Thank you all. I will probably end up using combined insight of all your answers. I will firstly attempt the ceiling mounted slider which was suggested by Hamish to get the most out of budget and detail. It is also scalable and flexible, which i find nice. I had not known about the mechanical automation possibilities (as well as possibility to trigger the cameras) with an arduino/pi until i researched the non-automated suggestion. Also allows for improvements in development due to the number of steps & parts. Not everything needs to be right upon purchase. \$\endgroup\$
    – Marc
    Apr 1, 2016 at 12:02

3 Answers 3

3
\$\begingroup\$

Depending on your output image resolution requirement, you have two options as I see it. The first option being the simplest and quickest but not necessarily the cheapest:

  1. Use a dSLR with a wide lens (probably a 24mm on full frame or something approaching a 16mm in DX format) and use the specific lens correction option for that particular lens to correct any distortion in post. Mount the camera on a boom arm with a suitably rigid tripod head on it so as to keep the film plane parallel to the subject. Swing the arm up to ceiling height and use a remote trigger to take the shot. Tethering or WiFi will make file transfer a lot simpler and will allow you to leave the camera in place while you swap out each subject. If you have a high resolution camera and decent lens, you will get large, reasonable quality images this way.

  2. If extreme resolution and accuracy is required (such as for larger than life reproductions or archival detail levels) you will need to build a sliding horizontal camera rail system that will allow you to take a matrix of images from exactly the same plane and then stitch them together in a panorama. You can do this in such a way that each panorama consists of just 3 linear images in portrait orientation (giving you files with approximately 2.8x the camera's resolution), or you can shoot from a lower height (or longer lens) and take a dozen or more images in a matrix/grid pattern (giving you files with tens of times the camera's resolution). Using this method you will want a lens with minimal distortion to simplify the stitching; a cheap 50mm lens typically is perfect unless you need to focus in the macro range. The rail system doesn't have to be complicated or expensive, but it does have to be rigid and square. Probably aluminium angles and a home made camera brace would work. You could probably build something suitable with little more than a drill, rivet gun, hack saw, file, duct tape and suitable length of aluminium angles; perhaps $NZ150.00 all up (about $US80.00). You could also get away with a cheaper camera with this option (though not recommended), since each individual image represents a smaller portion of the finished image.

If I was doing this as a big project and I wanted the level of detail that you indicate, I would go for option 2 and probably shoot it with a 60mm Micro Nikkor on full frame. I would shoot the entire thing under the best lighting I could manage (minimum of two studio strobes and probably big soft boxes or brollies), and I would shoot entirely with manual settings, using an aperture of f/8-16 depending on subject distance. This would likely require about 6-9 shots per subject and would give very sharp high resolution images with little to no distortion correction required and quite simple post processing. I would script all of the post processing using actions in Photoshop to ensure a ruthless level of consistency.

Edit: I realise option 2 does require manual positioning of the camera from frame to frame, but this can be done very quickly in a routine pattern for consistent results. If full automation is required however, you can get similar rail kits that are designed to drive computer aided cutting tools, so programming it to position a camera instead of a router wouldn't be too hard. You'd probably still have to fire the camera manually unless you wish to rewire a remote cable release and get the program to fire the shutter when in position (by shorting the two required terminals).

\$\endgroup\$
7
  • \$\begingroup\$ "Use a dSLR with a wide lens " - actually, the best way would be Sony NEX and Samyan 10mm F2.8 - which would give sufficient FoV, and the combo is not too expensive. \$\endgroup\$ Mar 27, 2016 at 11:58
  • \$\begingroup\$ It is also rectilinear, or at least easily correctable. \$\endgroup\$ Mar 27, 2016 at 12:23
  • \$\begingroup\$ Interesting, good to know. I assume the 10mm Samyang still has enough distortion to require correction though. Does it have any lens correction options in camera or software? \$\endgroup\$
    – HamishKL
    Mar 27, 2016 at 20:21
  • \$\begingroup\$ @Pinhollow Euri. Ah I see your linked example image shows some simple barrel distortion, so yes, the images would still need to be corrected - as would those from every single wide angle lens I've ever seen. \$\endgroup\$
    – HamishKL
    Mar 27, 2016 at 20:57
  • \$\begingroup\$ Samyang is a widely recognized MFG, there are profiles for Samyang in Adoe software (but not a profile for Sony E version unfortunately). Of course one would better try it to see if the result is fine - there are many samples in internet though. \$\endgroup\$ Mar 28, 2016 at 5:19
2
\$\begingroup\$

Well this is a bit complicated, but let us asume you are willing to really work on this, for example a special room on a museum or a collector's one.

Some ideas.

The cameras

Focal length and crop factor.

Initial notes: If you choose one with a 50mm lens, on a full sensor camera at a distance of 2m you can photograph an area of more or less 1.2 x .85 mts. (I did a really quick test)

On a crop factor camera, much cheaper ones, lets say with a factor of 1.5 you need a 35 mm lens.

You can reduce the number of images using a wider lens. For example 1.7 x 1.1 with a 24 mm lens on a croped factor.

But you need a "safe" zone, for the correction of the lens and the overlaping of the photos. I calculate that you need like 20 cm each side. That gives you about 1.3 x .7 mts... 1.5 x .9 becouse the safe zone is "shared" on consecutive frames.

Use a fixed focal lens, not a zoom one.

Remote controll and wifi

Even some old Eos cameras can be trigged remotely, for example https://www.google.com/search?q=canon+remote+trigger

You can use a wifi camera, or use a wifi SD card. But there is a chance you have a little mess reciving all data at once, so probably you can trigger one by one, using diferent channels on the triggers.

One RF-603 control can have like 7 channels, so probably that is a limitation.

Put them in a grid on the celling. You can for example make an array of 2x3 cameras (6 total) 2 colums, 3 rows. And after that decide what focal length you can use.

A robotic panorama head

This can give you a realy high resolution image. Look for a head that is programable regarding the number of photos and the angles to move.

https://www.google.com.mx/search?q=robotic+panohead

This would be a lot cheaper solution.

The lights

If you have non reflective surfaces, you can pute difuse lights on all the celling. But if the surfaces are reflective you probably need to go for the array method or the lights on the wall, but you will have a lot of diference across the area.

The software

If you take the photos on raw, you can prepare some actions either on photoshop or lightroom to correct the distorsions of the lens and have a flat projection.

\$\endgroup\$
2
  • \$\begingroup\$ Rafael, thank you for taking the time to answer my question. The surfaces are non reflecting. I find the panorama head idea interesting. By taking for example three photos (though with increasing perspective distortion) from a fixed point at the ceiling, the resolution will be higher than by just using one very wide angle lens from above. If the panorma solves the "length problem", do you think the width demand would be fully addressed by the capability of the wide angle lens? \$\endgroup\$
    – Marc
    Mar 29, 2016 at 10:29
  • \$\begingroup\$ Do not use a too wide angle lens. A panoramic head not only makes a "pan", but also the "tilt". The longer the focal length the higher the the resolution of resulting image. \$\endgroup\$
    – Rafael
    Mar 29, 2016 at 13:22
2
\$\begingroup\$

I do this sort of photography all the time and find that as long as you have good software capable of stitching, then the least hassle factor way is surely to split your subject into sections [1/2's or 1/4's or 1/6's] and let the software do the rest. Keep the lighting consistent, mount the camera from above if possible and slide the map, rug or whatever, through the legs of the tripod making sure there is some overlap on each image you take. Lightroom or PS will do the stitching reasonably quickly but you will of course end up with fairly large files. Good luck

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.