Grants:Project/Rapid/dllu/Line Scan Camera

Choose one or more of the following goals. You can add or delete goals as needed. You can add details to each goal. For example, if you want to add content on a specific subject.

Add unique photos which are impossible to create without a line scan camera
Expand prose on English Wikipedia of relevant articles
Engage and inspire other Commons photographers and Wikipedia editors
Develop open-source software for removing motion distortion from strip photography

Project Plan

Activities

Tell us how you'll carry out your project. Be sure to answer the following questions:
1. What content gaps will you be filling with the equipment?

All existing examples of strip photography on Wikimedia projects are of low resolution and/or are decades old.

The current Wikipedia articles on en:Strip photography, en:Slit-scan photography, en:Rollout photography and en:Photo finish are sparse and have very few examples produced with modern equipment. The lack of images hampers understanding of an imaging technique. Meanwhile, commons:Category:Photo finish is nearly devoid of Wikimedian-produced content, and half of the photos therein are not strip photos. Strip photography is an important technology that is used for artistic expression, industrial quality assurance, and for determining the winner in races. As opportunities arise, I will also be enriching Wikimedia Commons and English Wikipedia with the many other applications of a line scan camera, detailed in section 5.
2. What activities will you organize in the next 6 months to create content and use it on Wikimedia projects?

Equipped with the line scan camera, I will conduct numerous photo walks around San Francisco, at least twice per month, to create artistic images of the historic streetcars and other vehicles in the area. In May, I will travel to Xi'an, China, as well as Shanghai, for a yet undetermined period of time to take further images. Later in the year, I will also travel to Vancouver, BC for an undetermined period of time for additional photographic opportunities.
3. How will the equipment be shared/tracked/cared for by members of your community?

Upon request, the line scan camera will be lent out to nearby Wikimedians on a first-come-first-serve basis. The location of the equipment will be tracked using an online spreadsheet. When not in use, it will be stored at the residence of Daniel Lu.
4. How long will the equipment be used, and by whom?

The equipment is anticipated to be used for at least 5 years, primarily by me, and by whoever requests to borrow it.
5. Is there anything else you want to tell us about this project?

A line scan camera is a high-speed camera which only takes 1 row (or column) of pixels at a time. It can be used for many purposes, most of which cannot be replicated by conventional camera:

To take pictures of fast-moving vehicles for a high-resolution profile view.
To take 360 degree images of arbitrary small household items and artworks, e.g. File:Maya vase.jpg.
To illustrate the movement of various mechanisms and physical phenomena.
To create a very high resolution panorama without obvious discontinuities by panning the camera (i.e. as a en:Rotating line camera).
To capture very long objects, such as trains, streets, buildings, artworks, and so forth.
Archival of film.

As such, the photos created from this project will be of a very diverse nature, covering many different subjects to optimally make use of the many applications of this technology.

High quality line scan cameras are typically not designed for portable, outdoor use by consumers. Custom mounts and electronics (or a laptop computer) may be needed to use the camera. As a mechatronic engineer by training, I am capable of designing and constructing any necessary mounts (cf. https://lawrence.lu/engineering/ for some past projects from university). Note that the Necta camera chosen is a recent model that uses USB3, which is much easier to interface with modern portable laptops. The vast majority of other line scan cameras use Camera Link instead of USB3, which require their own power supplies and a PCI Express board mounted in a desktop computer.

If the camera is not stationary, or if the subject is not moving at a constant speed, distortions may occur. Moreover, artificial lighting tends to flicker in a manner that may affect the photo. Software corrections will be needed. As I am proficient in non-rigid image registration algorithms (cf. en:Point set registration, an article I started), I will be able to develop the necessary software for performing said corrections. Any software created for this purpose will be licensed under the appropriate open-source license.

Many hobbyist projects have used a converted flatbed scanner or film; however, although they are cheap, such cameras have poor scanning rate and are not suitable for fast-moving objects. The line rate of a typical flatbed scanner is about two orders of magnitudes worse than that of the requested line scan camera. A video camera, which can also simulate strip photography, has a typical frame rate that is yet another magnitude slower than that of the flat bed scanner.

Here is a list of some examples of slit scan photography by other photographers:

Elm's scans of trains with a DIY CCD camera: [1]
Panoramas of riversides in the Danube project: [2]
Adam Magyar's scans of subway trains and urban traffic [3] [4]. Note that his higher resolution projects used an industrial line scan camera, similar to the one proposed here ('Magyar made “Stainless” using an industrial camera “normally used at assembly lines for mass production,” he says.')
Jay Mark Johnson's abstract photography [5]

Impact

How will you know if the project is successful and you've met your goals? Please include the following targets and feel free to add more:

Number of photos uploaded: 400 over the next 12 months, of which
Number of quality images: 300
Number of featured pictures: 10
Number of photos used on Wikimedia projects: 200 (50%)
Number of articles created/improved: 50

Resources

What resources do you have? Include information on who is organizing the project, what they will do, and if you will receive support from anywhere else (in-kind donations or additional funding).

I, Daniel Lu, am equipped with conventional photographic equipment such as:

Manfrotto tripod
Two laptop computers and a powerful desktop computer with calibrated sRGB displays
Sony a7R camera
Assortment of lenses covering a focal length of 15mm to 300mm

Several of my lenses can be readily adapted to the desired line scan camera with little or no cost. Any additional equipment, such as extra lenses, custom mounts, a portable embedded computer attached to the camera, and so forth will be self-funded.

The project will be entirely conducted by myself. I have 11 featured photos on Wikimedia Commons and many dozens of quality images, as well as started or greatly expanded several technical articles on English Wikipedia.

What resources do you need? For your funding request, list bullet points for each expense and include a total amount.

NECTA N4K2-7C color line-scan camera [6], list price €1859 including F-mount adapter. (USD $1982 at current exchange rate)
Importing camera equipment to the USA is duty free
Shipping, insurance, and other misc fees: at least $18

Total: $2000

Endorsements

members are encouraged to endorse your project request here!

Support. This is a well-thought-out, worthy project. Also dllu is a great photographer, and the fact that the camera would be at the service of other Wikimedians as well is a great thing that could add a lot of value to this project. -- Ikan Kekek (talk) 21:49, 8 February 2017 (UTC)
Oppose I'm afraid I'm not convinced about this project. The results, taken by professionals, appear here by Jay Mark Johnson, in this long article and in this article. But Jay Mark Johnson's professional images use a "$85,000 rotating slit-scan camera". Another photographer, Adam Magyar is discussed here and his slit camera is largely self-made at low cost ($50). The industrial device proposed for this grant has resolution "2x2048" and I'm guessing to achieve colour there is a Bayer array over that which reduces the colour resolution by a factor of 2 in each direction, meaning a line of 1024 colour pixels. A 1:4 panorama is only 4 megapixels. I don't know what the image sharpness may be like, but it might possibly be no better than the output from a baby monitor. Do we have any examples of colour photos produced by this camera? I suspect not as it is aimed at industrial applications. Looking at the photos I've found online, some are certainly beautiful artworks but they are all quite distorted in some way. This is fine for judging a photo finish or a work of fine abstract art, but not for creating something likely to be generally useful on an educational project. If it merely produce examples of images created by a line scan camera, this is an expensive and time-consuming project. It would be easier and cheaper to ask some professional who is using such a camera to provide a freely licensed image -- they must have taken some that didn't have any commercial value and could be given away.

Secondly, I think the project goals are not reasonable. It isn't indicated whether the number of photos listed are to be made by this device, or in general by Dllu. For information, Dllu has uploaded about 600 images to Commons, 153 quality images and 11 featured images, over a period of about 2 years. Comparing the goal for 12 months, I do not see this as achievable for standard photography unless Dllu greatly increases the number and quality of his contributions. The percentage used on Wikimedia projects is also IMO ambitious. But if we assume that Dllu would continue taking standard photos even without the grant, how many more would be taken with this equipment? There is a significant risk the project fails and no images are produced. Another risk is that images are created but none are of good enough quality to meet the goals. I personally think that an optimistic result would be a small quantity of interesting photos. There is a chance that developing this camera takes so much of Dllu's time, that he contributes fewer images to Commons.

My suggestion is that Dllu reaches out to Adam Magyar to ask more details about his self-made camera, and then attempts a similar low-cost project. That could indeed be a fun project. I think for WMF, their educational mission is far more likely to succeed buy purchasing regular photographic lenses and equipment. For example, a macro lens (to photograph insects or small objects); or a tilt-shift lens (for great product photos with superb front-to-back sharpness); or a portrait lens if someone wanted to get serious about taking portraits for Commons; or a panoramic head for creating 360-degree panoramas. In the past, such equipment has been funded by WMF or a crowdfunding campaign, and has gone on to produce many highly educational photos. -- Colin (talk) 13:43, 9 February 2017 (UTC)

Thank you Colin for your honest comments. I appreciate that you've taken the time to carefully read my proposal.

Regarding Adam Magyar, the Medium article states that he created two cameras: the first one, using a low-cost flatbed scanner, was used in his project "Urban flow" and the second one, using an industrial line scan camera similar to the one I'm hoping to purchase, was used in his project "Stainless". He was dissatisfied with the poor scanning rate and low resolution of the cheap project.
The camera chosen actually has 2×4096. One of the lines has 4096 green pixels and the other has 2048 each of red and blue pixels, alternating. This results in 100% coverage by green pixels and 50% coverage by red and blue pixels, which is considerably better than the typical Bayer filter used in most cameras which has 50% green, 25% red, and 25% blue. Moreover, at 4096 pixels, this is comparable to modern cameras such as the Nikon D810 and Sony a7R which produces pictures of around 4900 pixels in height.
Regarding distortion, it is possible to produce pictures without distortion. For example, Adam Magyar's "Stainless" project has distortion-free pictures of subway trains which I think are very educational.
Regarding number of photos taken, I will have much more opportunity to travel this year than in any previous year and I hope to increase my photo output.

I remain confident that this project will produce photos that cannot be created by the millions of cameras already being used to create photos. Thanks again, and let me know if you have any other questions. Dllu (talk) 21:50, 9 February 2017 (UTC)

Thanks for the extra info about the sensor. I think you should ping more people (such as a message on talk FPC) to get varied opinions and possibly useful suggestions. I think, though, in terms of this grant proposal, you should indicate how many images/QI/FP/etc will be created/achieved with this speciality camera. It isn't relevant to the grant how many photos you will take/upload in general with your existing gear. Also, I think that such speciality equipment is unlikely to be shared by other photographers. This is true to some degree of all camera equipment unless one lives in a city with many Commons photographers who share the same brand of gear. Wrt subway photos, as Magyar comments in one article, taking photos on the subway isn't easy as tripods are generally not allowed and security are often suspicious of people taking photos with high-tech equipment. When I consider my own speciality lenses (a fisheye lens and my new 500mm lens) they are some of my favourite lenses, capable of taking images no other can, but also of limited applicability and acceptance. I wouldn't expect more than a few featured pictures each year with my special lenses, and they are still standard consumer optics one can buy in a shop. -- Colin (talk) 21:45, 12 February 2017 (UTC)

Hi Colin,

For QI, one way to achieve a large quantity of quality images is to set up the camera next to a busy street for a few hours. Each of the many different types of cars that passes by should be captured in sufficient technical quality to be considered a QI.
For train photos, there are many places where you can set up the camera to shoot trains outdoors without being in a station. Many subway systems have elevated sections that are visible from the outside; this is especially true of the two cities I will be spending the most time in (San Francisco's BART and Vancouver's Skytrain and Canada Line). For example, see Elm's 33200 x 1024 line scan photo of a JR E233 series train which I've also linked above. There's no need to risk arrest like Magyar did.
Regarding specialty equipment, you have a good point. The camera I've chosen has a standard Nikon F-mount, so it can accept any Nikon manual lens (APS-C or full frame), and it has a standard USB 3 interface that you can plug into any modern computer. It's easier to use than line scan cameras from a few years ago, which all use Camera Link, requiring a desktop computer with special hardware. user:Pine and I have been reaching out to gauge interest from WMNYC and WMDC, the two largest chapters in the US. I will also reach out to more FPC photographers and ask for their opinions.

Thanks again! Dllu (talk) 23:32, 12 February 2017 (UTC)

However, once you go outside, in good light, a conventional camera can capture a sharp image of a moving train or car. It won't provide the strange isolation-from-background that this camera would, though whether that's an advantage is debatable. The backgrounds I've seen from this camera seem to either be a striped effect or black. And even with a conventional camera, one can pan the camera with a moving vehicle to generate a sharp image of the vehicle along with blurred striped background that helps indicate motion. So, from an educational point of view, what is the advantage to photographing vehicles/transport using this camera? -- Colin (talk) 09:05, 13 February 2017 (UTC)

Some advantages include being able to capture very long vehicles like articulated buses, trams, and trains; and also to capture different types of vehicles against the same background to easily compare scale. Moreover, panning is never as sharp because the angle of the camera changes with respect to the vehicle during the panning. Dllu (talk) 21:13, 13 February 2017 (UTC)