Stereotowns

In 2016, Jonathan Mines and I created Stereotowns, a 3D videographic work for the academic journal [in]Transition.

 

Stereotowns title copy.jpg

This was, as far as we know, the first stereoscopic academic video essay although Mubi has some good articles on other academics working in the area of stereoscopic production. It involved a lot of trial and error (mostly on the part of the hardworking Jonathan) in order to put the images and ideas on screen. Below is a guide to how we achieved it in case anyone would like to follow and build upon our endeavors.It is also available in PDF form

Creating a Stereoscopic Videographic Work

Our video essay can be roughly divided into two parts, one that presents nineteenth century stereoscopic stills and one that cuts up and edits together stereoscopic footage from recent Hollywood blockbusters. Both aspects were brought together in Final Cut X and then uploaded as a single side-by-side file on YouTube. Below are details of the processes we used which are specific to the software we had available but should be easily transferable to other software programmes.

Working with stereoscopic stills

Copyright

Using 19th century stereoscopic images can be tricky territory in terms of copyright – the original image should no longer be under copyright, but a digital scan itself might be. Any modern content almost certainly will be copyrighted, but depending on your aims you might be protected by fair use. It isn’t possible to go into detail concerning all the varying contexts and circumstances around the copyright of stereoscopic images but the following sites can provide some good starting points:

http://fairuse.stanford.edu/ – general copyright and fair use information

http://creativecommons.org/ – very useful guides to creative commons licensing

http://creativecommons.org.nz/ – a New Zealand specific guide to creative commons

https://search.creativecommons.org/ – a creative commons search engine

https://www.flickr.com/creativecommons/ – flickr includes licensing with user images, making it easy to know your usage rights with those images. Additionally, the user upload system makes it easy to get in touch with the images owners and directly ask them for further information about licensing (or about the images themselves). We recommend checking out Wade Roush’s images and the Boston Public Library’s stereograph collections.

Manipulating analogue stereocards

Our use of still images from the 19th century posed a number of problems. How were we going to make analogue stereoscopic material, designed to be viewed with one of these  digitally viewable?

pedstereo

And how do we maintain a sense of authenticity, a trace of their original display mode, when we are so heavily processing the images?

Typically, 3D monitors will read side-by-side images as separate images and overlap them so that they can then be viewed in a stereoscopic mode. This means that the mechanics of modern stereoscopic productions are similar to those of 150 years ago, when stereographers typically produced side-by-side images in order to be combined in a viewing device, albeit an analogue one. For this reason, we were able to use the side-by-side sets produced in the nineteenth century but they needed photographic manipulation in order to operate as side-by-side images that could be read by a 3D monitor.

Jonathan worked on manipulating these images and tried to strike a careful balance between modernising the images while remaining true to their intended viewing experiences. To achieve this, he preserved each image’s original frame in order to simulate the original viewing context, while reworking the images into a 16:9 aspect ratio that would match the rest of the content being used in the video essay.

He split the scan of the stereoscopic card into 3 images (the reasons for this will become clear soon). In order to do so he used:

  • An analogue stereoscopic image (for example, Wade Roush’s copy of the “Sphinx and Pyramid”) Sphinx
  • A Photo Editor (he used Adobe Photoshop)
  • Video Editing Software (he used Final Cut Pro X)
  • A stereoscope plugin (he used Dashwood’s Stereo3D)

 

In the photo editor:

1. Remove the space around the edges of the scan of the stereocope (you might have to rotate the image slightly to preserve as much of the frame as possible). You can leave the space in the corners intact for now.

Step1

2. Maintaining aspect ratio, resize the image so that the height is 1080 pixels (or 720 if you’re working in 720p, or 2160 if you’re working in 2k, etc)

Step2

3. Save a copy of the image as “Image raw,” (you’ll be returning to this iteration of it).

4. Crop the image, so that you have a rectangle containing only the left and right images (i.e. no frame). As you can see we’re already compromising the intended viewing experience. Unfortunately, this is the simplest way of keeping the images aligned for a comprehensible stereoscopic viewing experience.

5. Save another copy of the image here as “Image_noframeraw”

Step4

6. Adjust your canvas size to be 50% of its current width, cropping out the right part of the image.

step5

7. The final image output needs to be 1920×1080 (or whatever size and aspect ratio you’re working with). This isn’t an exact science, and you will have to bear in mind that the frame will obstruct some of your final output. For this reason, Jonathan used a combination of stretching the image (while maintaining its aspect ratio), cropping it, and adding a small border to it (which is obstructed by the frame in the final output).

8. Save this image as “Image_left”.

9. Open “image_noframeraw” and repeat steps 6-8, instead cropping out the left part of the image, and saving the file as “image_right”. Make sure that you repeat your resizing process in step 7 exactly.

10. Open “image_raw”.

11. Remove the image inside the frame. You’ll probably want to feather this slightly – you might have to experiment with settings a little here.

step10

12. The frame remaining is for two images – we have to manipulate it into a single image frame that is 1080 x 960 (i.e. half of 1920). How you go about this will depend on the sections of the frame you wish to preserve. As Jonathan wanted to preserve the text (“Sphinx and Pyramids, Cairo, Egypt”) on the right part of the frame, he essentially selected and feathered the top left corner of the frame, placing it over the right part of the frame.

 

 

 

While this image is 960 pixels wide, it will be stretched to 1920 when viewed stereoscopically. As such, he manipulated the text to be half its width, as to not have it distorted by this process.

13. Remove the spaces around the edge of the frame. Because we made the images rectangular, the original frames will leave white space inside the frame – fill in this white space.
To remove the white spaces inside and outside the frame, Jonathan used the technique of duplicating and feathering other sections of the image but if you’re skilled in photoshop then you might find a better technique.

14. Save this image as a .png (to preserve transparency) named “Image_frame”.

Now most of the hard work is done and we’re ready to take our images into editing software such as Final Cut X.

In the Editing Software:

1. If you have a 3D monitor attached (highly recommended), make sure “A/V output is selected, so you can preview your final output stereoscopically.

2. Import “Image_left,” “Image_Right,” and “Image_Frame.” Place “Image_Left” and “Image_Right” on top of each other in your project/timeline.

Step1.png

3. Using your stereoscopic plug in, tell your editing software which image is left and which is right. With Dashwood’s Stereo3D, this means applying the “Identify Left” and “Identify Right” effects on the corresponding clips, as well as applying “Stereo3D Toolbox LE” on both.

4. Within your stereoscopic plugin, adjust the necessary variables. In Dashwood’s Stereo3D, we turned on “Use Global Output Mode” (this shows you the final output), and adjusted the convergence slider (which adjusts the point in the images where depth is at zero parallax). The vertical, horizontal, and zoom disparity sliders allow you to adjust for any differences in the source left/right images.

Step4.png

5.Place two “Image_Frame” clips above the “Image_Left” and “Image_Right” clips.

6. Adjust the x axis position of one of the “Image_Frame” clips to -480. The reason this is 480, is because it is the centre of the left half of the screen (1920/4 = 480).

Step6.png

7. Adjust x axis position of the second “Image_Frame” clip to +480.

8. You should now see both sets of the image side-by-side in your editing software but showing up as one, overlapping stereoscopic image on the 3D monitor

 

Adding stereoscopic titles

The next part of our project involved adding stereoscopic titles that conveyed ideas about the images to our intended audience. In order to do so we undertook the following steps:

1. Create 2 identical clips of text with your editing software’s in built title creation tool and place them one on top of the other in your project/timeline. Move them to +480 and -480 on the x axis respectively – this gives your text zero parallax, and will place the text in the very centre of the screen in your final output.

Step1.png

2. If you wish to adjust the placement of the text, first adjust the left screen text. Adjust the right screen text to the same y position as the left.

3. Adjust the x axis the same amount as you did the left. So, because my left text is at -820, I moved it -320 (-480 – -820 = -320) pixels. As such, the right hand text x axis should now be 160 (480 – 320)

Step3.png

4. At this point, the text still has zero parallax. To create a stereoscopic effect, simply adjust the x axis placements. Moving the text further apart will render it in positive parallax (appearing behind the screen). Moving the text closer together will render it in negative parallax (in front of the screen).

 

5. In Stereotowns, we often animated text to move it through parallax space. To do this, place the playhead at the point you wish the animation to start, and add a keyframe to the parameters you wish to adjust. To add a keyframe, hover your mouse to the left of the parameter you’re adjusting, and click the + sign that appears – when it turns yellow, the keyframe has been added.

Step5

6. Move the playhead to the next point in your animation, and adjust the keyframed variables – adjusting the values will automatically add keyframes.

7. To edit and move keyframes, right click the clip and select “show video animation” – the blue diamonds are the key frames. To move them, simply click and drag them. To edit them, place the playhead over the centre and adjust the values of your parameters.

Step7.png

 

Working with stereoscopic feature film footage

Inputting feature films

In the next part or our project we inserted footage from Hollywood 3D blockbusters to provide comparison points with the still images. We sourced files of the 3D films that had a side-by-side layout of the left and right eye image sets. Some files of 3D films come in an up-and-over format which would involve a great deal of manipulation on our part so it was easier to use files that were already in a side-by-side format. This meant that when we inputted them into the editing software the layout was similar to the layout of the still images.

Editing stereoscopic footage

Once you have inputted the files you should have a project of raw, stereo-ready clips. It can be helpful to consolidate raw material into “compound clips.” This tidies up your project, and makes it easier not to accidentally lose material that should fit together when editing. To do this, highlight the relevant clips and select “New Compound Clip.” You can edit this clip by double clicking it at anytime.

Step7

You can now edit the footage you have in your project as you would any other. At this stage it is useful to check the 3D monitor to see how any effects such as transitions and titles work stereoscopically.

 

Outputting and uploading the project

Outputting the project

You can output the project using the parameters you would normally use for video projects. For example, with Stereotowns, we outputted with a h.264 codec. Take into consideration that 3D monitors such as commercial 3D televisions will only play certain file types and codecs.

Once it has outputted, if you open your file in videoplaying software such as VLC or Quicktime you will see a side-by-side layout.

VLC view copy

When this file is played in a 3D monitor, the 3D monitor should be able to convert it into a stereoscopic format.

Uploading and displaying the project

YouTube currently supports stereoscopic content although, at the time of writing, it constantly shifts how it will display in different browsers and the options viewers have to change that display. Nonetheless, at the point of upload, you can designate your file as ‘3D’ meaning that regardless of what YouTube does with its display modes, it will always recognize your video as being stereoscopic. This means that any 3D monitor playing the video from YouTube will usually be able to convert it into stereoscopic format.

Follow your normal process for uploading to YouTube and in the ‘edit video’ section go to Advanced Settings. Click the 3D video box towards the bottom right side.

youtube copy.jpg

When it comes to viewing your video on YouTube on a 2D monitor. Currently:

In Firefox and Chrome it will give you an anaglyph view or 2D view option

anaglyph copy 2

In Safari it will only allow you to view it in side-by-side mode

youtube 3 copy

We hope these points have provided a useful guide for creating stereoscopic videographic work

 

Miriam Ross and Jonathan Mines

Victoria University of Wellington

April 2016

 

License: Attribution-ShareAlike
CC BY-SA

Advertisements
%d bloggers like this: