Ka of Isis is not optimized for this version of Internet Explorer
16 November, 2008 | Filed under “Final Cut Pro”
When, in 1985, Norwegian pop group a-ha with director Steve Barron made a second attempt at a video to promote their single “Take on me”, the combination of a new song mix by Alan Tarney, heavy MTV exposure, and the highly original video animation by Michael Patterson, made the recipe for a worldwide hit.
JavaScript must be enabled and Flash Player installed to experience this page as intended.
Now, a blend of masked video and hand drawn sketchy animation is hardly the best starting-point for an automized production method. As such, the “Take on me” video can never be more than an inspiration in an effort to “draw with charcoal in video”. I would not be surprised if there are Final Cut Pro (FCP) filters out there making a good job of it, but a lack of funding and understanding of how to create filters for FCP, sent me in the direction of Photoshop, now capable of video handling. Although, my initial optimism should soon turn sour in that department.
JavaScript must be enabled and Flash Player installed to experience this page as intended.
Sure, you can make global corrections like removing colour cast from a video sequence in Photoshop (and that is Photoshop Extended, mind you). The ever inspiring Russell Brown can show you that and you can expand on the technique. But I might as well tell you up front: If you want to manipulate individual frames in a video, you will at some point have to save out each frame (of a sequence) as an individual image to be able to manipulate it with Photoshop, either manually or with the aid of an action. That is, you could achieve your goal, but it would be extremely tedious and time consuming. Remember, 1 second of, e.g., PAL DV video translates to 25 frames. And how long is your sequence?
This initial dilemma leads to the first decision: There is no point in exporting a movie from FCP and then open it in Photoshop, only to save out frames from the latter. You should export the image sequence from FCP; that is where you start, anyway.
Nice! You head up to the File menu and look for an image sequence exporter. Yes, right there, under <Using QuickTime Conversion…>. But wait! You stop short in your tracks. There is no option that allows you to save the images as proper square pixels. If you rush ahead (and we stick with the PAL DV 4:3 example), your images will end up 720 pixels wide and 576 pixels tall and exhibit the squeezed appearance that reflects the difference between traditional non-square television “pixels” and a computer’s square pixels. 768x576 is the way to go when preparing images for PAL video.
Yes, yes, but Photoshop can deal with that, you say. Right you are, but of the 5 choices available for image resampling (Nearest Neighbor, Bilinear, Bicubic, Bicubic Smoother, and Bicubic Sharper) which one is the best for video? I do not know, so I leave it to the video program and not the image manipulation program to decide.
Now, those square pixels. How to get them? Where QuickTime fails, Compressor comes to the rescue. You should have the latter application on your system as part of the FCP installation. And there you have it: The second decision. From here on in the lights are green.
1. Prepare the video sequence in FCP. Mine was a 20 seconds exercise of bikers in Dublin, shot interlaced in DV 4:3 for PAL. At 25 frames per second that is 500 frames; 499 images with a black slug at the end. You may de-interlace at this stage but if your final movie is depending on interlacing, there is a short discussion on an alternative later on.
Target only the sequence of interest (you end up with a lot of images as it is and they digest hard disk space like locusts strip a crop field).
2. In FCP, go File > Export > Using Compressor…
Compressor opens with a Batch Name already filled in and with your Source Media in place in the Batch window. At this stage you have a choice of encoding presets, none of which will serve your purpose. You will have to prepare your own …
3. Show available presets by clicking on the <Presets> button.
4. To create a new preset, first click on the <+> button and then choose TIFF from the drop-down menu.
Your preset will be given the name Untitled TIFF. Double-click on the name and change it to something useful, say, TIFF Sequence 768x576.
5. Open the <Encoder> tab and accept TIFF as the Encoder and tiff (or tif) as the File Extension. Change the Frame Rate to match your movie’s frame rate, and check both of the remaining choices.
Important: It is imperative that you add leading zeros to frame numbers when the sequence is longer than a few frames. See discussion further down.
Fig. 1. Automatically add leading zeros to the names of sequence images.
6. Open the <Filters> tab to make sure that none of the filters are checked, with one possible exception: If you did not de-interlace you movie earlier, you can do so now, and you should! This way your original movie is not affected and each individual sequence image will be given over to Photoshop with its Sunday clothes and shined shoes on.
Check Deinterlacing and choose the proper Algorithm for your sequence. Not sure about that? Return to your FCP project and look for Field Dominance in the Browser window.
Note: That said, I am not too happy with the way Compressor de-interlaces as opposed to the native filter in FCP (Effects > Video Filters > Video > De-interlace). You will have to optimize for your own project. In this particular case it is even possible that no de-interlacing at all might actually add an interesting touch to the “Take on me” effect. I never tried, though. Please observe also that your version of FCP may reverse the field dominance as discussed by Joe Maller in his article Joe’s De-Interlacer.
7. Open the <Geometry> tab, leave cropping at 0/0/0/0, then choose Custom as your Frame size. Enter a value of 768 for Width and 576 for Height. Constraining to display aspect should automatically change to None, and you round it up with a Pixel aspect of square.
It would have been nice to run an AppleScript on output, but the <Actions> tab is for a rainy day.
8. Close the Presets window and hit <OK> when asked to Save.
Your new preset is now added to the list.
9. Back in the Batch window, choose the new preset from the list.
10. Change Destination if necessary.
11. Change Output Filename if necessary.
12. Run the batch by clicking on the <Submit> button.
Note: When Compressor baptizes the images in the sequence, the numbering starts with a 0 (zero) the way programmers prefer. When FCP exports an image sequence, the numbering kicks off with a 1 in the name. Remember that if you ever do a direct comparison.
To save time when adding the same effect to a sequence of images in Photoshop, you should create an action for the application. This way you can also come back at a later stage with different “footage”, re-use the settings and even set up batch production. Do as follows:
13. Open any 8 bits RGB image – but take heed: Do not work on an important image! If it is, make a duplicate/copy because you will record a saving step as part of the action.
14. Bring up the Actions panel (Window > Actions).
15. In the Actions panel, click on the <Create new action> button (if you do not want the new action to end up in the Default Actions folder, precede with Create new set, name it and choose it in the New Action dialogue window when it pops up).
Fig. 2. Creating an action in Photoshop.
16. Name the action appropriately.
17. When you choose <Record> in the New Action window, that is exactly what you do: Everything from now on gets recorded by Photoshop as part of the final action.
18. Image > Adjustments > Desaturate
19. Layer > New > Layer via Copy (aka “Jump copy”). The copy layer is named Layer 1.
20. With Layer 1 targeted (i.e., highlighted in the Layers panel): Image > Adjustments > Invert
21. Change Layer 1 blending mode from Normal to Color Dodge.
22. Filter > Blur > Gaussian Blur…
Radius: 2 pixels
23. Create a blank transparent layer above Layer 1 by clicking on the <Create a new layer> button in the Layers panel. The layer is named Layer 2.
24. Stamp a copy of what is visible in the image window to this layer by holding down the <alt> key while clicking the Layers panel’s flyout tab (the black triangle) and selecting Merge Visible.
25. Change Layer 2 blending mode to Linear Burn.
26. Target Layer 1.
27. “Jump copy” this layer to its own new layer. The new layer’s name is Layer 1 copy.
28. Click the visibility icon off for the original Layer 1 (but keep the copy layer active).
29. With Layer 1 copy active, choose: Filter > Blur > Gaussian Blur…
Radius: 50 pixels
30. Target Layer 2 (the uppermost layer).
31. Create a blank transparent layer above Layer 2. The new layer is named Layer 3.
32. Stamp a copy of what is visible with <alt> + Merge Visible.
33. Hit “D” for default Foreground (black) and Background (white) colours.
34. Filter > Artistic > Colored Pencil…
Pencil width: 1
Stroke Pressure: 15
Paper Brightness: 50
24. Layer Opacity: 76%
25. Flatten Image (<OK> the Discard hidden layers? question).
26. File > Save As…
Note: Name and destination is not important at this point. You will be more specific with the batch setup described below. Do, however, go for TIFF and LZW compression. LZW is non-destructive and saves hard disk space.
27. File > Close
Important: Do not skip the Close step. If you do, Photoshop may choke on the hundreds or thousands of pages it will then have to keep open once saved.
28. Click the <Stop playing/recording> button in the Actions panel.
Note: You may have noticed that when you changed the layer opacity to 76% as part of the steps above, Photoshop actually recorded two steps; one when you typed 7 and one when you finished of with 6. The latter is accepted as 76 as soon as focus is no longer on the Opacity field. You may save a nanosecond for every image processed if you delete the first of these two steps from the action. A detail, I know, but that’s me.
Tip: Video sequences typically consist of a lot of frames, and the more complex the action, the longer it will take to complete. Photoshop is set up by default to play the action in an Accelerated mode. This way each step will be performed as quickly as possible and each intermediate step will not necessarily appear on the screen. Which is exactly what you normally want (it is time for lunch, anyway). The alternatives are Step by Step and Pause For: _ seconds, and you find them as <Playback Options…> in the fly-out menu of the Actions panel.
The action is perfect for single image manipulation. Open a picture, run the action and presto! Charcoal drawing complete. But it is rather crippled when you have a ton of pictures dumped from Compressor. Time for batch processing!
29. In Photoshop, go File > Automate > Batch…
30. Configure your batch as something akin to Figure 3.
Fig. 3. Batch setup in Photoshop.
A short breakdown of the settings is in order: Once you have specified which action to Play, choose the folder with all the Compressor images as the Source. There are no “Open” commands in the action and Compressor leaves no subfolders. No need to check those options, then. Unless you plan on sitting (awake) in front of the computer screen while Photoshop grinds away, check Suppress File Open Options Dialogs and Suppress Color Profile Warnings. Do specify a different folder than the source folder for your Destination. Otherwise your original images will be overwritten (i.e., if you do not specify a different file naming scheme). Keep the Document Name default, but remove extension. If you do not remove the latter, you will end up with names like frame--00227.tiff.tif; an unnecessary duplication of the file suffix. And finally, if something goes wrong while processing one-two-many images, you still want Photoshop to continue. But let it leave a (plain text) error log with details. Store the file anywhere but in the same folder as your finished pictures (see further down).
31. To start the batch, hit <OK>.
Tip: You may take automation one step further with a Droplet. To get the details on that, in Photoshop, go Help > Photoshop Help… and search for droplet.
32. Open your FCP project.
33. In the menu, go Final Cut Pro > User Preferences…
34. In the <Editing> tab enter 00:00:00:01 as Still/Freeze Duration. This way imported images will occupy one frame each in the movie.
35. To import the images, go File > Import > Folder… and choose the folder with the effect images from Photoshop. The folder, with name intact, is referenced in the Browser window.
Note: If the log file is inside of this folder, delete the file prior to import. Otherwise FCP will try to include it when you import all of the images in one chunk as a folder.
36. Drag the folder into the Timeline to add the images to the movie. The images are positioned in time according to their names and FCP automatically adjusts the size of each image to suit the project sequence (720x576 pixels in this case).
37. Test the movie.
Important: If your movie seems to jump back and forth you may have forgotten one small detail earlier when you exported each frame as a still image: Add leading zeros to frame numbers. If you open the folder in the Browser, you will see images numbered …0, …1, …10, …100, …101 … …109, …11, …110 etc. You get the drift. Unless you want to spend a lot of time renumbering all those pictures or reshuffle them in the Timeline, you really have to go back into Compressor and add this important preset detail.
Apart from that, have fun with your virtual charcoals and please write me a line if you come up with something clever, like post-treatment in FCP to polish the effect.
If you hate the thought of building the Photoshop action, head for the Downloads section and rip it. Then load the action from within Photoshop’s Actions panel.
If you are curious as to what a-ha’s first video for “Take on me” looked like, there is a small passage on their official community website.
Acknowledgement: Thanks to Shan Canfield for showing me how to merge visual elements to a new layer.