Here you can find some experimental scenes we created during the development of FlashFilterLab. We selected the more interesting ones in order to show you the possibilites of this tool. We added a few words to each scene which describe the main idea behind that effect and hopefully give you some directions in which you can start your very own experiments. If you did something interesting and would like to see it here, feel free to post it to email@example.com.
The experimental area is divided in to four categories. This main experimental page holds the entrance to the categories and the last four recent experimental entries.
These are our expectations what the experimental movie should offer: It should be interactive in some new way and make use of Flash filter’s possibilities to produce high quality visuals even with reduced download size. And mainly it should use the least possible conventional Flash key frame animation in order to reduce human resources due to time pressure. Oh, and I forgot to mention that there was not any professional equipment used for this movie to prove anyone is able to create a movie like that.
So, the challenge is given and the movie has to be finished by the end of September to make it a good start for FlashFilterLab. Let's just hope for the best...
The first possible story that came to our mind was the very old story of Adam and Eve and the well known expulsion from paradise. I can’t get the thought out of my head, that this little incident is the one that had the most influence on our lives and sealed our fate for ever. So it definitely looks like a story that is worth dealing with.
Of course we have our own version of the story, which will spice it up a little bit, but we try to keep it for ourselves till the end.
We will need two locations for the movie, the paradise and outside of it. For some reason, we know better how it looks outside, so that is the starting point. In our version, we use the two locations as two metaphors– the paradise for the unity and outside for the partiality.
We will use two separated small rooms for the scenes outside the paradise. Our actors Adam and Eve are reprensented by two dancers which should make the movie more like a contemporary bally than a conventional movie.
To give the viewer some unusual interactivity, we built a quite simple flash scene, which simulates a 3d room where the camera can be panned among one pane. The camera movement is generated by an available webcam based motion tracking FFL module created using the numbers returned by the threshold filter on a time delayed and difference-blended webcam input.
In our first test setup, an effect applied dancer is dancing in one place. The mouse movement simulates a simplified camera movement. The resulted scene became a quite popular one, which gave us the impulse to go for the next step. But with this impulse a new challenge turned up: In this test scene the dancing person is not moving from his place, which made it quite simple to transform its movie clip to match the camera’s view. But -of course- this cannot be like this in our final movie because our actors should definitely be moving around, in the given room.
So we had to come up with some new solutions. In order to reposition a moving object to match the viewpoint of the camera we somehow have to know the object’s distance to the camera. To cut things short, we took a simple webcam and captured the whole dancing scenario from the side as well and from a relative big distance in the front. This footage was to be the holder of the distance information measured between the main camera and the dancer. We simply imported this movie to Flash and put a dummy movie clip to the position of the dancer. Using a simple motion tween , we positioned our dummy using key frames to the position of the dancer. This way we had that dummy movie clip instance which we could easily use to retrieve its x position which also meant we could calculate a concrete value for the distance, we were just looking for. So there are no more problems to make a quite interesting 3d-like camera movement in realtime possible.
The portrait blur is a commonly used effect for removing smaller flaws from a face. The main idea is leaving the edges as original and add a small amount of blurring to anything else.
To determine the edges of the portrait we use a convolution filter with the edge detect preset. Then after we applied some smoothing via a blur node we use a threshold node to create a black-and-white image. This image will be white where we don't want to apply the blur - at the edges of the face.
With another blur node after the threshold we can make the transition between the blurred part and the edges more smooth. To create the non-blurred parts of the image we use a channel copy node so only the threshold-defined edges of the original picture will be seen.
To build the final picture simply add some blur to the original one and merge it below the the edges we've just created.
A similar effect can be be achieved using the bitmap threshold module which applies the Flash 8's BitmapData.threshold function. This alternative method uses less CPU but you cannot smooth the transition between the two parts of the image. This is because the module node simultaneously performs the function of the threshold and the channel copy node, so you cannot link a blur node between them.
This scene is based upon a threshold and a channel copy node. With a threshold node you can map the incoming movie's pixels' luminance. Every pixel that is brighter than the value you've given in the parameter will be white, the darker ones will be black.
You can use the resulting image as the alpha channel of another one. If the channel copy's source channel set to any color and the destination to alpha, the destination image will have holes where the thershold's output is black.
In this scene we used a gradient with a HSL effect on it and made a hole inside it by the means described above. With a glow effect one can enchance the visibility of the borders which also considers zero-alpha areas inside not to be the part of the image.
With a merge box we placed the original movie under the effected one so you can see the ground in different color but the flower itself keeps its own.
This scene's main idea is the same as in the previous one, but we changed the concept a little. The movie with the black background is used as a displacement map for the tiles. So the tiles will be distorted only where the hands are present. With a small amount of blurring the displacement looks more interesting as it adds gray shades to the edge of the white part.
With the channel copy we cut out the hands movie's black areas from the resulting image so it will be clearer that what's moving are actually hands. A small amount of blurring here gives the alpha channel some anti-aliasing effect due to the previously-mentioned gray shades.
In this scene we included a slider node so you can easly try what happens if the amount of blurring changes.