Skip

How do you make 'cinegraph' gifs?
April 19, 2011 3:24 PM   Subscribe

We've talked about how to make gifs before - but what about "cinegraphs?"

I'm curious what the process is behind these gifs? We talked about making gifs before here - but what makes these animated stills different (aside from super models & gorgeous framing)? I can't put my finger on the technical bit and it's driving me crazy!
posted by teststrip to Computers & Internet (7 answers total) 24 users marked this as a favorite
 
Copy the still portions of the image onto each frame of the gif, the moving part remaining untouched.
posted by fire&wings at 3:26 PM on April 19, 2011 [1 favorite]


Actually if I remember correctly you can use transparency so that the further layers just contain the parts that change. A lot of Livejournal icons use the technique to save on space because of LJ's limitations on how big your icons can be. Maybe look here under the animation tags?
posted by kmz at 3:55 PM on April 19, 2011


Using a program like Adobe After Effects, you would take video/continuous-shutter-photos from a dslr or high-resolution camcorder (which was mounted on a sturdy tripod to avoid shake or movement) and choose your favorite stillframe. You would use that stillframe/cropping as the main "background."

You would create a new layer using the same video file and match the cropping/positioning (if necessary)

Then, by defining areas called masks you would isolate the video section(s) that you want to use in the animation. This masked video layer sits on top of the background still frame.

The masks prevent video from playing outside of those defined regions. Since the models are fairly motionless and the camera completely static, the video regions should perfectly match the stillframe background.

With a few tweaks to the masks etc, it is a simple render into .gif format, and your animation is complete! The gif-pixelated-look hides the digital video noise that you might see within the mask boundaries (i.e. on the cake mix, you can see the masked area has a little bit of pixel movement about 4-10px outside of the mix falling...gotta look close!)

It's not a quick process by any means but it is the method I would choose.
posted by Khazk at 4:08 PM on April 19, 2011 [1 favorite]


Most of the time when you see an animated GIF it's been converted directly from a video clip and it looks like ass. The reason it looks like ass is because the GIF format was never designed for movie clips and you have to make tons of compromises to get a file of reasonable size: you have to reduce the color palette (which means posterization or dithering, both of which look like poo), you have to reduce the resolution (so they are tiny), and you have to decimate the frame rate. This kind of 'dumb' conversion where you give some program a video clip and out pops an animated GIF is suboptimal. GIF is very dumb, and it can't cope with parts of the frame changing a little -- it has to encode the entire difference. This means that your typical video source that has some amount of low-level variation (noise) of even the static background aspects will register as difference, and the GIF won't be able to just delta it. These examples you linked to are hand made, and using an absolutely static background, achieved through manually masking off everything that isn't moving. This results in much better GIF compression as you force it to only encode the differences where there is real motion, not where there is camera shake or h264 noise. By ensuring that you don't have to make all those compromises listed above and you can keep your nice 256 color palette with only minimal dithering. That's why those images (and the If we don't, remember me ones) all look so slick.
posted by Rhomboid at 4:31 PM on April 19, 2011 [2 favorites]


You can open them up in GIMP; each animation frame appears as a new layer.

Only the background image is full frame; the others act as "stamps" on top of what went before.

You can position each frame independently as well- they don't have to always start at the same XY locations. This means you can have multiple independent animations going, you just have to do the math to get the timing right.

Way back in the day (back when netscape 2 started supporting them) we used to compose these on the command line through a mix of perl and dead reckoning- it was really nice when GUI tools came along that could let you actually see the animations you were working on.

There's a pretty good reference on wikipedia of course.
posted by jenkinsEar at 6:08 PM on April 19, 2011 [1 favorite]


What I do to make a very crude version of this effect is (in photoshop, with the animation as a video layer) take the still shot, put it over the animation layer, and cut out holes to show the moving bits. That obviously falls apart when you get to something complex like this, though.
posted by btfreek at 7:32 PM on April 19, 2011 [1 favorite]


Here's another tutorial on creating Animated GIFs from movie scenes.
posted by labnol at 11:48 PM on April 19, 2011 [2 favorites]


« Older Where can I go to sell custom ...   |  I have to buy a fridge and I d... Newer »
This thread is closed to new comments.


Post