anna ☆

psst: i usually always create in photoshop cs5, but for this effect you need a recent version of photoshop because it's using transform keyframes (i think cs5 doesn't let you do that, or i just don't know how to lol). i used cc 2019 for this.

sorry if this is lengthy or has too much or too little details haha, but i hope it's comprehensible! english is not my first language so i also apologize in advance for any mistakes!

I. PREPARING THE CASSETTE

so, starting with the png here:

i removing everything i didn't want on the cassette png with the brush tool by just drawing the right color over the unwanted text. for the color, i then went to image > adjustments > hue/saturation and in the red tab, i played with the hue slider to get that purple color. finally, i added some text to my liking, and this is what i ended up with:

(not necessary but: i also selected the white lines on each side with the magic wand tool because i wanted these lines to be transparent. once your selection is done, go right click > layer via cut. it will create a new layer of the cutout you just made. you just need to disable or delete the layer to make the selection (lines) transparent.)

at this point you want to have only 2 layers: the revamped cassette and the text layer. you can remove the text layer actually, and just add the title back at the end, as it is not necessary for this effect. i just like to have the visual.

if you have multiple layers, you need to select all of them (except the song title layer), right click on the png layer and click on merge layers. this will create one layer with all the editing you made on the cassette. if you think you will need to edit this later though, i would save the file as a psd before merging the layers.

II. THE EFFECT

okay, so now that you have your cassette, make sure your video timeline is activated, not frame animation, and you are ready to go.

first, you want to create a perfect circle shape around one of the reel with the elliptical marquee tool (hold shift while dragging the circle). make sure it covers the entire area that will later be rotated. make sure this circle is perfectly centered around the reel or otherwise the animation will be a bit lopsided.

then right click on this selection and go "layer via copy". this will create a layer of only that circle selection. important step: right click on that new layer and go "convert to smart object". the layer should look like that, i've renamed mine:

now if you go to your timeline and open that new smart object layer, you will see that you have 3 keyframe options. we only need the transform one.

go to the start of the timeline and activate the transform animation by clicking on the stopwatch button. a keyframe will be created automatically.

to create the actual animation, move the position of the cursor on the timeline further, i put mine at the 01:00f mark so it's easier to create the right timing.

then what you want to do is select the reel smart object layer and hit ctrl + T. a box will appear and this is how you will make the reel rotate.

to rotate the reel shape, move your cursor near the blue box on your canvas and drag it until you have rotated the shape halfway through and hit enter. another keyframe will be created and if you play your animation, the reel should rotate on itself for half a turn

move your position on the timeline to 02:00f and do the same thing: select the left reel smart object, hit ctrl + T, rotate for another half turn, and hit enter. this third keyframe should be the last one needed for the animation and you should have a full animated rotation of the reel.

play your animation, and adjust the speed to your liking by dragging the keyframes on the timeline (but make sure they stay within the same distance from each other). the closer the keyframes are, the faster the animation are, and the further they are, the slower it'll be.

then you can just trim the smart object to your animation's length, and duplicate (right click the smart object > duplicate layer) this layer the amount of times needed (i find this less finicky than duplicating keyframes), and placing them one after the other. three full turns should be enough. this is what my timeline looks like right now:

and my animation for the left side looks like this:

as you can see, we can see the little "dents" peeking through behind the animation. we don't want that! to remove it, select the revamped cassette layer (that should be under the reel smart object), and create another perfect circle around it with the marquee tool. this time make sure it's smaller than the previous one, it just needs to cover these dents.

then right click on this selection on your canvas and go "layer via cut". this will create a new layer with that selection, and all you need to do is to disable it. this is removing the information in that circle.

once you are happy and the animation works, you can just delete that cut layer. now the animation is done and looks like this:

III. SECOND ANIMATION

one you have done it on the left side, you just gotta do the same thing on the right side. you can also try duplicating it, but i found it finicky for some reason (or maybe i'm just not used to the controls of this 2019 photoshop version?).

this is what i have once i've done the same thing on the right reel:

once i am happy with the speed and everything, i want to have only one layer so it's easier to use on gifs. first, i will save this animation as a psd file, in case i want to reuse it. then i am removing the song title layer and will be flattening everything and creating frames from this animation. to do so i am using the "save" action from here.

i'm not sure why it does that, but it's creating a couple of frames where the reels are a bit offset from their position everytime there's a full circle done, so i just delete these 5-6 frames. you can also change the speed here, but by default it should be 0.05.

once you are happy with it, just turn these frames into a smart object with the video timeline again (convert frame animation to video timeline and select all the frame layers > right click > convert to smart object)

now you have a smart object that is ready to be used anywhere!

IV. FINAL TOUCHES

for my particular speak now gifset, i have multiple layers of the animated cassette on each gif:

1, bottom one - cassette layer set to the blending mode "hard light" and set to opacity 86%:

2, middle - this same cassette layer set to hard light, but with the opacity at 100%, AND with a layer mask so it's only applied to the animated reels (i wanted them to show up more):

3, top one - and finally a third layer with another layer mask because i wanted the white label and speak now area to be less see through. it's set to the normal blending mode and the opacity is at 75%

and then i just added the song title on top at 100% opacity and normal blending mode, and added some drop shadows, and tada!

I. CHOOSING THE RIGHT SHOTS

the ultimate trick to pull this off is to choose the right image. in order to do the double exposure, you need a silhouette shot that has these:

a defined and dark silhouette with a background that is not too busy

enough contrast between the silhouette and the background

the silhouette should take at least 50% of the space

not too much movement

here are a few examples of why they work and why they won't:

gwen: perfect example since this shot is already quite contrasted with a defined silhouette. there won't be a lot of work needed to make this one work.

merlin: not a great example because even tho there's a somewhat good contrast between him and the background, the silhouette is just too bright, not dark enough.

arthur: another good example, even if there are some bright spots on his face and armor. since he's not moving too much, you can definitely brush some black over him to make his silhouette darker (i'll explain/show later)

morgana: this one could work because the contrast is great, but of course her skintone is very bright against the black clothing. that being said, since the movement is not too bad, it could be possible to brush some black over her and move these layers with keyframes (as mentioned for arthur's example). i haven't tried it tho, but i think it would work well enough.

once you have your silhouette shot, you need another gif for the double exposure. what works best, in my opinion, are:

wide, large shots

shots with no to little camera movement (no pan, zoom, etc), but the subjects in the shot can have little movement of course

pretty cinematography/scenery shots

i find these are easier to find and make it work, it's not as "precise" as with silhouette shots. it's mostly just trial and error to see what works best with the silhouettes.

II. PREPPING THE SILHOUETTE

for the effect to work, we want a silhouette that's dark as possible. i'm gonna use the gwen and arthur shots as examples.

for the gwen gif, i started by sharpening, and then upped the contrast by quite a lot so her silhouette is mostly black, while retaining some nice details. i've used only 3 layers here:

selective color layer: in the blacks tab, playing with the black slider (value: +10)

brightness/contrast layer: added a lot of contrast (+61) and a bit of brightness (+10)

black and white layer: on top, its blending mode set to soft light and at 20% opacity. gives a bit more depth and contrast

then for the arthur example, i've also sharpened it first, and added contrast layers in this order (the skintone looks horrible, but it won't matter soon lol):

levels layer: black slider at 0, grey slider at 0.76, white slider at 104

selective color layer: in the blacks tab, black slider at +10

brightness/contrast layer: brightness at +1-, contrast at +47

black and white layer: on top, its blending mode set to soft light and at 20% opacity. gives a bit more depth and contrast

as you can see, half of his face is still quite bright. to correct that, create a new empty layer and put it between the gif and the coloring layers.

using a really soft brush and the black color, brush some black over his face and body on that new empty layer. you can edit the layer's opacity if you want, i've set mine to 71%. since arthur doesn't move much here, there's no need to keyframe this layer's position. for the morgana example, this is where you'd need to play with keyframes to make it work. here's where i'm at now after this:

you can always edit this layer later if you need, after doing the double exposure blending.

once the silhouette is all ready, you can put all layers in a group and rename it (i've renamed mine silhouette).

III. BLENDING

now the fun part! import the wide/scenery shot in photoshop, then resize it to the same height of your silhouette gif. make sure the gif is a smart object layer, and sharpen it. finally, bring this gif onto the silhouette canvas (by right clicking the smart object > duplicate layer). once you have both gifs onto your canvas, put the wide shot gif layer in a group, and set this group's blending option to screen.

you can then position the wide/scenery gif the way you like it in the canvas. this is how it looks for both examples after i've done that:

if the blending mode screen doesn't give you the best result, so you can play around with other blending modes (such as lighten and linear dodge in these particular cases), but generally speaking, screen is the real mvp here haha.

IV. COLORING

now that the double exposure effect is done, we need to color the gifs to bring them together. i went with simple coloring here, simply enhancing the colors that were already there. just make sure that the coloring layers for each gif are in their respective groups. here's how i've colored both examples:

gwen silhouette group: i added a gradient map layer on top of the contrast layers in black to green and set the blending mode to color

scenery shot group: multiple coloring layers, with a green color fill layer (blending mode set to color), with a layer mask so it only affects the top half of the gif

for the arthur gif, i did something very similar but with warmer colors. i didn't use a gradient map for arthur though:

arthur silhouette group: i made the yellow warmer, closer to orange/red, with a hue/saturation layer, and added more vibrance. didn't feel like it needed a gradient map layer here though.

wide shot group: basic coloring layers to enhance colors from the merlin & daegal shot, and an orange color fill layer set to the color blending mode.

at this point you're pretty much done. just need to add some final touches and typography (if you want).

V. FINAL TOUCHES

a small and completely optional detail, but i wanted to soften the edges of the wide gifs. to do so i've duplicated the smart object gif layer and removed the sharpening filters (right click on smart filter > clear smart filters). put this layer on top of the other smart object layers (but still below the coloring).

then with this same layer still selected, go to filter > blur > gaussian blur... > 10px. this will give you a very blurry gif, but we only want the edges of the canvas to be softer. so add a layer mask to this layer. with a very large and soft brush (mine was at 0% hardness and about 800px size), brush some black onto the layer mask to remove the blur in the middle of the gif.

you can play with this layer's opacity or gaussian blur amount if you want (by double clicking on the gaussian blur smart layer filter). here how both examples look with this gaussian blur layer:

you can also mask some of wide/scenery gifs if you'd prefer, so it shows less outside of the silhouette. just put a layer mask on that whole wide shot group and brush some black or grey on the layer mask. it's what i did for the gwen gif, with a very soft brush and i set the mask density to 72% (i kept the arthur one as is tho):