Hello again! Isn't it still in the middle of the night? OK, let's see. If plugins are not an option, you have to go the hard way. I don't think that there is a problem with 1000 pictures itself, it's just an enormous task to prepare everything. If you don't mind this, you are good to go. As descriped above, I would break down everything to a single comp which contains the image and a mask only. Creating the mask from a shape layer is a bit more handy, but you can have the mask directly applied on the image as well. Then you would have to duplicate this comp and switch the image by selecting the layer in the comp, hold ALT key and dragging the new image over the selected. This replaces only the footage, everything else will stay. So you can have already an animation applied and it will still be there. You'll have to repeat this for all your images and it's a matter of your perseverance if you can do it a 1000 times. Alternativly you can write a script, or maybe find one, which does exactly this. If you have no experience in AE scripting and programming, you are faster doing it by hand. Next step would be to add the animation, but this kind of depends on the animation itself. Say you want to rotate every circle-image in 3D space, so they make a flip, best practice is to apply this animation to the precomp. You would put the precomp including image and mask into a new comp of the same dimensions, switch on 3D and make the animtion. The next step is to put every precomp including the animation into your "main" comp, forming the final mosaic. This is where AE won't be happy in dealing with 1000 layers. It will work, but you might run into a effect that you cannot see the layers in the UI anymore (some years ago I created a massive triangle-soup with 800 layers or so and ran into this). To prevent this you can split the final mosaic into 4 250 layer comp and finally place the 4 comps into your main comp. When creating the mosaic, you can use expression to place every individual precomp. For horizontal placement, the next layers x-position is the width of the previous layer (or a constant value, since every precomps have the same dimensions). When you have one horizontal line, you can precomp this. You do this for all lines and then you just need to vertical place all horizontal lines. You can use the same expression logic, but now the layers y-position is the height of the previous layer, or constant value. There are more sophisticated methods using modulo but if you can create horizontal procomps, there is no need to go crazy here. If not already in use, you should get the script MoveAnchorPoint (there should be a free version still available) for this project. Also TrueLayerDuplicator will come in handy: https://www.vdodna.com/blog/true-layer-duplicator-script-for-after-effects/ And check you TrueCompDuplicator, might be useful as well. *Martin
... View more