That's just a big image overlayed onto the underlying image grid to tint it. Basic blending modes stuff.
How would I do the animation of the smaller images dropping into place? As I said I am an absolute beginner at After Effects.
The technique you are trying to copy is using advanced animation and alignment techniques to distribute layers and animate their path in 3D space. There are several shots in the example video. I would start with each one as a separate comp then edit each comp in Premiere Pro or if your transitions are simple you can edit them in After Effects.
When starting to lay out a project like this you want to think in reverse and work backwards breaking down each movement into separate animations so you don't get overwhelmed. In this example you want to end up with something like this:
The intro to the video says drop 100 photos but this grid is 33 X 19 so if each grid is going to be a unique photo you'll need 627 photos to end up with something like the preceding shot. The grid lines can be achieved with the Generate>Grid effect so that part is easy.
As I look carefully at this image I can see lots of duplicates in the images so maybe you could get away with 100 of them or less. The easiest way to get from the overlay to the final image is to start with 2 layers, one which contains the grid of photos and a photo of the boy. To achieve the mix that you see you could put the grid above the photo of the boy and set then blend mode to overlay and animate the color. You can get different looks by changing the blend modes. Now you have figured out how to do the last shot of the sequence. You just need to figure out how to get the images in a grid.
This would be a good time to talk about image size. If you have 600 or so 20 megapixel images in your composition all scaled down to these tiny squares it's going to take a ton of time and resources to resample each one of them. You are also going to have to figure out a way to crop and position each one of them in their square frame unless they were all square photos. It's probably going to be a very good idea to use the batch process in Photoshop to crop each photo to a square and resize them to something that is going to be more usable in After Effects. You could put each photo inside a separate square comp that was a reasonable size for the project and then scale and position each photo in their own comp. Doing this manually would take a long time so an expression would help here speed things up. As long as you didn't turn on Collapse transformations with for the nested comps AE would not have to recalculate the pixels for every frame. This isn't as fast as resizing the images before you start but it would help. Wait, I'm getting ahead of myself.
The preceding shot is a camera panning across the grid of photos. The camera doesn't see all of the photos so this shot does not need as many images as the following shot. You will still need to figure out how to get the images into a grid and use that technique to build the last shot in the sequence.
Here's where things start go get more complicated. Every image needs to end up in a specific position so it will be easier to distribute all of the images on the grid and then back up one image at a time and fly them up and off the grid. To get the images into proper position and to make the camera move easy you'll need to make each image a 3D layer and rotate the image in X so that you can build the grid on the Y Z plane. Again thinking backwards, without the help of expressions or scripting, the easiest way to get all of the images lined up in a grid on the Y Z plane would be to start with 2D images and arrange them in a grid, then add a null, then make all of the layers and the null 3D, then parent all of the images to the null, then rotate the null 90º in X. To distribute the layers you could use a 3rd party plug-in like Echo Space from Trapcode, or write an expression, or enable the grid in AE and drag the images around manually. Once you have all of the images laid out in your grid you could start animating each one so they move up in Y, add in the little flutter before they settle into position and then move them up and out of frame. This can be done manually but it is very difficult to get it to look natural or you can write an expression. If you base the expression on a random time or a marker or the layer's index (number) then you could automate this process. If you do it by hand you can get the first image to fall into place to look right and then copy the keyframes and paste them to a new layer with the CTI at a different time. Work area is going to be much larger than the camera's view so you'll need to work in multiple views and lay things out where you can see what you are doing. Once you get all of the images falling into the grid (or at least all of the images you'll need for the shot you can animate the camera so that sit dolly's past the grid of pictures at the right speed and the right angle.
The shot that precedes this one is just a different camera move using a different angle.
There are also shots of individual images falling past the camera. This is just a series of layers with position keyframes or expressions making them fall.
The close up shots give us an idea about the size requirements of the images.
This is about as close as the camera ever gets to an image. The image is about as wide as your comp so if your are working in a 1080 project then the comp size is 1920 pixels wide by 1080 pixels high. Your images would not need to be any bigger than 1920 X 1920. Each image has a white boarder. If the images have been cropped to square ones you can add the boarder in Photoshop or you can add a stroke effect in AE to a mask on the image, or you can make this a part of the square comp you are using for each of the images. Fortunately there doesn't seem to be any curl in the images so that simplifies things. If you wanted to add a little bend to the images as they fall into place you could add the page turn transition or use a mesh warp to simulate a curved layer. Anything more complex than that would be overkill.
So there you have it. If I were starting from scratch and did not have any 3rd party plug-ins or animation presets for falling images or making a grid and I wanted recreate the sample video I'd need about 2 or 3 eight hour days. With scripts, animation presets and expressions I could probably do it in about a day. Buying the template would probably reduce the workload to 2 or 3 hours.
Thank you so much for your in-depth and detailed reply. It's truly much appreciated, I'm going to get to work on this now.
Did you manage to replicate the project...? Could share the technique please...(Beginner)