Breaking image into tiles

[Albo]

I am using an example image.

I would like to break the image along the grid lines and have  the grid patern become image tiles.

 

What would be the way to do this.

At the moment I cn select individual tiles and export but it would take a long time to do.

The tiles will be imported as PNG's in a film.

 

Thanks,

 

[gdenby]

Hi, Albo,

 

I don't have the time now to step thru this fully. 

 

I mostly use Designer, not Photo. Here's what i would do in Designer, I suppose the same might be possible in Photo.

 

Create the vector shape array of diamonds, and half diamonds. Be sure to have them in accurate position. Group them. Use the fill tool, choose bitmap, and place an image onto the group, sizing appropriately.

 

In export persona, double click each geometric shape, and use the context mouse menu to "create slice."  Repeat as needed.

 

I don't understand what you refer to when you say the PNG's will be imported into a film, but the above should make the PNGs one at a time, but need to be numbered/named as needed.

[Albo]

Thanks for the info.

I will try that.

As for the PNG. I simply meant that PNG will be exported.

 

UPDATE: I tried that but it puts the same image in every shape. I want the image to be represented in evey shape but as the whole original picture.

Each shape should represent one part of the global image used.

[R C-R]

4 hours ago, Albo said:

UPDATE: I tried that but it puts the same image in every shape. I want the image to be represented in evey shape but as the whole original picture.

Each shape should represent one part of the global image used.

I think this would work but it would require placing the bitmap at the same full (100%) size on each individual vector shape, with each bitmap set to a global origin of 0x & 0y. This would be very tedious & require careful alignment.

 

A slightly less tedious method in Affinity Photo that still requires a ton of work would be to use the Pen Tool's "Selection" button with 'snap to grid' enabled to create a selection from each triangular or diamond shaped grid area one at a time (yikes!), then -- making sure the original background layer is (re)selected in the Layers panel -- to use copy & paste to create each tile as a new pixel layer. Once that is done, switch to the Export persona, click on the Layers tab, select all the individual tile layers & click the "Create Slice" button at the bottom of that tab. You can then switch to the Slices tab, uncheck the box next to the original background layer, & click the "Export Slices" button at the bottom of that tab. I suggest creating a new folder for the exported pngs to keep things simple.

 

You can ignore the warning triangles about duplicate paths because a suffix will automatically be added (like "Background_2.png") to each exported png file.

 

Obviously, this is far from ideal but it will work, as (I hope) this Image grid partial.afphoto 'proof of concept' file should show. It is not very accurate because I just had your jpg screenshot to work with so I used the Inpainting brush to get rid of the jpg's white grid lines & I was not super careful about aligning the jpg to the grid I created in AP, but you should be able to get accurate results using the original file.

 

I just did the top row of tiles (I try to maintain the appearance of actually having a life apart from this forum ) but I also included the tile exports.zip so you can see what those 8 exported tiles look like.

[lepr]

.

 

[R C-R]

4 hours ago, owenr said:

gdenby's solution does work if you do the "sizing appropriately" step

Well, duh!  When I wrote that this would "require placing the bitmap at the same full (100%) size on each individual vector shape," I completely overlooked the "Group them" step! Obviously, applying the image as a bitmap to the grouped vector shape array makes it unnecessary to apply the image to each shape individually, making this much more practical. In fact, when I tried this in Affinity Photo, the image came in at the appropriate size, so I did not even have to worry about that.

 

I did notice one potential minor issue that I think is inherent in creating any diagonal tiling pattern: the diagonal edges of each exported png shape are antialiased, so there will be faint edge lines visible when the tiles are assembled to form the original image in a film or whatever. I am not sure what a suitable workaround for that would be.

[Aammppaa]

Here's what I'd do in Designer

1. Draw the array of diamond tiles…
   1. Power duplicate an array of squares
   2. Rotate 45 degrees
   3. Squash them vertically.
2. Place your bitmap image behind the diamonds and size / position appropriately.
3. Cut (Ctrl+X) the bitmap
4. Select the first of your diamonds and Paste Inside (Ctrl + Alt + V)
5. Repeat for each diamond.
6. In Export Personal select all diamonds in the Layers Panel, and click Create Slice
7. Export your PNGs

This method means that you are only placing and sizing the bitmap once, and then pasting it inside each of the diamonds.

15 minutes ago, R C-R said:

faint edge lines visible when the tiles are assembled to form the original image in a film or whatever. I am not sure what a suitable workaround for that would be.

 

Assuming that the tiles all join at the same moment, I would switch  to the original complete image in the last frame. If they join individually, then it is more complicated… you need to use the animated tiles themselves as a mask which gradually reveals the original complete image. And if the tiles join to form sub-groups which them continue to animate, then it gets trickier still.

 

Alternatively you add lots of motion blur and lighting effects, do everything at 4x actual size and downscale to minimize the artifacts!

 

Alternative…

Don't do any of the work in Affinity!

Use your motion graphics app and start with the complete image, which you then explode into tiles.

Run the animation backwards.

[R C-R]

4 minutes ago, Aammppaa said:

5. Repeat for each diamond.

This seems like a lot more work than applying a bitmap fill to the group of shapes in one step.

[Aammppaa]

Oops - yes I missed the 'Group' in the original solution too.

 

Can't think of any advantage to more laborious methods such as mine.

[lepr]

.

 

 

[lepr]

.

 

[R C-R]

1 hour ago, owenr said:

Grouping is not important. You can select all the diamonds and triangles then do the bitmap fill and drag a handle to get exactly the same result as with a group.

What I found, at least with the photos I was testing with in Affinity Photo, is that applying the bitmap fill to the group automatically sized it to the proper h & w dimensions for the entire group. It might be different with 1:1 aspect ratio photos but my tests were not done with them.

 

1 hour ago, owenr said:

The crucial step is to drag a handle of the Fill Tool. At the instant you start dragging, the "bitmap" expands to the collective extent of the selected objects.

I noticed that when I tried creating & grouping just some of the shapes to begin with & then creating & adding more shapes to the group afterwards, but as above, there was no need to adjust the handle(s) with my test files if I put every shape into the group first.

[lepr]

.

[firstdefence]

 

[lepr]

.

[R C-R]

3 hours ago, owenr said:

@R C-R, firstdefence's video shows the initial mapping of the texture to each member of the Group as happens to me, so how are you constructing your document so that the entire Group is initially textured with the "bitmap" scaled to the extent of the Group?

I am not sure but one or more of the following might have something to do with it:

 

1. As I said, all my work was done in Affinity Photo, not Affinity Designer.

2. I sized each new document to be exactly the same size as the bitmap I used for that document.

3. I created & adjusted the grid to match the dimensions of each document (so the grid had only full sized diamonds & half diamonds).

4. At the edges I created triangular shapes so that no shape extended beyond the edges of the document in any direction.

5. Initially, "Maintain fill aspect ratio" was not enabled, nor was it necessary (I guess) because the dimensions of the document & bitmap were already matched.

6. Throughout the whole procedure I used the minimum number of snapping options necessary for each step (including enabling "Snap to spread" only when creating the edge triangles).

 

Basically, I set everything up the way I would if I was doing this for a 'real' project, so no small edge tiles, if that makes any sense.

[firstdefence]

a section recompiled

 

Slice image up.afdesign

[lepr]

.

 

[R C-R]

1 hour ago, owenr said:

Thanks.

Photo and Designer do not differ for me and, as far as I understand, Albo, gdenby, firstdefence and myself get one behaviour with the Group fill and you strangely get another.

Does it make any difference if you do as I did in my step 4 & do not allow any of the shapes to extend past the edge of the document, or if as in my step 2 the document is sized to the same dimensions as the bitmap? Just guessing but I think one of those things might explain why we get different results.

[lepr]

.

[firstdefence]

7 hours ago, R C-R said:

the document is sized to the same dimensions as the bitmap

I did just that R C-R, re the neatness of the shapes I just blocked them in to get a working model but it would have been better if I'd booleaned the parts that went beyond the canvas, I'll do that when I redo the video to make it a tidier result. I did quickly attempt to get a clean layout of the grid so small bits didn't exist but failed at that because I was just being lazy.

 

I think the next big issue with this is the naming of the slices into a coherent format i.e left to right, top to bottom?

[R C-R]

1 hour ago, firstdefence said:

I think the next big issue with this is the naming of the slices into a coherent format i.e left to right, top to bottom?

Give or take the edge triangles, power duplicating first from left to right & then down probably will help with that. Otherwise, it means tediously checking/moving each shape in the Layers panel into the preferred order.

[firstdefence]

37 minutes ago, R C-R said:

Give or take the edge triangles, power duplicating first from left to right & then down probably will help with that. Otherwise, it means tediously checking/moving each shape in the Layers panel into the preferred order.

Yes I've just done that, I created the diamond, placed it in the highest left position then power duplicate (Cmd + J) but went over about 6 diamonds so I could power duplicate the row diagonally down and to the right, I found this the quickest and neatest method to fill the grid, then just deleted any diamond that was wholly out of the canvas. 

 

 

Flattened the edges.

 

 

The top was done the same way as the sides but the base line was done by creating boxes and geometry subtracting each diamond and you end up with a much neater matrix

[R C-R]

1 hour ago, firstdefence said:

... I created the diamond, placed it in the highest left position then power duplicate (Cmd + J) but went over about 6 diamonds so I could power duplicate the row diagonally down and to the right ...

I haven't tried that but it seems like it would not create a very coherent row-column shape or slice order. If that is important, probably best to create the top left triangle (Pen Tool, snapping to both grid & spread), power duplicate across for the top row of triangles, then draw the leftmost edge triangle, then the first diamond, power duplicate across for the second row, add the rightmost edge triangle, duplicate & move down, & so on. The bottom row would require special attention if the grid doesn't line up perfectly.

 

I know this is a slower way to create the shapes but I think it is the easiest & least error-prone way to get a coherent row-column shape order. Naming the slices should take care of itself in the export process.

 

Regarding matching the grid to the image (so no truncated diamonds & just 4 triangle shapes for the edges), I have assumed that for animation/movie work it would be desirable to match the grid to the image precisely. There are several ways to do that, depending on what is acceptable, like stretching the image to fit the grid via resampling & exporting that as a png (or whatever) to use as the bitmap; adding a border/frame to the image & exporting that for the bitmap; or cropping the image slightly.

[firstdefence]

@R C-R It works surprisingly well, so if you start with the bottom curve and work your way up the layers you'll see that it follows a desirable order: Diamond order.afdesign