Creating a flash-able disk image from a folder

First, create an empty file of the desired size that the disk image will be, using dd. The below will create a 1GB file (count*bs)

dd if=/dev/zero of=disk.img bs=512M count=2

Then, use losetup and the -P flag to use the newly created image as a loopback device.

losetup -P /dev/loop0 disk.img

Once the loopback device is created, you should be able to create a partition map on it using fdisk. (If you don't want to use fdisk, another disk partitioning program should work)

fdisk /dev/loop0

Inside of fdisk, you can now create your partitions. For instance, g will create a GPT map, and n will add a new partition to it. When you are done, type write.

Once the partition map is created, the system will reload the partitions, and you should now at least have a /dev/loop0p1 device, representing the first partition. We can run whichever mkfs we want on this partition now:

mkfs.ext4 /dev/loop0p1

Once the filesystem is created, it can be mounted into an empty folder as if it were a real device.

mkdir /mnt/diskimage
mount /dev/loop0p1 /mnt/diskimage

You can put files or folders inside of the mount point now, and they will be written to the disk.img file as you write to the folder. When you're done, you can unmount the filesystem, and disconnect the loopback device.

umount /mnt/diskimage
losetup -d /dev/loop0

Now the disk.img file should contain the exact desired filesystem, and can be flashed to one or more target devices using dd on unix/linux or a program like win32diskimager for windows.

dd if=disk.img of=/dev/sdX bs=500M