16 : 9 perhaps?
As Richard states, if you either Scale, or Crop your Still Images to the Frame Size in an HD Project, you should not have any black bars.
If you are in an SD Project, the 16:9 will be attained via non-square pixels, while your Still Images will most likely have square pixels. Then, you would Scale to about 0.2% larger than the pixel x pixel Frame Size of the Project.
The Aspect Ratio would stay the same, 16:9. Note that most still cameras shoot ~ 4:3, so when you Scale for the horizontal, you will have some extra pixels at the top and bottom of the image. You can adjust that "crop" to either the top only, or bottom only, with the Fixed Effect Motion>Position, and "move" the Image up, or down, as you need.
My Sony DSLR has the option to switch the still shooting mode to 16:9, other cameras may be able to do the same.
That is good to know. All of mine are locked into 4:3, so I need to do a bit of tweaking.
In general I always make a background layer that will be behind my photos if they don't take up the entire screen if I don't want to scale them. You can make a decent background very easily using titler and textures etc.
But like everyone else has stated if your footage is 16x9 then ideally 16x9 would a good option to go with for the photos as well. Although I think it's worth mentioning that IMO it's also a good idea too shoot your photos in a higher resolution than your footage in some cases this way if you crop the shot you have plenty to work with to still take up the entire screen without needing to scale. You can then also keyframe the photos in and out more with movement without having to worry that they won't take up the full screen.
Rumor had it that you could run into some export trouble if your still photographs didn't exactly match your sequence settings, and you were using certain CUDA cards. In other words, it wasn't enough to be 16:9 in a 16:9 sequence, you needed to be exactly 1920 x 1080 in a 1920 x 1080 sequence. The reason for this was supposedly something to do with a bug or three that didn't handle the required scaling well if scaling was indeed needed. And the fix was supposedly making the export without CUDA supposrt (IOW, software only, no hardware help), which make exports considerably (3x or more) slower.
Notice that I couched all that as supposition. Because it is. I've never run into this myself, I've just heard rumors. But it makes since that if your sequence is 1920 x 1080, everything that you include in that sequence should be 1920 x 1080, just to make it easier on the software and to make it faster to export.
My understanding is that, if you use some fairly big image's it will cause your CUDA card to run out of memory. I got this information from this page.
Recently though I have done 3 projects where clients have given tons of photos all at different resolutions. Most of the photos though were bigger than 1920x1080. Some of them were 3000x2000 and others right around 1920x1080. With the photos that were 3000x2000 I was able to do a lot of really nice pans and was really able to keyframe them and animate things very nicely IMO. I didn't have to turn off GPU based MPE acceleration and I also didn't run into any bugs. However my exports are always MUCH slower than normal when I use photos that big. So I have experinced the slow export times but I would still honestly recommend anyone to recieve the highest resolution photos they can get from their clients. Simply because you can always make a image smaller in photoshop. But you can only scale something up so much before the details begin too look terrible. Although Photoshop does do a wonderful job of scaling when compared to just scaling something in Premiere.